AMD and NVIDIA Launch Next-Gen GPUs with Advanced AI Features

MSI GeForce RTX 5070 Ti VANGUARD SOC Review

With a bit of overclocking with an app like MSI Afterburner, the GeForce RTX 5070 Ti can hit a boost clock speed of 3 GHz with headroom to go even further. Depending on the game, it only takes a few seconds and offers up a notable bump to performance. When it comes to the GeForce RTX 5070 Ti, no matter the model, you're looking at performance that sits comfortably in the range of the GeForce RTX 4080 and RTX 4080 SUPER when gaming in 4K. As these are still two of the most capable and powerful options for gaming at this resolution, with or without ray-tracing enabled, the GeForce RTX 5070 Ti is a fantastic GPU for playing cutting-edge cinematic games or simply firing up a competitive title for some high-refresh-rate action. A quick look at the MSI GeForce RTX 5070 Ti VANGUARD SOC The GeForce RTX 5070 Ti is built using a cutdown version of the GB203 chip found in the GeForce RTX 5080, another GPU in the GeForce RTX 50 Series line-up that is great for overclocking. Of course, when it comes to overclocking, you want a model with exceptional power delivery, cooling, and thermal design. The MSI GeForce RTX 5070 Ti VANGUARD SOC fits the bill, and like the RTX 5080 variant we reviewed, it ships with one of the fastest out-of-the-box overclocks with a 150 MHz boost on top of the 2452 MHz reference spec. The kicker is that even with this generous OC, the GPU temperature stays closer to 50 degrees, even after a few solid hours of gaming. As MSI's flagship GeForce RTX 5070 Ti, the VANGUARD is also the largest and most well-equipped in terms of thermal design, heatsinks, fans, and cooling. It's a premium option through and through, and with a price tag of $919.99 USD, it commands a decent premium over an MSRP model like the GeForce RTX 5070 Ti VENTUX 3X. So, with that in mind and the fact that the GPU barely breaks a sweat, we decided to give the GPU a modest manual overclock on top of the out-of-the-box to see what difference it made regarding cooling and in-game performance. Cooling-wise, you're still looking at an essentially silent GPU when in use, with GPU temperatures sitting in the 55-degree range and high-speed GDDR7 memory temperatures sitting closer to 60 degrees Celsius. And with performance that is up to 7-10% faster than the baseline GeForce RTX 5070 Ti specs, the MSI GeForce RTX 5070 Ti VANGUARD SOC is not only faster than the GeForce RTX 4080 SUPER, but it closes the gap between it and the GeForce RTX 5080. This is an impressive result, and this is based on a modest overclock using the same power draw as the out-of-the-box 'Gaming' mode BIOS. RTX Blackwell - NVIDIA's Gaming Architecture for the AI Era Below is a summary of NVIDIA's GeForce RTX 50 Series and RTX Blackwell architecture, applicable to all models. NVIDIA describes 'Neural Rendering,' which includes all previous versions of DLSS and the brand-new DLSS 4, as the 'next era for computer graphics.' They're not alone; the Lead System Architect for the PlayStation 5 Pro console, Mark Cerny, recently said that ray-tracing is the future of games and that AI will play an integral role in making that happen. DOOM: The Dark Ages developer id Software shared a similar sentiment, adding that the arrival of DLSS was an 'inflection point' for PC game visuals and performance and on par with the arrival of dedicated GPUs and programmable shaders. With the arrival of the Blackwell generation and the GeForce RTX 50 Series, AI is now being used to accelerate programmable shaders with the brand-new RTX Neural Shaders. Yes, these are actual neural networks that use live game data, and the power of Tensor Cores to do everything from compress textures, render lifelike materials with a level of detail impossible to match using traditional rendering methods, and even use AI to partially trace rays and then infer "an infinite amount of rays and bounces for a more accurate representation of indirect lighting in the game scene." RTX Mega Geometry is incredible in its own right; it essentially increases a scene's geometry detail and complexity (triangles or polygons) by up to 100x. 100 times the detail, it's hard to wrap your head around - but the added benefit in a game like Alan Wake 2 is dramatically improving the performance of the game's Full Ray Tracing or Path Tracing mode. With DLSS 4 and RTX Neural Shaders, NVIDIA's GeForce RTX 50 Series and RTX Blackwell architecture (which includes the same AI optimizations as data center Blackwell) can be viewed as the turning point for PC gaming - the moment when AI becomes integral to everything from designing a game to programming and then finally rendering it on a 4K display to play. DLSS 4 includes more goodies than NVIDIA's highly touted new Multi Frame Generation technology, but let's start there. DLSS 3's version of Frame Generation has evolved with DLSS 4, powered by Blackwell hardware and software, and an innovative use of AI to generate frames 40% faster while using 30% less VRAM. Switching to a new model also means that Frame Generation and Multi-Frame Generation could soon come to GeForce RTX 20, 30, and RTX 40 Series owners. DLSS 4 benefits all GeForce RTX gamers. With the 5th Generation of Tensor Cores in the GeForce RTX 50 Series delivering 2.5X more AI performance, NVIDIA's latest GeForce RTX 50 Series GPUs can execute five complex AI models - covering Super Resolution, Ray Reconstruction, and Multi Frame Generation in a couple of milliseconds. Part of the reason it happens so quickly is the addition of hardware Flip Metering, which shifts frame pacing to the Blackwell display engine - the result is frame rates of up to 4K 240 FPS and higher without stuttering issues. With up to 15 of every 16 pixels generated by AI, the result is up to 8X the performance when compared to native rendering or rasterized performance. DLSS Super Resolution and Ray Reconstruction are also switching to a new 'Transformer' model, with over double the parameters and four times the compute requirement. This is one of the most exciting aspects of the GeForce RTX 50 Series, as it pushes DLSS into a new realm of image quality and performance. The best part is that it will work on all GeForce RTX GPUs; however, there will be a performance hit compared to running it on an RTX 50 Series GPU. Already available in games, DLSS 4's Transformer model is another DLSS 2.0-like moment for the technology, and the results speak for themselves. Even better, DLSS 4 is integrated into the NVIDIA App with a new 'DLSS Override' feature that allows users to experience the latest tech without waiting for a path or game update. DLSS 4 is built to be backward compatible, with 75 games and apps supported (so far). It doesn't stop there, as the new AI Management Processor (AMP) allows AI models to share the GPU with graphics workloads. As a result, expect to see digital humans in games alongside AI assistants like NVIDIA's Project G-Assist becoming more prevalent in the coming years. This filters down to the creator side, with AI assistants for streamers, who will also benefit from the GeForce RTX 50 Series' expanded creator features. RTX Blackwell introduces 4:2:2 chroma-sampled video encoding and decoding. The ninth-generation NVENC encoder also improves AV1 and HEVC quality. RTX Blackwell is a game changer for creators and editors, especially with the new low-voltage and cutting-edge GDDR7 memory that dramatically improves memory bandwidth and speed. Here's a look at the specs for the GeForce RTX 5070 Ti compared to the previous generation's GeForce RTX 4070 Ti and GeForce RTX 4070 Ti SUPER, as well as the GeForce RTX 5080 and GeForce RTX 4080. As mentioned above, the new GeForce RTX 5070 Ti and GeForce RTX 5080 are built using the same GB203 chip. The latter features 20% more hardware in the form of CUDA, Tensor, and RT Cores, which means it's more powerful. Conversely, the GeForce RTX 5070 Ti is more efficient because it uses 17% less power (on paper) when under full load and can still deliver excellent 4K gaming performance. It shares a similar memory configuration, with 16GB of fast GDDR7 memory on a 256-bit bus, providing an impressive 896 GB/sec of bandwidth - which is excellent for playing games at higher resolutions. Speaking of, 4K is where the GPU excels and can offer a sizable improvement of the previous generation's GeForce RTX 4070 Ti and GeForce RTX 4070 Ti SUPER, even though the 40 Series and 50 Series are built on a similar custom TSMC 4N process. Not seeing a process node shrink is unusual for a GeForce RTX generation and could be one of the reasons why there's not a massive performance uplift in specific workloads or scenarios. However, as the MSI GeForce RTX 5070 Ti VANGUARD SOC is a supercharged model, it offers a 24% improvement to 4K gaming over the GeForce RTX 4070 Ti SUPER in the games we tested, which is an impressive feat when you factor in that it only has 6% more CUDA Cores and RT Cores. This shows you there's much more to a GPU generation and chip architecture than moving to a newer and smaller process. RTX Blackwell is built for the AI era, where raw performance meets cutting-edge hardware and software designed to improve performance, enhance overall efficiency, and deliver the future today. DLSS 4's switch to a more complex transformer model for the GeForce RTX 50 Series (which is backward compatible with previous GeForce RTX generations) is a game changer for AI-powered upscaling and image fidelity. More and more titles are being released with RT every year, and the new transformer model for the DLSS Ray Reconstruction denoiser significantly improves image quality compared to native rendering. The difference is night and day, and it's downright remarkable that a game like Cyberpunk 2077 or Alan Wake 2 can look that much more impressive running at 150 FPS with DLSS and Frame Generation than running at 30 FPS natively. On the ray-tracing front, NVIDIA's new RTX Neural Shaders leverage AI to compress textures, render complex environments, assist in calculative ray bounces, and more - all to provide a 4X or 8X improvement to image quality and performance as opposed to a modest generational bump in raw performance. We're just starting to see these technologies appear in games, so it will take some time to get a clearer picture of what Neural Shaders mean for the future of PC gaming. The VANGUARD is a brand new design from MSI, making its debut with the GeForce RTX 50 Series and more powerful models like the GeForce RTX 5070 Ti. It is a premium design similar in some respects to the SUPRIM models from the company and represents the current flagship option for its range of GeForce RTX 5070 Ti cards. Of all the RTX 5070 Ti GPUs we've reviewed so far, it's the most powerful and the largest in terms of size and weight - it's a triple slot GPU that weighs in at 1945 grams - making it virtually identical, physically, MSI's GeForce RTX 5080 VANGUARD SOC. The thermal performance here is second to none, thanks to MSI's new STORMFORCE fans that are quiet and powerful, the advanced vapor chamber for the GPU and VRAM, the core pipes designed for maximum contact and heat dissipation, and the use of high-quality components and thermal padding throughout the design. Put it all together, and you've got MSI's HYPER FROZR THERMAL DESIGN, which sees GPU temperatures stay below 55 degrees Celsius, even under load and manually overclocking the memory and clock frequencies. Visually, the VANGUARD design is one of MSI's more impressive-looking GPUs - especially if you're a fan of stylish RGB lighting. The lightning bolt that runs diagonally across the GPU is met by two transparent MSI badges that light up, with grey and black two-tone flourishes that converge on the metal backplate with the iconic MSI dragon. Our only criticism is that it might all be a little overkill and that even for those wanting a premium build with plenty of cooling power and OC potential, a more compact card like MSI's revamped GAMING TRIO design could be the way to go, especially when you're not talking about a GeForce RTX 5080 or the flagship RTX 5090. The Games and Tests PC gaming not only covers a wide range of genres and styles, from indie games with simple 2D graphics to massive 3D worlds lit by cutting-edge real-time ray tracing technology. With that, the needs and requirements of each gamer vary. High refresh rates and latency reduction become more important than flashy visuals or playing at the highest resolution possible for those who live and breathe fast-paced competitive games. For those who want to live in a cinematic world and become a key player in an expansive narrative, ray-tracing, and high-fidelity visuals are a stepping stone toward immersion. Our chosen benchmarks cover various games, engines, APIs, and technologies. For the GeForce RTX 5070 Ti, all tests are run at 4K and 1440p and include results for performance-boosting Super Resolution technologies like NVIDIA DLSS 4 - including Frame Generation and the new Multi Frame Generation. In many ways, DLSS numbers are more important in 2025 than native rendering - a title with ray tracing isn't meant to be played without Super Resolution. Also, DLSS technologies like Ray Reconstruction and the new RTX Mega Geometry dramatically improve visual fidelity and detail compared to native rendering. However, our benchmark results are still sorted using 'raw performance' or native rendering. Here's the breakdown of games, settings, and what's being tested. It's fantastic to have an enthusiast-class GPU that delivers 5-10% more performance with a modest manual overclock. For all our benchmarks, we boosted the memory and clock speeds by 250 MHz and found that it was stable and made a real difference in performance when added to the 150 MHz OC you get out of the box. This makes the MSI GeForce RTX 5070 Ti VANGUARD SOC 34% faster than the GeForce RTX 4070 Ti for 4K gaming and 24% faster than the GeForce RTX 4070 Ti SUPER refresh. In addition, it's 8% faster than the GeForce RTX 4080 and 4% faster than the GeForce RTX 4080 SUPER. The 83 FPS average, which climbs to 104 FPS with DLSS enabled, puts it within 10% of the GeForce RTX 5080 with reference specs. This makes the MSI GeForce RTX 5070 Ti VANGUARD SOC a 4K powerhouse, alongside being a beast for 1440p gaming or pairing with an Ultrawide 3440x1440 cinematic aspect ratio display. And with DLSS, which includes the new and impressive DLSS 4 transformer model for Super Resolution and Ray Reconstruction, you're getting enough of an additional performance boost that you'll be able to hit 100+ FPS in most games running in 4K without any tinkering. With Radeon's relatively lackluster RT performance, the MSI GeForce RTX 5070 Ti VANGUARD SOC is a much better all-rounder than the Radeon RX 7900 XTX. With a third of the games in our benchmark suite featuring ray-tracing, the GeForce RTX 5070 Ti VANGUARD is 14% faster on average for 4K gaming. However, there are titles where the GeForce RTX 5070 Ti and RTX Blackwell seem to underperform compared to Radeon. Call of Duty: Black Ops 6 is still a game where Radeon hardware shines, likewise when it comes to Cyberpunk 2077 with RT turned off. If you go through the individual benchmark results below, you'll notice a lot of titles where the GeForce RTX 5070 Ti VANGUARD, RTX 4080, RTX 4080 SUPER, and Radeon RX 7900 XTX are all within a few percentage points of each other. Average Gaming Performance - 1440p Results When it comes to 1440p gaming, the overclocked MSI GeForce RTX 5070 Ti VANGUARD SOC is still faster than the GeForce RTX 4080 SUPER and AMD's RDNA 3 Radeon RX 7900 XTX flagship - on average. The average performance of 139 FPS without DLSS enabled is around 80% faster than the GeForce RTX 3070 Ti at this resolution and 50% faster than the GeForce RTX 3080. These are two GPUs that are still high up on the monthly Steam Hardware Survey results, which makes the MSI GeForce RTX 5070 Ti VANGUARD SOC a pretty impressive upgrade for those with a GeForce RTX 30 Series GPU looking for a new enthusiast-class card. The good news is that DLSS 4 still delivers fantastic results at 1440p, which bodes well for the upcoming GeForce RTX 5070 and firing up hardware-intensive Path Tracing modes in titles like Alan Wake 2. For pure 1440p gaming, the gen-on-gen upgrade over the GeForce RTX 4070 Ti SUPER drops to 18% - which is still a respectable result but not earth-shattering. Regarding GPUs that are GeForce RTX 4070 Ti and higher, you've got an excellent option for high-refresh-rate gaming at this resolution. For the MSI GeForce RTX 5070 Ti VANGUARD SOC reviewed here, competitive gamers can hit 320 FPS in Counter-Strike 2, while the single-player DOOM Eternal hits 288 FPS with ray-tracing enabled. Even though there are differences, pretty much every GPU you see on the list above is a capable GPU at this resolution; the MSI GeForce RTX 5070 Ti VANGUARD SOC just so happens to be one of the best. Benchmarks - 3DMark Synthetic Tests 3DMark offers a suite of synthetic benchmarks built to test GPUs in various scenarios. 3DMark Steel Nomad is a cutting-edge DirectX 12 benchmark with newer, modern rendering techniques designed to push GPUs to their limit. The 'Light' version tests at 1440p, while the main Steel Nomad benchmark tests pure native 4K rendering. Port Royal is a benchmark focusing exclusively on real-time ray tracing for lighting effects like reflections, shadows, and more. 3DMark's new Steel Nomad benchmarks are the official successors to Time Spy. They're designed for modern AAA gaming and run on modern engines like Unreal Engine 5 - running at 1440p and 4K without ray-tracing or upscaling. Looking at the MSI GeForce RTX 5070 Ti VANGUARD SOC's score, you're looking at a result that sits alongside the GeForce RTX 4080 and GeForce RTX 4080 SUPER, which is reflected in most of the real-world gaming benchmarks we carried out. The score demonstrates the decent generational uplift over the GeForce RTX 4070 Ti and how modern GPU architectures like RTX Blackwell are better equipped for 2025 gaming. Port Royal is a synthetic benchmark for ray-tracing, and here, the MSI GeForce RTX 5070 Ti VANGUARD SOC delivers a score of 6.5% higher than the GeForce RTX 4080 SUPER and a whopping 28% higher than the flagship RDNA 3 GPU, the Radeon RX 7900 XTX. This result can be seen in some titles; however, for the most part, we found that the RT performance of the overclocked MSI GeForce RTX 5070 Ti VANGUARD SOC was on par with the GeForce RTX 4080 SUPER - not faster. With AMD's GeForce RTX 5070 Ti competitor, the new Radeon RX 9070 XT, set to arrive in a matter of days, this chart highlights that RDNA 4's new flagship mid-range offering will need to offer better than Radeon RX 7900 XTX RT to be competitive with GeForce RTX. DLSS 4, Frame Generation, and the new Multi-Frame Generation DLSS 4 and Multi Frame Generation are impressive bits of technology, thanks mainly to the overall improvements to performance and latency on the Frame Generation side and the new 'Transformer' model for Super Resolution and Ray Reconstruction. We used the DLSS 'Quality' mode preset for these benchmarks, often delivering better-than-native image quality. Multi Frame Generation is an evolution of the Frame Generation technology that NVIDIA introduced with the GeForce RTX 40 Series, which leverages AI to generate up to 3 additional frames to deliver a massive improvement to performance or the overall smoothness of a game's presentation, motion clarity, and feel. With the correct settings and the right game, it is like watching digital sorcery. Sure, there are artifacts, and a frame-by-frame analysis will make these obvious - however when you're looking at 200+ FPS, it's not something you notice. When paired with DLSS 4's new transformer model for AI Super Resolution, RTX Blackwell's new AI scheduling, and 'Flip Metering,' the effect is seamless as long as the input frame rate is at least 60 FPS - preferably 70-80 FPS. Multi Frame Generation is not a one-click fix for poor performance or a poorly optimized game. Still, it does mean that decent performance becomes excellent, and games with CPU or other limitations can see those limitations disappear. For a lot of gamers, 80 FPS in a cinematic title like Dragon Age: The Veilguard running in 4K on the MSI GeForce RTX 5070 Ti VANGUARD SOC with ray-tracing and the DLSS 'Quality' preset would be more than enough performance. What Multi Frame Generation brings to the table is the ability to maximize the refresh rate of your display, and as our GPU test bench includes a 4K 240 Hz OLED display, the result is that it can do precisely that. Play Dragon Age: The Veilguard at 4K 240 Hz, and you immediately notice the additional smoothness and performance. Path Tracing Performance - 1440p Path Tracing, or Full Ray Tracing, arrived with the GeForce RTX 40 Series and DLSS 3 and is leveling up with the GeForce RTX 50 Series and DLSS 4. It's only possible thanks to AI technologies like DLSS Super Resolution, Ray Reconstruction, and RTX Neural Shader technology like RTX Mega Geometry. It's designed specifically for these technologies, and we're only including native or rasterized performance to highlight just how intensive it is on a GPU as powerful as the GeForce RTX 5070 Ti. In fact, outside of the massive increase in performance, these games also look notably worse without DLSS 4. Remedy is one of the most talented PC gaming developers working today. Scratch that; Remedy is one of the most talented game developers, period - going back to the original Max Payne, which blew us away with its slow-motion bullet-time action and incredible (for the time) visuals. Alan Wake 2 is the culmination of decades of technical experience, and running on the studio's in-house engine, it's a stunning showcase for what the MSI GeForce RTX 5070 Ti VANGUARD SOC can do. It features full support for DLSS 4 and Multi Frame Generation, and it's the first game to incorporate the brand-new RTX Mega Geometry tech for rendering path-traced environments. Star Wars Outlaws is another DLSS 4 showcase, especially regarding Ray Reconstruction and how the new transformer model makes the DLSS Balanced preset look superior to native rendering. These benchmarks show that the MSI GeForce RTX 5070 Ti VANGUARD SOC is a powerhouse for 1440p Path Tracing with DLSS 4 and, depending on the title and settings, a viable option for taking Full Ray Tracing or Path Tracing into the 4K realm. Temperature and Power Efficiency The MSI GeForce RTX 5070 Ti VANGUARD SOC didn't break a sweat when overclocked and remained quiet throughout all benchmarking. Our manual overclock drew a little more power than the non-OC or modest OC models we tested, but not drastically. As a 300W GPU, it's also the most efficient GPU in the RTX Blackwell line-up so far, showcasing that the architecture scales incredibly well - delivering superior performance to the GeForce RTX 4070 Ti SUPER and GeForce RTX 4080 while using similar power. With manual overclocking, you can push the MSI GeForce RTX 5070 Ti VANGUARD SOC beyond 300W, and when testing this side of the GPU, we didn't see the power draw reach up to 310 or 315W. Final Thoughts The GeForce RTX 5070 Ti is a fantastic GPU for PC gaming. It offers a massive performance boost compared to the GeForce RTX 30 Series and a sizable bump to 4K gaming compared to the previous generation's GeForce RTX 4070 Ti. In reviewing the MSI GeForce RTX 5070 Ti VANGUARD SOC, which is MSI's flagship RTX 5070 Ti, we were blown away by how quiet and calm it stayed even when overclocked. It also shows that, like the GeForce RTX 5080, there's enough headroom to boost performance with a few tweaks, delivering up to 7-10% more performance in some games and workloads. The build quality is excellent, and MSI's new VANGUARD design for the GeForce RTX 50 Series is akin to getting a new and impressive SUPRIM design to play with. Unfortunately, the build quality you get, OC headroom, and premium cooling come at a cost, with the $919.99 price point sitting notably higher than the $750 MSRP. Granted, MSI has MSRP models like the VENTUS 3X we reviewed on launch day, and for many, that would probably be the way to go. The MSI GeForce RTX 5070 Ti VANGUARD SOC is absolutely for those who want to overclock the GeForce RTX 5070 Ti, increase power limits, and find that stable point where it leaves the GeForce RTX 4080 SUPER behind. Overclocking aside, the show's real star is how well the GeForce RTX 5070 Ti benefits from NVIDIA's new DLSS 4 updates - including the RTX 50 Series exclusive Multi Frame Generation. With powerful raw performance the new transformer AI model for Super Resolution and Ray Reconstruction delivers better-than-native image quality when gaming in 4K. And with that, you've got 4K 100+ FPS in most titles, with or without ray-tracing enabled. And with the shift to Neural Shaders and AI-enhanced features and technologies powering more and more titles, the GeForce RTX 5070 Ti is more than equipped to handle the next couple of years of gaming.

ASUS PRIME GeForce RTX 5070 Ti OC Edition Review

The GeForce RTX 5070 Ti, the successor to the GeForce RTX 4070 Ti and the GeForce RTX 4070 Ti SUPER, has launched with an MSRP of $749 - $50 cheaper than the $799 price point of these two GPUs. However, as shortages, price hikes, and limited availability have plagued the GeForce RTX 50 Series launch, you'd be hard-pressed to find one of these MSRP models a week from launch. The ASUS PRIME GeForce RTX 5070 Ti is an MSRP model. On paper, the value is undoubtedly here - 4K gaming performance that is 19% faster than the GeForce RTX 4070 Ti SUPER and an impressive 29% faster than the GeForce RTX 4070 Ti from 2023. In our suite of gaming benchmarks that include titles with ray-tracing, you're looking at RTX 5070 Ti performance that is on par with the GeForce RTX 4080 SUPER. If you look at our chart covering the average performance for 4K gaming, the ASUS PRIME GeForce RTX 5070 Ti OC Edition delivers identical performance to the GeForce RTX 4080 SUPER Founders Edition. As we're talking about one of the most impressive enthusiast-class GPUs from 2024, starting from $999, this is a win. A quick look at the ASUS PRIME GeForce RTX 5070 Ti OC Edition At this resolution, the GeForce RTX 5070 Ti also benefits from the arrival of DLSS 4 (the new transformer AI Super Resolution is available to all GeForce RTX gamers), where using the 'Balanced' or 'Performance' preset for DLSS Super Resolution upscaling delivers image quality on par with DLSS 3's 'Quality' preset. What this means is that in addition to the raw performance uplift over previous 70-class GPUs, there's essentially a free double-digit boost to performance from software alone. With the GeForce RTX 50 Series arriving on a similar custom TSMC 4N process as the GeForce RTX 40 Series, we've seen a more modest gen-on-gen improvement to raw performance overall. However, the RTX Blackwell architecture includes several improvements and advancements for Neural Rendering in addition to the latest generation of Tensor and RT Cores. Where the original GeForce RTX 4070 Ti, with its 12GB of VRAM, excelled when you set the resolution to 1440p, the GeForce RTX 5070 Ti shines in 4K. And the story becomes even more interesting when looking at games with heavy ray tracing or Full Ray Tracing, where DLSS 4's updated Ray Reconstruction denoiser is like flicking a light switch. What looked pretty good before now looks stunning. Raw performance will always be an essential metric for evaluating a GPU for PC gaming; however, with the arrival of ray-tracing and technologies like DLSS, it no longer represents the real-world performance you can expect when playing a game like Alan Wake 2 or Cyberpunk 2077 on the ASUS PRIME GeForce RTX 5070 Ti OC Edition. It's more of a baseline, a measure of 'is it enough for DLSS to take over'? We always like to highlight this, as getting lost in a sea of numbers is easy. Case in point: The GeForce RTX 50 Series marks the arrival of new RTX Neural Shaders, AI models that run in real-time to improve environment detail, assist in calculating ray bounces, compress textures, and more. With Remedy adding RTX Mega Geometry support to Alan Wake 2, Full Ray Tracing in that game now runs up to 15% faster on GeForce RTX GPUs, and it's a technology that is all about leveraging the power of RTX Blackwell's new AI hardware and architecture. Of course, none of this is relevant to a competitive title like Counter-Strike 2 - pure performance and efficiency are king. And here, the ASUS PRIME GeForce RTX 5070 Ti OC Edition delivers 313 FPS in 1440p and 180 FPS in 4K using the game's highest-quality graphics setting. RTX Blackwell - NVIDIA's Gaming Architecture for the AI Era Below is a summary of NVIDIA's GeForce RTX 50 Series and RTX Blackwell architecture, applicable to all models. NVIDIA describes 'Neural Rendering,' which includes all previous versions of DLSS and the brand-new DLSS 4, as the 'next era for computer graphics.' They're not alone; the Lead System Architect for the PlayStation 5 Pro console, Mark Cerny, recently said that ray-tracing is the future of games and that AI will play an integral role in making that happen. DOOM: The Dark Ages developer id Software shared a similar sentiment, adding that the arrival of DLSS was an 'inflection point' for PC game visuals and performance and on par with the arrival of dedicated GPUs and programmable shaders. With the arrival of the Blackwell generation and the GeForce RTX 50 Series, AI is now being used to accelerate programmable shaders with the brand-new RTX Neural Shaders. Yes, these are actual neural networks that use live game data, and the power of Tensor Cores to do everything from compress textures, render lifelike materials with a level of detail impossible to match using traditional rendering methods, and even use AI to partially trace rays and then infer "an infinite amount of rays and bounces for a more accurate representation of indirect lighting in the game scene." RTX Mega Geometry is incredible in its own right; it essentially increases a scene's geometry detail and complexity (triangles or polygons) by up to 100x. 100 times the detail, it's hard to wrap your head around - but the added benefit in a game like Alan Wake 2 is dramatically improving the performance of the game's Full Ray Tracing or Path Tracing mode. With DLSS 4 and RTX Neural Shaders, NVIDIA's GeForce RTX 50 Series and RTX Blackwell architecture (which includes the same AI optimizations as data center Blackwell) can be viewed as the turning point for PC gaming - the moment when AI becomes integral to everything from designing a game to programming and then finally rendering it on a 4K display to play. DLSS 4 includes more goodies than NVIDIA's highly touted new Multi Frame Generation technology, but let's start there. DLSS 3's version of Frame Generation has evolved with DLSS 4, powered by Blackwell hardware and software, and an innovative use of AI to generate frames 40% faster while using 30% less VRAM. Switching to a new model also means that Frame Generation and Multi-Frame Generation could soon come to GeForce RTX 20, 30, and RTX 40 Series owners. DLSS 4 benefits all GeForce RTX gamers. With the 5th Generation of Tensor Cores in the GeForce RTX 50 Series delivering 2.5X more AI performance, NVIDIA's latest GeForce RTX 50 Series GPUs can execute five complex AI models - covering Super Resolution, Ray Reconstruction, and Multi Frame Generation in a couple of milliseconds. Part of the reason it happens so quickly is the addition of hardware Flip Metering, which shifts frame pacing to the Blackwell display engine - the result is frame rates of up to 4K 240 FPS and higher without stuttering issues. With up to 15 of every 16 pixels generated by AI, the result is up to 8X the performance when compared to native rendering or rasterized performance. DLSS Super Resolution and Ray Reconstruction are also switching to a new 'Transformer' model, with over double the parameters and four times the compute requirement. This is one of the most exciting aspects of the GeForce RTX 50 Series, as it pushes DLSS into a new realm of image quality and performance. The best part is that it will work on all GeForce RTX GPUs; however, there will be a performance hit compared to running it on an RTX 50 Series GPU. Already available in games, DLSS 4's Transformer model is another DLSS 2.0-like moment for the technology, and the results speak for themselves. Even better, DLSS 4 is integrated into the NVIDIA App with a new 'DLSS Override' feature that allows users to experience the latest tech without waiting for a patch or game update. DLSS 4 is built to be backward compatible, with 75 games and apps supported (so far). It doesn't stop there, as the new AI Management Processor (AMP) allows AI models to share the GPU with graphics workloads. As a result, expect to see digital humans in games alongside AI assistants like NVIDIA's Project G-Assist becoming more prevalent in the coming years. This filters down to the creator side, with AI assistants for streamers, who will also benefit from the GeForce RTX 50 Series' expanded creator features. RTX Blackwell introduces 4:2:2 chroma-sampled video encoding and decoding. The ninth-generation NVENC encoder also improves AV1 and HEVC quality. RTX Blackwell is a game changer for creators and editors, especially with the new low-voltage and cutting-edge GDDR7 memory that dramatically improves memory bandwidth and speed. Here's a look at the specs for the GeForce RTX 5070 Ti compared to the previous generation's GeForce RTX 4070 Ti and GeForce RTX 4070 Ti SUPER, as well as the GeForce RTX 5080 and GeForce RTX 4080. The GeForce RTX 50 Series not arriving with a notable process node shrink is surprising, as this is usually the catalyst for NVIDIA to deliver new architecture with improved efficiency while adding more room for more hardware. The GeForce RTX 5070 Ti only features 16% more CUDA Cores, Tensor Cores, and RT Cores than the GeForce RTX 4070 Ti and only 6% more than the GeForce RTX 4070 Ti SUPER refresh from 2024. However, as we see up to 30% more performance when gaming in 4K, the RTX Blackwell architecture offers much more than the spec sheet suggests. The significant cutting-edge improvement here, in terms of hardware, is the shift to fast GDDR7 memory - with the 16GB on the GeForce RTX 5070 Ti delivering a massive 33% increase in memory bandwidth compared to the 16GB on the GeForce RTX 4070 Ti SUPER. Compared to the GeForce RTX 4070 Ti's 12GB of GDDR6X memory, you're looking at 77% more memory bandwidth. Memory plays an essential role in playing games in 4K, especially titles with detailed cinematic ray-traced visuals, so this jump alone could be one of the main reasons the ASUS PRIME GeForce RTX 5070 Ti OC Edition excels when bumping up the resolution to 4K. Like the GeForce RTX 5080 and RTX 5090, the RTX 5070 Ti is a PCIe Gen5 card and includes NVIDIA's latest display engine for GeForce RTX that adds support for DisplayPort 2.1 - which has enough bandwidth to support a 4K 12-bit HDR signal at a whopping 480 Hz without compression. For creators, the NVENC decoder and encoder updates mean higher-quality AV1 video for streaming and capture. Taking a slight detour from gaming and content creation, NVIDIA's continued leadership in all things AI means that the new Tensor Cores in the GeForce RTX 50 Series add support for FP4 for double the raw AI TOPS performance when compared to the GeForce RTX 4070 Ti. As an MSRP model, the ASUS PRIME GeForce RTX 5070 Ti OC Edition does see a modest boost in the reference clock speed; it ships with an additional 30 MHz out-of-the-box - which isn't enough to make a notable difference to performance. That said, there's plenty of overclocking headroom on the GeForce RTX 4070 Ti - so firing up ASUS's excellent GPU Tweak III app is an easy way to get a few extra frames in several games. When we reviewed the flagship ROG ASTRAL GeForce RTX 5080 from ASUS, we were impressed by its premium build. However, the four-fan GPU was also one of the largest and heaviest we've ever encountered, so it was a nice change of pace to slot in the ASUS PRIME GeForce RTX 5070 Ti OC Edition into our test bench. This compact 2.5-slot RGB-free GPU from ASUS represents the most affordable option in its line-up, but everything about it still feels premium and chock-full of excellent touches. For example, the power connector has a thermal pad to cool the connection and power cable. The ASUS PRIME GeForce RTX 5070 Ti includes three of the company's impressive Axial-tech fans designed for low noise and high performance, with a silent 0dB mode when the GPU temperature drops below 50 degrees Celsius. Throw in dual ball fan bearings for durability, and ASUS has crafted a GPU that will last you years. This extends to ASUS switching from thermal paste on the GPU to a Phase-Change GPU Thermal Pad, which not only offers optimal heat dissipation but will extend the life of the GPU compared to the traditional thermal paste method of GPU and heatsink contact. Design-wise, the ASUS PRIME GeForce RTX 5070 Ti is a great-looking GPU, too, with the prominent 'PRIME' branding on the curved side, the vented metal backplate, and all of the little screws, lines, and rivets giving off a look of a high-tech piece of equipment. Thermal performance is also excellent, with a dual BIOS switch offering a 'Quiet' mode that offers a less aggressive fan curve while keeping the same high-performance settings. The Games and Tests PC gaming not only covers a wide range of genres and styles, from indie games with simple 2D graphics to massive 3D worlds lit by cutting-edge real-time ray tracing technology. With that, the needs and requirements of each gamer vary. High refresh rates and latency reduction become more important than flashy visuals or playing at the highest resolution possible for those who live and breathe fast-paced competitive games. For those who want to live in a cinematic world and become a key player in an expansive narrative, ray-tracing, and high-fidelity visuals are a stepping stone toward immersion. Our chosen benchmarks cover various games, engines, APIs, and technologies. For the GeForce RTX 5070 Ti, all tests are run at 4K and 1440p and include results for performance-boosting Super Resolution technologies like NVIDIA DLSS 4 - including Frame Generation and the new Multi Frame Generation. In many ways, DLSS numbers are more important in 2025 than native rendering - a title with ray tracing isn't meant to be played without Super Resolution. Also, DLSS technologies like Ray Reconstruction and the new RTX Mega Geometry dramatically improve visual fidelity and detail compared to native rendering. However, our benchmark results are still sorted using 'raw performance' or native rendering. Here's the breakdown of games, settings, and what's being tested. The ASUS PRIME GeForce RTX 5070 Ti OC Edition is one of the most powerful and affordable 4K gaming cards, assuming you can pick one up for $749. The 80 FPS average at this resolution, captured from 14 games, makes it 29% faster than the GeForce RTX 4070 Ti, 9% faster than the Radeon RX 7900 XTX, 60% faster than the GeForce RTX 3080, and a whopping 105% faster than the GeForce RTX 3070 Ti. These two Ampere GPUs are still some of the most used among PC gamers today, and as it makes more sense to upgrade every other generation, the ASUS PRIME GeForce RTX 5070 Ti OC Edition is a fantastic choice. With the DLSS 'Quality' preset for Super Resolution, the 80 FPS average becomes 101 FPS, so you're looking at 100+ FPS in many games. When you factor in the better-than-native image quality of DLSS 4's new transformer model, this makes the ASUS PRIME GeForce RTX 5070 Ti OC Edition a 4K gaming beast. DLSS is one of those must-have pieces of technology and a key reason to pick up a GeForce RTX GPU, and the arrival of DLSS 4 puts a lot of pressure on AMD to deliver with FSR 4. Naturally, this review is written from the perspective of not knowing how the upcoming Radeon RX 9070 XT will perform - but we will be comparing RDNA 4's flagship to the GeForce RTX 5070 Ti in the coming weeks, so stay tuned for that. Looking at the individual game results and the ASUS PRIME GeForce RTX 5070 Ti's performance uplift over the GeForce RTX 4070 Ti and other cards varies. For example, Call of Duty: Black Ops 6 runs 19% faster in 4K on the Radeon RX 7900 XTX - one game where Radeon GPUs have a clear advantage. However, jump over to Cyberpunk 2077 using the RT Ultra preset, and the ASUS PRIME GeForce RTX 5070 Ti is 50% faster than the Radeon RX 7900 XTX. Cyberpunk 2077 sees significant gains on the GeForce RTX 50 Series; without ray-tracing, the ASUS PRIME GeForce RTX 5070 Ti is 44% faster than the GeForce RTX 4070 Ti at this resolution. Turn on ray-tracing, and this number drops to 30%, which is still impressive. However, jump over to DOOM Eternal in 4K, and the RTX 5070 Ti is only 10% faster than the GeForce RTX 4070 Ti SUPER. Average Gaming Performance - 1440p Results With impressive 4K gaming performance, the ASUS PRIME GeForce RTX 5070 Ti OC Edition is also an excellent option for high-refresh-rate 1440p gaming or pairing with an ultrawide display with a resolution and pixel count between 1440p and 4K. However, there are diminishing returns at this resolution as you run into bottlenecks or other limitations. For example, on our test system, Warhammer 40K: Space Marine 2 runs just as fast on the ASUS PRIME GeForce RTX 5070 Ti as it does on the RTX 4070 Ti SUPER, RTX 4080, RTX 4090, RTX 5080, and RTX 5090. This is the exception rather than the rule. Still, with fewer gains, you're looking at average 1440p gaming performance that is now only 20% faster than the GeForce RTX 4070 Ti, 16% faster than the GeForce RTX 4070 Ti SUPER, 77% faster than the GeForce RTX 3070 Ti, and 5% faster than the Radeon RX 7900 XTX. However, the ASUS PRIME GeForce RTX 5070 Ti's 1440p gaming performance is still on par with the GeForce RTX 4080 SUPER, so that means you're getting 93 FPS in Dragon Age: The Veilguard with ray-tracing and 121 FPS in Horizon Forbidden West. Benchmarks - 3DMark Synthetic Tests 3DMark offers a suite of synthetic benchmarks built to test GPUs in various scenarios. 3DMark Steel Nomad is a cutting-edge DirectX 12 benchmark with newer, modern rendering techniques designed to push GPUs to their limit. The 'Light' version tests at 1440p, while the main Steel Nomad benchmark tests pure native 4K rendering. Port Royal is a benchmark focusing exclusively on real-time ray tracing for lighting effects like reflections, shadows, and more. The 4K-based Steel Nomad test, which is an official benchmark from 3DMark that replaced the tried and true Time Spy test, is turning out to be a great indicator of how the GeForce RTX 50 Series performs in modern AAA titles. Here, the ASUS PRIME GeForce RTX 5070 Ti delivers a score 31% higher than the GeForce RTX 4070 Ti and 20% higher than the GeForce RTX 4070 Ti SUPER. Even though it sits slightly lower in the hierarchy than the GeForce RTX 4080 and RTX 4080 SUPER, the score is close enough that it matches what we found in the newer and more visually intensive games in our benchmark suite. The 3DMark Port Royal results are interesting because it puts the ray-tracing performance of the ASUS PRIME GeForce RTX 5070 Ti slightly higher than the GeForce RTX 4080 SUPER, which is not the case for a number of games with ray-tracing. In titles with heavy ray-tracing or Full Ray Tracing, the GeForce RTX 4080 SUPER slightly edges out the ASUS PRIME GeForce RTX 5070 Ti. However, as you're looking at performance that sits within the margin of error, this is neither here nor there, as both cards are excellent options for RT gaming. DLSS 4, Frame Generation, and the new Multi-Frame Generation DLSS 4 and Multi Frame Generation are impressive bits of technology, thanks mainly to the overall improvements to performance and latency on the Frame Generation side and the new 'Transformer' model for Super Resolution and Ray Reconstruction. We used the DLSS 'Quality' mode preset for these benchmarks, often delivering better-than-native image quality. Multi Frame Generation, which is exclusive to the new GeForce RTX 50 Series, builds on the improved Frame Generation performance of the DLSS 4 suite of AI technologies by offering you the option to generate up to three additional frames. It's impressive, as the increased smoothness and motion clarity means you can push performance to match the refresh rate of your display - which makes it perfect for 1440p or 4K monitors with 200+ Hz to play with, like the MSI MAG 321UPX QD-OLED 4K 240 Hz display we use for all benchmarking. But it's a technology that needs a decent input frame rate to do its magic, which for us is around 70 to 80 FPS, as anything lower can lead to some noticeable latency with controls. The quality of the AI-generated frames has improved with DLSS 4; however, as Multi Frame Generation 4X means three out of every four frames are AI-generated, some artifacts can degrade image clarity and stability. Alongside a slight increase in latency, this is the trade-off for the smoothness you get - making it more suitable for single-player or cinematic games with ray-tracing features. Multi Frame Generation is one of those 'seeing is believing' pieces of technology, and it's worth turning on in a game like Cyberpunk 2077 or Dragon Age: The Veilguard, as you can immediately see and feel the benefit. Path Tracing Performance - 1440p Path Tracing, or Full Ray Tracing, arrived with the GeForce RTX 40 Series and DLSS 3 and is leveling up with the GeForce RTX 50 Series and DLSS 4. It's only possible thanks to AI technologies like DLSS Super Resolution, Ray Reconstruction, and RTX Neural Shader technology like RTX Mega Geometry. It's designed specifically for these technologies, and we're only including native or rasterized performance to highlight just how intensive it is on a GPU as powerful as the GeForce RTX 5070 Ti. In fact, outside of the massive increase in performance, these games also look notably worse without DLSS 4. Remedy's Alan Wake 2 features some of the most technically advanced visuals ever seen in a PC game, with Full Ray Tracing, RTX Mega Geometry, DLSS 4's new transformer model for Super Resolution and Ray Reconstruction, and support for both Frame Generation 2X and Multi Frame Generation 4X. Playing the game without DLSS significantly degrades Path Tracing image quality so the 46 FPS native rendering you see here is with using the excellent DLAA mode. Getting 85 FPS without Frame Generation is very impressive, as it offers a crisp, smooth, and responsive experience. Here, the ASUS PRIME GeForce RTX 5070 Ti's performance is around 20% faster than the GeForce RTX 4070 Ti and RTX 4070 Ti SUPER - making it a better overall GPU for pushing cutting-edge visuals. For the ASUS PRIME GeForce RTX 5070 Ti, we finally benchmarked and tested Indiana Jones and the Great Circle with the game's latest DLSS 4 update that adds support for the new transformer model for both Super Resolution and Ray Reconstruction. It's a game-changer at 1440p, as it adds a lot more detail to all lighting with improved ray-traced shadows. The fact that you can get 100+ FPS with DLSS Super Resolution is fantastic, as the more advanced model is more hardware-intensive. The RTX 4070 Ti and RTX 4070 Ti SUPER results use the older CNN model for DLSS, so you're still looking at a performance improvement of around 20%. Temperature and Power Efficiency The ASUS PRIME GeForce RTX 5070 Ti OC Edition features a relatively quiet fan curve. With the GPU temperature topping out at around 65 degrees Celsius in our stress test, there's still room to overclock and push performance even higher. The efficiency is also impressive, making it an RTX Blackwell GPU that can do more with less - which is great to see after the 575W beast that is the GeForce RTX 5090. Final Thoughts As an MSRP GPU, the ASUS PRIME GeForce RTX 5070 Ti is an excellent choice for those looking for a powerful GPU to handle 1440p and 4K gaming with detail settings pushed to their limit with ray-tracing enabled. Looking at the GeForce RTX 50 Series line-up, so far, it's also the card that benefits the most from DLSS 4's new transformer model for Super Resolution as it allows you to use the 'Balanced' or 'Performance' preset and get better image quality than native rendering. Everything looks cleaner, sharper, and more detailed - and this feature alone is a game changer as it offers an additional double-digit gain to performance on top of the up to 30% uplift you get over the GeForce RTX 4070 Ti. ASUS's PRIME design is also compact, clean, and stylish. Thanks to its high-quality build, it runs quiet and cool. The GeForce RTX 5070 Ti represents the best value GPU for 4K gaming at its price point. However, with the upcoming release of AMD's Radeon RX 9070 XT around the corner, it will be interesting to see how the two stack up in a head-to-head battle. AMD has a lot of ground to make up when it comes to ray-tracing and FSR 4 image quality, so if you're a gamer who loves playing cinematic titles with stunning visuals - odds are that GeForce RTX 50 Series GPUs like the RTX 5070 Ti will still have the edge here. In the end, we want to see competition, especially if it means pricing for the GeForce RTX 5070 Ti will sit at around $750 USD for this reason. In a perfect world, we would have loved to have seen the GeForce RTX 5070 Ti launch with a price point of around $650, as this would have opened up a new level of performance and ray-traced visuals to more gamers. However, with the prices of everything going up, that might not be realistic. The ASUS PRIME GeForce RTX 5070 Ti is an impressive GPU if you can pick one up at a reasonable price point. With its advanced AI capabilities and the arrival of RTX Neural Shaders, it's poised to power next-gen PC games.

XFX Radeon RX 9070 QuickSilver OC review

Introduction The GPU reviewed for this article is formidable in cooling, noise levels as well as performance. Meet the Radeon RX 9070 from XFX, in particular the QuickSilver OC edition. It runs at a boost clock of 2.70 GHz while consuming 250W of power, striking a clear balance between processing capability and energy use. It packs 16GB of GDDR6 memory, which means it can handle high-resolution textures and complex graphics tasks without issue. With a PCIe 5.0 x16 slot, this card is set up to keep data moving quickly, an essential factor in today's computing tasks. Its dimensions are 35.5 cm by 14 cm by 7.4 cm, making it a well-proportioned piece of hardware designed for modern systems. AMD chose to skip the RX 8000 series naming, opting for a streamlined product lineup that emphasizes mainstream and high-end options, also the naming schema now matches the Ryzen series (9000). This initial GPU series includes both the RX 9070 and RX 9070 XT models. Products that sit in a 500 USD pricing space, simplifying the selection process for consumers looking for a balance between performance and affordability. The RX 9070 XT offers ray tracing capabilities that now rival NVIDIA's RTX 4080, targeting gamers requiring high-end performance with sweet eye-candy while at it. This is significant as AMD has reduced its focus on uber-high-performance GPUs. Instead, AMD reaffirmed its commitment to the high-end market by advancing its RDNA 4-based GPUs. Aside from the new GPUs, AMD's upcoming release schedule includes FidelityFX Super Resolution 4 and Ryzen 9 9000X3D processors with additional cores, all scheduled for the end of January 2025. So yeah, we have a lot to talk about and show you. First let's glimpse over the specifications. The Radeon RX 9070 series comes in two flavors: the RX 9070 and the RX 9070 XT. The most striking improvement lies in the number of AI and ray tracing (RT) accelerators, which have seen a noticeable boost compared to previous generations. The RX 9070 XT packs 64 RDNA 4 compute units, 64 RT accelerators, and 128 AI accelerators, making it a clear step up from its non-XT sibling. AMD's increased investment in AI-driven computing is evident, with AI TOPS performance reaching 1557 TOPS for the XT model. While NVIDIA has long held an advantage in AI and machine learning workloads with its Tensor Cores, AMD seems to be making a strong effort to close the gap. This could have major implications for AI-assisted gaming, upscaling technologies like FSR (FidelityFX Super Resolution), and content creation applications. On the raw speed front, AMD has pushed the boost clock to an impressive 2.97 GHz on the RX 9070 XT, while the RX 9070 settles at 2.54 GHz. This indicates that AMD is optimizing for higher clock speeds, potentially improving gaming performance, especially in high refresh rate scenarios. However, the power consumption numbers tell an interesting story. The RX 9070 XT draws 304W, which is significantly higher than the 220W of the standard RX 9070. This suggests that AMD is pushing the XT model to its limits, possibly in an effort to keep up with (or surpass) NVIDIA's upcoming RTX 5000 series. Gamers will need a capable power supply, and cooling solutions will play a crucial role in maintaining performance. Both the RX 9070 and RX 9070 XT come with 16GB of GDDR6 video memory, which should be adequate for modern gaming at 1440p and 4K resolutions. AMD is keeping up with modern connectivity standards by supporting PCIe 5.0 x16, which ensures these GPUs are future-proofed for high-speed data transfer. While PCIe 4.0 has been more than sufficient for gaming so far, PCIe 5.0 could be a game-changer for high-bandwidth workloads like AI training and video editing on the long term. On the display front, DisplayPort 2.1a and HDMI 2.1b support means gamers can enjoy higher refresh rates, better HDR support, and 8K gaming capabilities. This is particularly relevant for those investing in the next generation of high-refresh-rate 4K and ultrawide monitors. AMD has shown ambition with the RX 9070 lineup, doubling down on AI acceleration, ray tracing, and high clock speeds. While it's too early to declare these GPUs as NVIDIA-killers, they certainly have the potential to shake up the market. Specification AMD Radeon RX 9070 XT AMD Radeon RX 9070 Architecture RDNA 4 RDNA 4 Manufacturing Process 4nm 4nm Transistor Count 53.9 billion 53.9 billion Die Size 357 mm² 357 mm² Compute Units 64 56 Ray Accelerators 64 56 AI Accelerators 128 112 Stream Processors 4096 3584 Game GPU Clock 2400 MHz 2070 MHz Boost GPU Clock Up to 2970 MHz Up to 2520 MHz Peak Single Precision Throughput Up to 48.7 TFLOPS Up to 36.1 TFLOPS Peak Half Precision Throughput Up to 97.3 TFLOPS Up to 72.3 TFLOPS Peak INT8 AI TOPS Throughput Up to 779 TOPS w/ Sparsity Up to 578 TOPS w/ Sparsity Peak INT4 AI TOPS Throughput Up to 1557 TOPS w/ Sparsity Up to 1156 TOPS w/ Sparsity Peak Texture Fill-Rate Up to 730.3 GT/s Up to 564.5 GT/s ROPs 128 128 Peak Pixel Fill-Rate Up to 190.1 GP/s Up to 161.3 GP/s AMD Infinity Cache™ 64 MB (3rd. Gen.) 64 MB (3rd. Gen.) Memory 16GB GDDR6 16GB GDDR6 Memory Speed 20 Gbps 20 Gbps Memory Bus Interface 256-bit 256-bit PCIe Interface PCIe 5.0 x16 PCIe 5.0 x16 Total Board Power 304 W 220 W Radeon RX 9070 comes with the Navi 48 GPU, sporting a rectangular design that measures about 30 by 13 mm and covers roughly 390 mm². That area estimate is based on a TSMC N4 process. It packs 56 compute units -- just a bit less than the 60 CUs found in Navi 32 -- while sticking to a 256-bit memory interface. The extra space on the die likely goes toward improved ray tracing accelerators and AI features, which power stuff like FidelityFX Super Resolution 4. With RDNA 4, AMD revamps the compute units to boost instructions per clock and push clock speeds higher compared to RDNA 3. Under normal conditions, the Radeon RX 9070 is expected to match the performance of a Radeon RX 7900, and it might pull ahead when heavy ray tracing comes into play. Clock speeds fall between 2.07 GHz and 2.54 GHz, delivering the horsepower modern games and apps need, while graphics memory chugs along at an effective 20 Gbps.

Sapphire Radeon RX 9070 Pulse review

Introduction The Sapphire Pulse version of the Radeon RX 9070 is built on an advanced RDNA 4 architecture, engineered to meet the demands of modern graphical workloads. This GPU operates with a boost clock of 2.54 GHz and has a power consumption of 220W, establishing a technical balance between performance and energy efficiency. It is equipped with 16GB of GDDR6 video memory, ensuring the smooth handling of complex textures and high-resolution assets during intensive computing tasks. The device supports PCIe 5.0 x16, which offers improved data transfer rates essential for next-generation applications and alleviating bottlenecks in data-intensive operations. In addition, it provides versatile connectivity options, featuring DisplayPort 2.1a and HDMI 2.1b interfaces that support high refresh rate displays, advanced HDR, and various 4K configurations. AMD chose to skip the RX 8000 series naming, opting for a streamlined product lineup that emphasizes mainstream and high-end options, also the naming schema now matches the Ryzen series (9000). This initial GPU series includes both the RX 9070 and RX 9070 XT models. Products that sit in a 500 USD pricing space, simplifying the selection process for consumers looking for a balance between performance and affordability. The RX 9070 XT offers ray tracing capabilities that now rival NVIDIA's RTX 4080, targeting gamers requiring high-end performance with sweet eye-candy while at it. This is significant as AMD has reduced its focus on uber-high-performance GPUs. Instead, AMD reaffirmed its commitment to the high-end market by advancing its RDNA 4-based GPUs. Aside from the new GPUs, AMD's upcoming release schedule includes FidelityFX Super Resolution 4 and Ryzen 9 9000X3D processors with additional cores, all scheduled for the end of January 2025. So yeah, we have a lot to talk about and show you. First let's glimpse over the specifications. The Radeon RX 9070 series comes in two flavors: the RX 9070 and the RX 9070 XT. The most striking improvement lies in the number of AI and ray tracing (RT) accelerators, which have seen a noticeable boost compared to previous generations. The RX 9070 XT packs 64 RDNA 4 compute units, 64 RT accelerators, and 128 AI accelerators, making it a clear step up from its non-XT sibling. AMD's increased investment in AI-driven computing is evident, with AI TOPS performance reaching 1557 TOPS for the XT model. While NVIDIA has long held an advantage in AI and machine learning workloads with its Tensor Cores, AMD seems to be making a strong effort to close the gap. This could have major implications for AI-assisted gaming, upscaling technologies like FSR (FidelityFX Super Resolution), and content creation applications. On the raw speed front, AMD has pushed the boost clock to an impressive 2.97 GHz on the RX 9070 XT, while the RX 9070 settles at 2.54 GHz. This indicates that AMD is optimizing for higher clock speeds, potentially improving gaming performance, especially in high refresh rate scenarios. However, the power consumption numbers tell an interesting story. The RX 9070 XT draws 304W, which is significantly higher than the 220W of the standard RX 9070. This suggests that AMD is pushing the XT model to its limits, possibly in an effort to keep up with (or surpass) NVIDIA's upcoming RTX 5000 series. Gamers will need a capable power supply, and cooling solutions will play a crucial role in maintaining performance. Both the RX 9070 and RX 9070 XT come with 16GB of GDDR6 video memory, which should be adequate for modern gaming at 1440p and 4K resolutions. AMD is keeping up with modern connectivity standards by supporting PCIe 5.0 x16, which ensures these GPUs are future-proofed for high-speed data transfer. While PCIe 4.0 has been more than sufficient for gaming so far, PCIe 5.0 could be a game-changer for high-bandwidth workloads like AI training and video editing on the long term. On the display front, DisplayPort 2.1a and HDMI 2.1b support means gamers can enjoy higher refresh rates, better HDR support, and 8K gaming capabilities. This is particularly relevant for those investing in the next generation of high-refresh-rate 4K and ultrawide monitors. AMD has shown ambition with the RX 9070 lineup, doubling down on AI acceleration, ray tracing, and high clock speeds. While it's too early to declare these GPUs as NVIDIA-killers, they certainly have the potential to shake up the market. Specification AMD Radeon RX 9070 XT AMD Radeon RX 9070 Architecture RDNA 4 RDNA 4 Manufacturing Process 4nm 4nm Transistor Count 53.9 billion 53.9 billion Die Size 357 mm² 357 mm² Compute Units 64 56 Ray Accelerators 64 56 AI Accelerators 128 112 Stream Processors 4096 3584 Game GPU Clock 2400 MHz 2070 MHz Boost GPU Clock Up to 2970 MHz Up to 2520 MHz Peak Single Precision Throughput Up to 48.7 TFLOPS Up to 36.1 TFLOPS Peak Half Precision Throughput Up to 97.3 TFLOPS Up to 72.3 TFLOPS Peak INT8 AI TOPS Throughput Up to 779 TOPS w/ Sparsity Up to 578 TOPS w/ Sparsity Peak INT4 AI TOPS Throughput Up to 1557 TOPS w/ Sparsity Up to 1156 TOPS w/ Sparsity Peak Texture Fill-Rate Up to 730.3 GT/s Up to 564.5 GT/s ROPs 128 128 Peak Pixel Fill-Rate Up to 190.1 GP/s Up to 161.3 GP/s AMD Infinity Cache™ 64 MB (3rd. Gen.) 64 MB (3rd. Gen.) Memory 16GB GDDR6 16GB GDDR6 Memory Speed 20 Gbps 20 Gbps Memory Bus Interface 256-bit 256-bit PCIe Interface PCIe 5.0 x16 PCIe 5.0 x16 Total Board Power 304 W 220 W Radeon RX 9070 comes with the Navi 48 GPU, sporting a rectangular design that measures about 30 by 13 mm and covers roughly 390 mm². That area estimate is based on a TSMC N4 process. It packs 56 compute units -- just a bit less than the 60 CUs found in Navi 32 -- while sticking to a 256-bit memory interface. The extra space on the die likely goes toward improved ray tracing accelerators and AI features, which power stuff like FidelityFX Super Resolution 4. With RDNA 4, AMD revamps the compute units to boost instructions per clock and push clock speeds higher compared to RDNA 3. Under normal conditions, the Radeon RX 9070 is expected to match the performance of a Radeon RX 7900, and it might pull ahead when heavy ray tracing comes into play. Clock speeds fall between 2.07 GHz and 2.54 GHz, delivering the horsepower modern games and apps need, while graphics memory chugs along at an effective 20 Gbps.

XFX Radeon RX 9070 XT Mercury Magnetic review

Introduction XFX has been quietly preparing for the launch of its Radeon RX 9070 series, and yeah, we've had their card laying around ever since January already. The high-end RX 9070 XT cards, part of the MERCURY series, will reportedly come in four different versions. Two of these models sport Magnetic Air cooling which we review - they're marked with an "M" for Magnetic Air on the box - while the other two use regular fan designs. The cards have a factory tweak at around 3.1 GHz and three 8-pin power connectors, indicating a solid focus on reliable power delivery and cooling. Looking further into the lineup, there are also hints of additional models in the RX 9070 series. The QUICK SILVER range appears to include three models: one white Magnetic Air option and two versions (one black and one white) that use standard fans. These models skip the RGB lighting but borrow design cues from the premium MERCURY series. Additionally, there are two SWIFT models, available in black and white, which seem geared towards those who need a straightforward, budget-friendly option. AMD chose to skip the RX 8000 series naming, opting for a streamlined product lineup that emphasizes mainstream and high-end options, also the naming schema now matches the Ryzen series (9000). This initial GPU series includes both the RX 9070 and RX 9070 XT models. Products that sit in a 500 USD pricing space, simplifying the selection process for consumers looking for a balance between performance and affordability. The RX 9070 XT offers ray tracing capabilities that now rival NVIDIA's RTX 4080, targeting gamers requiring high-end performance with sweet eye-candy while at it. This is significant as AMD has reduced its focus on uber-high-performance GPUs. Instead, AMD reaffirmed its commitment to the high-end market by advancing its RDNA 4-based GPUs. Aside from the new GPUs, AMD's upcoming release schedule includes FidelityFX Super Resolution 4 and Ryzen 9 9000X3D processors with additional cores, all scheduled for the end of January 2025. So yeah, we have a lot to talk about and show you. First let's glimpse over the specifications. The Radeon RX 9070 series comes in two flavors: the RX 9070 and the RX 9070 XT. The most striking improvement lies in the number of AI and ray tracing (RT) accelerators, which have seen a noticeable boost compared to previous generations. The RX 9070 XT packs 64 RDNA 4 compute units, 64 RT accelerators, and 128 AI accelerators, making it a clear step up from its non-XT sibling. AMD's increased investment in AI-driven computing is evident, with AI TOPS performance reaching 1557 TOPS for the XT model. While NVIDIA has long held an advantage in AI and machine learning workloads with its Tensor Cores, AMD seems to be making a strong effort to close the gap. This could have major implications for AI-assisted gaming, upscaling technologies like FSR (FidelityFX Super Resolution), and content creation applications. On the raw speed front, AMD has pushed the boost clock to an impressive 2.97 GHz on the RX 9070 XT, while the RX 9070 settles at 2.54 GHz. This indicates that AMD is optimizing for higher clock speeds, potentially improving gaming performance, especially in high refresh rate scenarios. However, the power consumption numbers tell an interesting story. The RX 9070 XT draws 304W, which is significantly higher than the 220W of the standard RX 9070. This suggests that AMD is pushing the XT model to its limits, possibly in an effort to keep up with (or surpass) NVIDIA's upcoming RTX 5000 series. Gamers will need a capable power supply, and cooling solutions will play a crucial role in maintaining performance. Both the RX 9070 and RX 9070 XT come with 16GB of GDDR6 video memory, which should be adequate for modern gaming at 1440p and 4K resolutions. AMD is keeping up with modern connectivity standards by supporting PCIe 5.0 x16, which ensures these GPUs are future-proofed for high-speed data transfer. While PCIe 4.0 has been more than sufficient for gaming so far, PCIe 5.0 could be a game-changer for high-bandwidth workloads like AI training and video editing on the long term. On the display front, DisplayPort 2.1a and HDMI 2.1b support means gamers can enjoy higher refresh rates, better HDR support, and 8K gaming capabilities. This is particularly relevant for those investing in the next generation of high-refresh-rate 4K and ultrawide monitors. AMD has shown ambition with the RX 9070 lineup, doubling down on AI acceleration, ray tracing, and high clock speeds. While it's too early to declare these GPUs as NVIDIA-killers, they certainly have the potential to shake up the market. Reference Specification AMD Radeon RX 9070 XT AMD Radeon RX 9070 Architecture RDNA 4 RDNA 4 Manufacturing Process 4nm 4nm Transistor Count 53.9 billion 53.9 billion Die Size 357 mm² 357 mm² Compute Units 64 56 Ray Accelerators 64 56 AI Accelerators 128 112 Stream Processors 4096 3584 Game GPU Clock 2400 MHz 2070 MHz Boost GPU Clock Up to 2970 MHz Up to 2520 MHz Peak Single Precision Throughput Up to 48.7 TFLOPS Up to 36.1 TFLOPS Peak Half Precision Throughput Up to 97.3 TFLOPS Up to 72.3 TFLOPS Peak INT8 AI TOPS Throughput Up to 779 TOPS w/ Sparsity Up to 578 TOPS w/ Sparsity Peak INT4 AI TOPS Throughput Up to 1557 TOPS w/ Sparsity Up to 1156 TOPS w/ Sparsity Peak Texture Fill-Rate Up to 730.3 GT/s Up to 564.5 GT/s ROPs 128 128 Peak Pixel Fill-Rate Up to 190.1 GP/s Up to 161.3 GP/s AMD Infinity Cache™ 64 MB (3rd. Gen.) 64 MB (3rd. Gen.) Memory 16GB GDDR6 16GB GDDR6 Memory Speed 20 Gbps 20 Gbps Memory Bus Interface 256-bit 256-bit PCIe Interface PCIe 5.0 x16 PCIe 5.0 x16 Total Board Power 304 W 220 W Radeon RX 9070 XT comes with the Navi 48 GPU, sporting a rectangular design that measures about 30 by 13 mm and covers roughly 390 mm². That area estimate is based on a TSMC N4 process. It packs 64 compute units -- just a bit more than the 60 CUs found in Navi 32 -- while sticking to a 256-bit memory interface. The extra space on the die likely goes toward improved ray tracing accelerators and AI features, which power stuff like FidelityFX Super Resolution 4. With RDNA 4, AMD revamps the compute units to boost instructions per clock and push clock speeds higher compared to RDNA 3. Under normal conditions, the Radeon RX 9070 XT is expected to match the performance of a Radeon RX 7900 XT, and it might pull ahead when heavy ray tracing comes into play. Clock speeds fall between 2.5 GHz and 2.97 GHz, delivering the horsepower modern games and apps need, while graphics memory chugs along at an effective 20 Gbps.

SAPPHIRE Radeon RX 9070 XT NITRO+ review

Introduction The Sapphire NITRO+ model based on the RX 9070 XT SKU represents a carefully engineered graphics card designed with a focus on efficient thermal management and robust construction. Its design incorporates a mesh structure that facilitates improved airflow throughout the unit, ensuring that heat is efficiently dissipated during intensive workloads. An innovative aspect of this model is the discreet placement of the 16‑pin power connector, which is concealed behind a removable backplate. This also is the first instance of a Radeon model integrating a 16‑pin connector, a design choice that reflects an evolution in power delivery architecture. Despite the new connector configuration, the card's overall power consumption is engineered to remain below 375W. From a physical and mechanical perspective, the Sapphire NITRO+ is built with a three-slot thickness and equipped with three dedicated cooling fans. This multi-slot design is complemented by an integrated mesh panel that not only enhances the airflow but also contributes to the stability of operating temperatures over extended periods of high-load operation. The design is underpinned by several technical innovations, including the Free Flow Cooling Design, which is intended to optimize the movement of air through the system, and the use of high-conductivity materials like Honeywell's PTM7950 TIM. Additionally, the integrated cooling module, featuring AeroCurve fan blades and optimized composite heatpipes, is designed to uniformly distribute and dissipate heat across critical components. Such a configuration ensures that even during demanding tasks -- whether for gaming, content creation, or overclocking -- the card maintains consistent performance and reliability. Numerous technical measurements and images validate these design choices, providing clarity regarding the construction and intended operational performance of the card. Speaking of performance, this card clocks in at a Boost frequency of 3060 MHz with a 350W TBP. AMD chose to skip the RX 8000 series naming, opting for a streamlined product lineup that emphasizes mainstream and high-end options, also the naming schema now matches the Ryzen series (9000). This initial GPU series includes both the RX 9070 and RX 9070 XT models. Products that sit in a 500 USD pricing space, simplifying the selection process for consumers looking for a balance between performance and affordability. The RX 9070 XT offers ray tracing capabilities that now rival NVIDIA's RTX 4080, targeting gamers requiring high-end performance with sweet eye-candy while at it. This is significant as AMD has reduced its focus on uber-high-performance GPUs. Instead, AMD reaffirmed its commitment to the high-end market by advancing its RDNA 4-based GPUs. Aside from the new GPUs, AMD's upcoming release schedule includes FidelityFX Super Resolution 4 and Ryzen 9 9000X3D processors with additional cores, all scheduled for the end of January 2025. So yeah, we have a lot to talk about and show you. First let's glimpse over the specifications. The Radeon RX 9070 series comes in two flavors: the RX 9070 and the RX 9070 XT. The most striking improvement lies in the number of AI and ray tracing (RT) accelerators, which have seen a noticeable boost compared to previous generations. The RX 9070 XT packs 64 RDNA 4 compute units, 64 RT accelerators, and 128 AI accelerators, making it a clear step up from its non-XT sibling. AMD's increased investment in AI-driven computing is evident, with AI TOPS performance reaching 1557 TOPS for the XT model. While NVIDIA has long held an advantage in AI and machine learning workloads with its Tensor Cores, AMD seems to be making a strong effort to close the gap. This could have major implications for AI-assisted gaming, upscaling technologies like FSR (FidelityFX Super Resolution), and content creation applications. On the raw speed front, AMD has pushed the boost clock to an impressive 2.97 GHz on the RX 9070 XT, while the RX 9070 settles at 2.54 GHz. This indicates that AMD is optimizing for higher clock speeds, potentially improving gaming performance, especially in high refresh rate scenarios. However, the power consumption numbers tell an interesting story. The RX 9070 XT draws 304W, which is significantly higher than the 220W of the standard RX 9070. This suggests that AMD is pushing the XT model to its limits, possibly in an effort to keep up with (or surpass) NVIDIA's upcoming RTX 5000 series. Gamers will need a capable power supply, and cooling solutions will play a crucial role in maintaining performance. Both the RX 9070 and RX 9070 XT come with 16GB of GDDR6 video memory, which should be adequate for modern gaming at 1440p and 4K resolutions. AMD is keeping up with modern connectivity standards by supporting PCIe 5.0 x16, which ensures these GPUs are future-proofed for high-speed data transfer. While PCIe 4.0 has been more than sufficient for gaming so far, PCIe 5.0 could be a game-changer for high-bandwidth workloads like AI training and video editing on the long term. On the display front, DisplayPort 2.1a and HDMI 2.1b support means gamers can enjoy higher refresh rates, better HDR support, and 8K gaming capabilities. This is particularly relevant for those investing in the next generation of high-refresh-rate 4K and ultrawide monitors. AMD has shown ambition with the RX 9070 lineup, doubling down on AI acceleration, ray tracing, and high clock speeds. While it's too early to declare these GPUs as NVIDIA-killers, they certainly have the potential to shake up the market. Reference Specification AMD Radeon RX 9070 XT AMD Radeon RX 9070 Architecture RDNA 4 RDNA 4 Manufacturing Process 4nm 4nm Transistor Count 53.9 billion 53.9 billion Die Size 357 mm² 357 mm² Compute Units 64 56 Ray Accelerators 64 56 AI Accelerators 128 112 Stream Processors 4096 3584 Game GPU Clock 2400 MHz 2070 MHz Boost GPU Clock Up to 2970 MHz Up to 2520 MHz Peak Single Precision Throughput Up to 48.7 TFLOPS Up to 36.1 TFLOPS Peak Half Precision Throughput Up to 97.3 TFLOPS Up to 72.3 TFLOPS Peak INT8 AI TOPS Throughput Up to 779 TOPS w/ Sparsity Up to 578 TOPS w/ Sparsity Peak INT4 AI TOPS Throughput Up to 1557 TOPS w/ Sparsity Up to 1156 TOPS w/ Sparsity Peak Texture Fill-Rate Up to 730.3 GT/s Up to 564.5 GT/s ROPs 128 128 Peak Pixel Fill-Rate Up to 190.1 GP/s Up to 161.3 GP/s AMD Infinity Cache™ 64 MB (3rd. Gen.) 64 MB (3rd. Gen.) Memory 16GB GDDR6 16GB GDDR6 Memory Speed 20 Gbps 20 Gbps Memory Bus Interface 256-bit 256-bit PCIe Interface PCIe 5.0 x16 PCIe 5.0 x16 Total Board Power 304 W 220 W Radeon RX 9070 XT comes with the Navi 48 GPU, sporting a rectangular design that measures about 30 by 13 mm and covers roughly 390 mm². That area estimate is based on a TSMC N4 process. It packs 64 compute units -- just a bit more than the 60 CUs found in Navi 32 -- while sticking to a 256-bit memory interface. The extra space on the die likely goes toward improved ray tracing accelerators and AI features, which power stuff like FidelityFX Super Resolution 4. With RDNA 4, AMD revamps the compute units to boost instructions per clock and push clock speeds higher compared to RDNA 3. Under normal conditions, the Radeon RX 9070 XT is expected to match the performance of a Radeon RX 7900 XT, and it might pull ahead when heavy ray tracing comes into play. Clock speeds fall between 2.5 GHz and 2.97 GHz, delivering the horsepower modern games and apps need, while graphics memory chugs along at an effective 20 Gbps.

Radeon RX 9070 XT (ASUS TUF OC) review

Introduction It's March 2025, and AMD is knocking on our door with a load of new hardware. At CES 2025 AMD hinted at the Radeon RX 9070 XT, part of its latest RX 9000 series based on RDNA 4 architecture. The launch is positioned to compete with NVIDIA's RTX 4070, 4080 and 5070 Ti, and comes at a time of low market activity right after the Christmas 2024 season. First up this, if you were expecting the Radeon RX 8000 series as release, you're wrong. AMD chose to skip the RX 8000 series naming, opting for a streamlined product lineup that emphasizes mainstream and high-end options, also the naming schema now matches the Ryzen series (9000). This initial GPU series includes both the RX 9070 and RX 9070 XT models. Products that sit in a 500 USD pricing space, simplifying the selection process for consumers looking for a balance between performance and affordability. The RX 9070 XT offers ray tracing capabilities that now rival NVIDIA's RTX 4080, targeting gamers requiring high-end performance with sweet eye-candy while at it. This is significant as AMD has reduced its focus on uber-high-performance GPUs. Instead, AMD reaffirmed its commitment to the high-end market by advancing its RDNA 4-based GPUs. Aside from the new GPUs, AMD's upcoming release schedule includes FidelityFX Super Resolution 4 and Ryzen 9 9000X3D processors with additional cores, all scheduled for the end of January 2025. So yeah, we have a lot to talk about and show you. First let's glimpse over the specifications. The Radeon RX 9070 series comes in two flavors: the RX 9070 and the RX 9070 XT. The most striking improvement lies in the number of AI and ray tracing (RT) accelerators, which have seen a noticeable boost compared to previous generations. The RX 9070 XT packs 64 RDNA 4 compute units, 64 RT accelerators, and 128 AI accelerators, making it a clear step up from its non-XT sibling. AMD's increased investment in AI-driven computing is evident, with AI TOPS performance reaching 1557 TOPS for the XT model. While NVIDIA has long held an advantage in AI and machine learning workloads with its Tensor Cores, AMD seems to be making a strong effort to close the gap. This could have major implications for AI-assisted gaming, upscaling technologies like FSR (FidelityFX Super Resolution), and content creation applications. On the raw speed front, AMD has pushed the boost clock to an impressive 2.97 GHz on the RX 9070 XT, while the RX 9070 settles at 2.54 GHz. This indicates that AMD is optimizing for higher clock speeds, potentially improving gaming performance, especially in high refresh rate scenarios. However, the power consumption numbers tell an interesting story. The RX 9070 XT draws 304W, which is significantly higher than the 220W of the standard RX 9070. This suggests that AMD is pushing the XT model to its limits, possibly in an effort to keep up with (or surpass) NVIDIA's upcoming RTX 5000 series. Gamers will need a capable power supply, and cooling solutions will play a crucial role in maintaining performance. Both the RX 9070 and RX 9070 XT come with 16GB of GDDR6 video memory, which should be adequate for modern gaming at 1440p and 4K resolutions. AMD is keeping up with modern connectivity standards by supporting PCIe 5.0 x16, which ensures these GPUs are future-proofed for high-speed data transfer. While PCIe 4.0 has been more than sufficient for gaming so far, PCIe 5.0 could be a game-changer for high-bandwidth workloads like AI training and video editing on the long term. On the display front, DisplayPort 2.1a and HDMI 2.1b support means gamers can enjoy higher refresh rates, better HDR support, and 8K gaming capabilities. This is particularly relevant for those investing in the next generation of high-refresh-rate 4K and ultrawide monitors. AMD has shown ambition with the RX 9070 lineup, doubling down on AI acceleration, ray tracing, and high clock speeds. While it's too early to declare these GPUs as NVIDIA-killers, they certainly have the potential to shake up the market. Please do note that the TUF OC models run 2-3% faster than the reference baseline performance. Specification AMD Radeon RX 9070 XT AMD Radeon RX 9070 Architecture RDNA 4 RDNA 4 Manufacturing Process 4nm 4nm Transistor Count 53.9 billion 53.9 billion Die Size 357 mm² 357 mm² Compute Units 64 56 Ray Accelerators 64 56 AI Accelerators 128 112 Stream Processors 4096 3584 Game GPU Clock 2400 MHz 2070 MHz Boost GPU Clock Up to 2970 MHz Up to 2520 MHz Peak Single Precision Throughput Up to 48.7 TFLOPS Up to 36.1 TFLOPS Peak Half Precision Throughput Up to 97.3 TFLOPS Up to 72.3 TFLOPS Peak INT8 AI TOPS Throughput Up to 779 TOPS w/ Sparsity Up to 578 TOPS w/ Sparsity Peak INT4 AI TOPS Throughput Up to 1557 TOPS w/ Sparsity Up to 1156 TOPS w/ Sparsity Peak Texture Fill-Rate Up to 730.3 GT/s Up to 564.5 GT/s ROPs 128 128 Peak Pixel Fill-Rate Up to 190.1 GP/s Up to 161.3 GP/s AMD Infinity Cache™ 64 MB (3rd. Gen.) 64 MB (3rd. Gen.) Memory 16GB GDDR6 16GB GDDR6 Memory Speed 20 Gbps 20 Gbps Memory Bus Interface 256-bit 256-bit PCIe Interface PCIe 5.0 x16 PCIe 5.0 x16 Total Board Power 304 W 220 W Radeon RX 9070 XT comes with the Navi 48 GPU, sporting a rectangular design that measures about 30 by 13 mm and covers roughly 390 mm². That area estimate is based on a TSMC N4 process. It packs 64 compute units -- just a bit more than the 60 CUs found in Navi 32 -- while sticking to a 256-bit memory interface. The extra space on the die likely goes toward improved ray tracing accelerators and AI features, which power stuff like FidelityFX Super Resolution 4. With RDNA 4, AMD revamps the compute units to boost instructions per clock and push clock speeds higher compared to RDNA 3. Under normal conditions, the Radeon RX 9070 XT is expected to match the performance of a Radeon RX 7900 XT, and it might pull ahead when heavy ray tracing comes into play. Clock speeds fall between 2.5 GHz and 2.97 GHz, delivering the horsepower modern games and apps need, while graphics memory chugs along at an effective 20 Gbps.

Radeon RX 9070 (ASUS TUF OC) review

Introduction The Radeon RX 9070 is engineered for high-performance gaming and content creation, featuring an advanced RDNA 4 architecture optimized for modern graphical demands. With a boost clock of 2.54 GHz and a power consumption of 220W, this GPU offers a balanced blend of performance and efficiency, making it well-suited for 1440p and 4K gaming environments. It comes equipped with 16GB of GDDR6 video memory, ensuring that complex textures and high-resolution assets are handled smoothly during intensive applications. Supporting PCIe 5.0 x16, the Radeon RX 9070 benefits from faster data transfer rates, which is crucial for next-generation workloads and reducing bottlenecks in data-intensive processes. Additionally, the GPU features versatile connectivity options with DisplayPort 2.1a and HDMI 2.1b, facilitating support for high refresh rate displays, improved HDR, and even 4K gaming scenarios. AMD chose to skip the RX 8000 series naming, opting for a streamlined product lineup that emphasizes mainstream and high-end options, also the naming schema now matches the Ryzen series (9000). This initial GPU series includes both the RX 9070 and RX 9070 XT models. Products that sit in a 500 USD pricing space, simplifying the selection process for consumers looking for a balance between performance and affordability. The RX 9070 XT offers ray tracing capabilities that now rival NVIDIA's RTX 4080, targeting gamers requiring high-end performance with sweet eye-candy while at it. This is significant as AMD has reduced its focus on uber-high-performance GPUs. Instead, AMD reaffirmed its commitment to the high-end market by advancing its RDNA 4-based GPUs. Aside from the new GPUs, AMD's upcoming release schedule includes FidelityFX Super Resolution 4 and Ryzen 9 9000X3D processors with additional cores, all scheduled for the end of January 2025. So yeah, we have a lot to talk about and show you. First let's glimpse over the specifications. The Radeon RX 9070 series comes in two flavors: the RX 9070 and the RX 9070 XT. The most striking improvement lies in the number of AI and ray tracing (RT) accelerators, which have seen a noticeable boost compared to previous generations. The RX 9070 XT packs 64 RDNA 4 compute units, 64 RT accelerators, and 128 AI accelerators, making it a clear step up from its non-XT sibling. AMD's increased investment in AI-driven computing is evident, with AI TOPS performance reaching 1557 TOPS for the XT model. While NVIDIA has long held an advantage in AI and machine learning workloads with its Tensor Cores, AMD seems to be making a strong effort to close the gap. This could have major implications for AI-assisted gaming, upscaling technologies like FSR (FidelityFX Super Resolution), and content creation applications. On the raw speed front, AMD has pushed the boost clock to an impressive 2.97 GHz on the RX 9070 XT, while the RX 9070 settles at 2.54 GHz. This indicates that AMD is optimizing for higher clock speeds, potentially improving gaming performance, especially in high refresh rate scenarios. However, the power consumption numbers tell an interesting story. The RX 9070 XT draws 304W, which is significantly higher than the 220W of the standard RX 9070. This suggests that AMD is pushing the XT model to its limits, possibly in an effort to keep up with (or surpass) NVIDIA's upcoming RTX 5000 series. Gamers will need a capable power supply, and cooling solutions will play a crucial role in maintaining performance. Both the RX 9070 and RX 9070 XT come with 16GB of GDDR6 video memory, which should be adequate for modern gaming at 1440p and 4K resolutions. AMD is keeping up with modern connectivity standards by supporting PCIe 5.0 x16, which ensures these GPUs are future-proofed for high-speed data transfer. While PCIe 4.0 has been more than sufficient for gaming so far, PCIe 5.0 could be a game-changer for high-bandwidth workloads like AI training and video editing on the long term. On the display front, DisplayPort 2.1a and HDMI 2.1b support means gamers can enjoy higher refresh rates, better HDR support, and 8K gaming capabilities. This is particularly relevant for those investing in the next generation of high-refresh-rate 4K and ultrawide monitors. AMD has shown ambition with the RX 9070 lineup, doubling down on AI acceleration, ray tracing, and high clock speeds. While it's too early to declare these GPUs as NVIDIA-killers, they certainly have the potential to shake up the market. AMD submitted the ASUS TUF OC model, let me clearly state that though this should be considered an MSRP reference performing product, you can expect this card to run 2-3% faster than the baseline reference. Specification AMD Radeon RX 9070 XT AMD Radeon RX 9070 Architecture RDNA 4 RDNA 4 Manufacturing Process 4nm 4nm Transistor Count 53.9 billion 53.9 billion Die Size 357 mm² 357 mm² Compute Units 64 56 Ray Accelerators 64 56 AI Accelerators 128 112 Stream Processors 4096 3584 Game GPU Clock 2400 MHz 2070 MHz Boost GPU Clock Up to 2970 MHz Up to 2520 MHz Peak Single Precision Throughput Up to 48.7 TFLOPS Up to 36.1 TFLOPS Peak Half Precision Throughput Up to 97.3 TFLOPS Up to 72.3 TFLOPS Peak INT8 AI TOPS Throughput Up to 779 TOPS w/ Sparsity Up to 578 TOPS w/ Sparsity Peak INT4 AI TOPS Throughput Up to 1557 TOPS w/ Sparsity Up to 1156 TOPS w/ Sparsity Peak Texture Fill-Rate Up to 730.3 GT/s Up to 564.5 GT/s ROPs 128 128 Peak Pixel Fill-Rate Up to 190.1 GP/s Up to 161.3 GP/s AMD Infinity Cache™ 64 MB (3rd. Gen.) 64 MB (3rd. Gen.) Memory 16GB GDDR6 16GB GDDR6 Memory Speed 20 Gbps 20 Gbps Memory Bus Interface 256-bit 256-bit PCIe Interface PCIe 5.0 x16 PCIe 5.0 x16 Total Board Power 304 W 220 W Radeon RX 9070 comes with the Navi 48 GPU, sporting a rectangular design that measures about 30 by 13 mm and covers roughly 390 mm². That area estimate is based on a TSMC N4 process. It packs 56 compute units -- just a bit less than the 60 CUs found in Navi 32 -- while sticking to a 256-bit memory interface. The extra space on the die likely goes toward improved ray tracing accelerators and AI features, which power stuff like FidelityFX Super Resolution 4. With RDNA 4, AMD revamps the compute units to boost instructions per clock and push clock speeds higher compared to RDNA 3. Under normal conditions, the Radeon RX 9070 is expected to match the performance of a Radeon RX 7900, and it might pull ahead when heavy ray tracing comes into play. Clock speeds fall between 2.07 GHz and 2.54 GHz, delivering the horsepower modern games and apps need, while graphics memory chugs along at an effective 20 Gbps.

AMD's Radeon RX 9070 XT is the most exciting GPU to launch in years

It's a terrible time to be a PC gamer. Prices have skyrocketed since before the worldwide lockdowns, and one has to cast one's thoughts back almost a decade to when graphics cards launched at reasonable prices. Even flagship GPUs could be installed in many systems, though whether you should was an entirely different question. Fast-forward to 2025, we're at yet another crossroad. Nvidia turned left, launching new GeForce RTX 50 series graphics cards with an emphasis on software and AI developments. Prices are as obscene as they have been, and Nvidia has had a plethora of issues since its launch. Intel took a sharp right with two 1440p GPUs costing less than $300 in an attempt to get anyone to purchase one. AMD has -- thankfully -- decided to head straight. Launching the Radeon RX 9070 and RX 9070 XT, AMD completely changed the way it names its graphics cards. Gone are the days of 5000, 6000, and 7000, and we're now seeing a homogenized effort to make it easier for gamers to compare SKUs between AMD and Nvidia. These two GPUs are going up against Nvidia's GeForce RTX 5070 and RTX 5070 Ti, though they are considerably cheaper and potentially better. Having used both the RX 9070 and RX 9070 XT extensively for a solid week, I've come to the conclusion that AMD has successfully launched two fantastic cards for 1440p and 4K gaming. The new RDNA 4 architecture is excellent for rasterized performance, and FSR 4 seems to be quite the upgrade over previous-gen FSR implementations. AMD needed to offer something special to compete against Nvidia, which is a step in the right direction. About this review: AMD provided XDA with an RX 9070 XT samples for this review but had no input to its contents outside of pre-launch driver support. ASRock Radeon RX 9070 XT Steel Legend 8.5 / 10 The Radeon RX 9070 XT is one of AMD's first GPUs powered by RDNA 4, capable of smooth 4K gaming with powerful internals and more advanced upscaling and frame generation technologies. It's the graphics card to buy for 4K gaming. Pros & Cons Excellent 4K performance Vastly improved ray tracing and upscaling FSR 4 can more than double FPS Butter-smooth 4K gaming for less than $600 FSR 4 support is somewhat limited Still not quite as good as Nvidia for RT Nvidia still holds ground on AI $599 at Amazon AMD Radeon RX 9070 XT pricing and availability Graphics cards are the least affordable PC part gamers can buy for their system. Want the latest flagship Nvidia GPU? That'll be $2,000 at least, please. Thankfully, Intel is looking to claw some market share, which saw a super-aggressive launch for the B-series, offering 1440p gaming for less than $300. AMD typically followed Nvidia's pricing, undercutting its main rival slightly, but the company did a 180 this generation. The RX 9070 costs $549, and the RX 9070 XT sits at just $599. That's $600 for a 4K graphics card with machine learning (ML) upscaling and all the latest bells and whistles with RNDA 4. Promising frame rates in the three digits for many games at high settings is not a bad deal for PC gamers. This GPU effectively replaces the RX 7900 series yet manages to cost a full $400 less than the RX 7900 XTX alone. AMD faces some stiff competition in the $500 segment with the RTX 5070 having just released, but we'll be closely looking at these two GPUs, and AMD is most certainly coming out ahead. Throw in the latest AI-based upscaling using hardware, and Team Red is finally on a more level playing field with Nvidia. Make no mistake that unlike the RTX 5070, which is positioned as a 1440p GPU, the RX 9070 and 9070 XT are designed for 4K. RX 9070 XT Shader Units 128 Ray Accelerators/Cores 64 AI Accelerators/Cores 64 Stream Processors 4096 Base Clock Speed 1660 MHz Boost Clock Speed 2970 MHz Memory Clock Speed 2518 MHz Memory Capacity 16 GB GDDR6 Memory Bus 256-bit Memory Bandwidth 644.6 GB/s Power Draw 304 W Architecture RDNA 4 Process 5 nm Price $599 Expand RX 9070 Shader Units 128 Ray Accelerators/Cores 56 AI Accelerators/Cores 112 Stream Processors 3,584 Base Clock Speed 1330 MHz Boost Clock Speed 2520 MHz Memory Clock Speed 2518 MHz Memory Capacity 16 GB GDDR6 Memory Bus 256-bit Memory Bandwidth 644.6 GB/s Power Draw 220 W Architecture RDNA 4 Process 5 nm Price $549 Expand AMD's RX 9070 XT is the best $600 4K GPU Finally, PC gamers have some good news Close There was a lot riding on AMD's newest generation of graphics cards. Intel worked some magic with its B580 GPU, and Nvidia's DLSS continues to perform wonderfully with ray tracing. AMD's existing FSR implementation left a lot to be desired. So much so that I prefer to play games without it enabled, sacrificing visual fidelity slightly with no ray tracing and other settings configured accordingly. AMD states RDNA 4 has an uplift of two times that of RDNA 3, and I believe them. That's not the case with FSR 4 and these latest GPUs. The RX 9070 and RX 9070 XT are powered by RDNA 4, AMD's most recent GPU architecture. RDNA 4 brings a new compute unit for improved gaming performance, significant generational uplifts in ray tracing and AI acceleration (through new cores and FSR 4), boost clocks approaching 3.0 GHz, a new media engine for improved streaming and recording quality, and a hefty amount of GDDR6 VRAM. Unlike Nvidia, which focused on AI performance yet managed to fall short with its RTX 5070 besting an RTX 4090 claims, AMD went all-out with raster and compute workload improvements. This provided significant uplifts in performance and efficiency, something AMD expects to continue into the future, even with new rendering techniques making their way into gaming. We've got countless upgrades over RDNA 3, which enabled AMD to increase performance with fewer compute units. Close The most vital part of RDNA 4 is the third-gen RT Accelerators - essentially AMD's version of Nvidia Tensor Cores. Ray tracing is more commonplace in today's gaming world, and this trend will only see more games support, if not outright require, the technology to be enabled for the best experience. Having more advanced RT Accelerators allows the RX 9070 and RX 9070 XT to spread their wings and run some serious numbers. AMD states RDNA 4 has an uplift of two times that of RDNA 3, and I believe them. A big push with the RX 90 series is machine learning (ML) for upscaling. A cluster of second-gen AI Accelerators, which handle all the AI model processing. Compared to RDNA 3, we're seeing up to eight times the uplift with full support for FP8, something that was missing with previous-gen graphics cards. With the enhanced media engine and improved Adrenalin software and driver support, the RX 90 series has all the necessary tools for epic 4K gaming. Related RDNA 4 could be game-changing for AMD against Intel and Nvidia AMD's latest GPU generation has the potential to be a hit. Posts AMD was right: The best 4K gaming for $600 This is the one to buy Close The RX 7900 series was okay at 4K gaming. It wasn't super smooth, but it handled the latest games at the time well. Issues arose when one attempted to configure ray tracing, which would send FPS into the floor. FSR 3 and all other available tech at the time helped, but it was nowhere near as good as DLSS on Nvidia GPUs. Putting the new RX 90 series to the test, I was excited to see how they performed at 4K with RT enabled. To cut a long story short, the RX 9070 XT is brilliant. Native 4K performance with rasterization is excellent compared to the RX 7900 series, which cost a fair bit more than what these GPUs are retailing at -- though the RX 7900 GRE did launch at a low price in select markets. To put the RX 9070 XT to the test, I used a high-end configuration on one of the test benches, so we can be sure the GPUs are running as expected by AMD. CPU AMD Ryzen 9 9950X RAM 32 GB DDR5-7200 G.SKILL Trident Z5 RGB Motherboard Gigabyte B650 Aorus Elite AX Ice OS Windows 11 (24H2) AMD Driver 25.3.1 I tested the RX 9070 XT at 1440p and 2160p to get an idea of how it performed at the more popular resolutions. You won't be buying one of these for FHD gaming, and if you are considering a GPU for such a scenario, the Intel B-series is a solid pick. I tested both GPUs with a variety of games, some used with our existing GPU tests and others recommended by AMD, to see how FSR 4 performs. Nvidia's gen-on-gen performance uplift has been lackluster at best, AMD had plenty of room to grow, and the RX 9070 XT is an absolute powerhouse. Although Nvidia's gen-on-gen performance uplift has been lackluster at best, AMD had plenty of room to grow, and the RX 9070 XT is an absolute powerhouse compared to existing AMD GPUs. AMD's RX 9070 isn't far behind, being based on the same architecture with fewer compute units, Ray Accelerators, AI Accelerators, and Stream Processors. Game RX 9070 XT RX 9070 RTX 5070 RX 7900 GRE RX 7900 XTX Cyberpunk 2077 (Ultra) QHD: 135.2 QHD, RT: 51 UHD: 61.1 UHD, RT: 24.7 QHD: 127 QHD, RT: 43.2 UHD: 56 UHD, RT: 21.2 QHD: 104.9 QHD, RT: 46.1 UHD: 48.3 UHD, RT: 20.8 QHD: 99 QHD, RT: 26 UHD: 48.5 UHD, RT: 11.4 QHD: 123 QHD, RT: 37 UHD: 64 UHD, RT: 21 Ray tracing performance also saw a considerable boost, as did general gaming. Cyberpunk 2077 has become the new go-to game for testing RT performance, and the RX 9070 and RX 9070 XT didn't disappoint. Results should improve as games start to add FSR 4 support, but in terms of raw power, AMD is looking mighty against Nvidia. Firing up FSR 3 in supported titles with frame generation did improve things, but your mileage may vary, and you may run into minor visual artifacts, depending on the game. Game RX 9070 XT RX 9070 RX 7900 GRE DOOM Eternal (Ultra Nightmare) QHD: 313.5 UHD: 209.1 UHD, RT: 147.7 QHD: 274.1 UHD: 176.6 UHD, RT: 126.3 QHD: 301 UHD: 201 UHD, RT: 119 Final Fantasy XVI (Ultra) QHD: 119.2 UHD: 62.6 QHD: 107.3 UHD: 54.7 QHD: 96.4 UHD: 45.2 Forza Horizon 5 (Extreme) UHD: 105 UHD, FSR 3: 229 UHD: 113.3 UHD, FSR 3: 207.5 UHD: 71 UHD, FSR 3: 157 Shadow of the Tomb Raider (Highest) UHD: 122.3 UHD, FSR 3: 211.2 UHD: 112 UHD, FSR 3: 187.7 UHD: 86.4 UHD, FSR 3: 169.3 The Witcher 3 (Ultra+) QHD: 122.9 UHD: 63.2 UHD, RT: 31.8 QHD: 105.3 UHD: 54.1 UHD, RT: 26.7 QHD: 94.3 UHD: 40.7 UHD, RT: 23.1 Holding out against the RTX 5070 and performing well against older 7900 series shows how capable these new Radeon GPUs are with lower TDPs, lower prices, and more efficient platforms. AMD's right to focus on the rasterization gains with RDNA 4, which is a welcome sight against Nvidia's FG and AI push. Related Best GPUs in 2025: Our top graphics card picks Picking the right graphics card can be difficult given the sheer number of options on the market. Here are the best graphics cards to consider. Posts Testing FSR 4 and AMD's frame-gen tech Fake frames or fake flames? Activating the features in games can prove challenging since not many titles support it, and you'll be relying on the Adrenalin software to showcase support. There are a few changes to make to get the most out of FSR, too. You'll need to play around with both in-game and Adrenalin software settings to achieve the sweet spot, but once you're there, the results are positive. FSR 4 can be enabled on top of FSR 3.1, switching over to ML-based algorithms. I attempted to get FSR 4 working on SMITE and Kingdom Come Deliverance II, two completely different games. A fast-paced MOBA such as SMITE requires low latency and stable FPS to be competitive in combat, whereas KCD II is a (mostly) laid-back role-playing game with an emphasis on story, environments, and gameplay. The results in SMITE were nothing short of astounding, going from 150 FPS at native 4K to a whopping 398 with FSR 4. Like DLSS 4 frame generation, you'll need a decently high FPS to make the most of AMD's AI implementation. FSR3 wasn't too effective at improving performance without causing latency or graphical issues, but from what I've been able to see with the RX 90 series, these are largely addressed with RDNA 4. SMITE was responsive with FSR 4 and FG enabled, and I didn't notice any artifacts with the UI elements, character models, ability animations, or environment. The gameplay was smooth, fluid, and comparable to native 4K. I'll need to spend more time testing it in other titles, but so far, it seems positive. Game RX 9070 XT RX 9070 SMITE 2 (Maximum) Native UHD: 150.2 UHD FSR 3 (Quality): 274.9 UHD FSR 4 (Quality): 397.7 Native UHD: 153.4 UHD FSR 3 (Quality): 265.3 UHD FSR 4 (Quality): 325.3 Kingdom Come Deliverance II (Ultra) Native UHD: 63.2 UHD FSR 3 (Quality): 173.3 UHD FSR 4 (Quality): - Native UHD: 55.6 UHD FSR 3 (Quality): 149.4 UHD FSR 4 (Quality): - I did experience trouble getting FSR 4 to work with KCD II, even though AMD stated support would be introduced and Adrenalin detected support through the game. FSR 3 was fine, but when attempting to move up to FSR 4, both the game and driver would crash. Considering FSR 3 with FG offered a sizable boost to frame rates, I would expect another 50% or so. I'll update this review should a driver/game update address this. AMD and game developers need to work on adding FSR 3.1 support in the coming weeks to ensure both GPUs can run as expected, should you wish to utilize frame-gen tech with the latest iteration of FSR. Related 5 reasons DLSS and FSR are game-changers for budget gaming PCs Budget gaming PCs deal with aa ton of constraints. DLSS and FSR help ease the load Posts 2 You won't even know it's running Silent as a mouse We received the ASRock Steel Legend RX 9070 and 9070 XT graphics cards from AMD for testing, and both ran almost silently. The RX 9070 XT hit 305 W according to various measuring tools, with a temperature reading of just 56 degrees Celcius (81C for the memory) and fan speed of just 1,500 RPM. Even on an open-air test bench, you could barely hear the three fans spinning. They both share the same build and design, rocking a thick heatsink and three powerful blowers. All that heft is managed with a large backplate and reinforced metal frame. There's a silent mode for when the graphics card is barely working, and numerous large heatpipes direct heat away from the GPU and other vital components when pushed hard. It's impressive how effective these cooler designs have become with the latest technology from AMD, Intel, and Nvidia. Even at 4K with all the bells and whistles enabled, the card still runs cool. Related RDNA 4 preview: AI improvements and architectural optimization RDNA 4 is expected to be the next-generation of AMD graphics architecture, and we've got some details of what to expect. Posts Should you buy the AMD Radeon RX 9070 XT? You should buy the RX 9070 XT if: You want to send a message to Nvidia. You want an excellent 4K GPU. You don't wish to spend more than $600 on a single PC part. You are upgrading from an RTX 40 or RX 70 series (or older). You shouldn't buy the RX 9070 XT if: You primarily play at 1080p, as this thing is overkill. You don't want a fantastic GPU at a reasonable price. You do a lot of AI work or upscaling -- Nvidia is still king. Yes, you should buy this GPU. Review complete. Okay, I'll elaborate slightly. You should buy one of the RX 9070 XT if you're upgrading from essentially anything other than an RTX 4080 Ti or RTX 4090. The RX 9070 series is that good. The latest AI and RT cores from AMD have considerably more guts to them, making easy work of heavier workloads, such as ray tracing and advanced graphics configurations. The RX 9070 does get overshadowed by its XT sibling, however, making this the go-to for me with its $50 MSRP increase. AMD has stepped up the challenge of offering an excellent yet well-priced 4K GPU in the wake of Nvidia's floundering RTX 50 series launch. For that $50, you're getting a well-rounded GPU with more compute cores, stream processors, and AI and RT cores, resulting in higher frame rates and a better overall experience at 4K with advanced graphic options enabled. I always recommend spending as much as you can on the GPU to save upgrading earlier, and although the 9070 is a solid 4K GPU, I would buy the 9070 XT if your budget can stretch that little bit further. This thing is absolutely brilliant at rasterized loads, ray tracing, and even some upscaling with the latest version of FSR. Either way, AMD and Intel have won this early 2025 graphics card launch window. The Intel Arc B580 is a masterpiece in 1440p performance and pricing, and AMD has stepped up the challenge of offering an excellent yet well-priced 4K GPU in the wake of Nvidia's floundering RTX 50 series launch. ASRock Radeon RX 9070 XT Steel Legend 8.5 / 10 $599 at Amazon

GIGABYTE Radeon RX 9070 GAMING OC Review - Fantastic Performance, Questionable Value

Priced way too close to the Radeon RX 9070 XTFSR 4 has a lot of catching up to do to match DLSS adoptionOC model, so the price will be higher than $549 RDNA has arrived, with AMD releasing the Radeon RX 9070 and the flagship Radeon RX 9070 XT, two cards that are set to compete directly with NVIDIA's brand-new GeForce RTX 5070 and GeForce RTX 5070 Ti. Unlike NVIDIA's mid-range and enthusiast offerings, the Radeon RX 9070 XT, at $599, is only 9% more expensive than the $549 Radeon RX 9070. This muddles the Radeon RX 9070 value proposition quite a bit, as you've got a situation where spending a bit more is enough to unlock a new tier of performance. Granted, the Radeon RX 9070, especially in overclocked GIGABYTE Radeon RX 9070 GAMING OC form, is not dramatically slower than the Radeon RX 9070 XT. However, the aggressive pricing of the latter is enough to make you choose that over this. A quick look at the GIGABYTE Radeon RX 9070 GAMING OC How much slower are we talking? Looking at 1440p and 4K gaming, the difference is around 14%. With this in mind, with the same MSRP or price point as NVIDIA's GeForce RTX 5070, the overclocked GIGABYTE card we received from AMD for review is around 5% faster on average for 1440p gaming and 6% faster when you bump up the resolution to 4K when stacked up against the Founders Edition model. With RDNA 4's massive improvement to ray-tracing performance and FSR 4, which even means Path Tracing is possible on the Radeon RX 9070, AMD has effectively leveled the playing field. However, DLSS 4 and support for NVIDIA's RTX suite of technologies are enough to swing the pendulum in favor of the RTX 5070. But even here, the shadow of the Radeon RX 9070 XT looms large, a card that outperforms the GeForce RTX 5070 in every way - including raw ray-tracing performance in games like Cyberpunk 2077. So where does that leave the Radeon RX 9070, especially OC models like the GIGABYTE Radeon RX 9070 GAMING OC, which ships with a 180 MHz boost to clock speeds? OC models usually command a premium, and with only $50 separating the MSRP of the Radeon RX 9070 and Radeon RX 9070 XT, there's not a lot of wiggle room. It's no secret that GeForce RTX is the dominant brand for discrete GPUs, so for the baseline Radeon RX 9070 to make as significant an impact as the Radeon RX 9070 XT, its price should have been $449 or $479. And that's because, pricing aside, the GIGABYTE Radeon RX 9070 GAMING OC is a fantastic GPU for 1440p gaming, FSR 4 delivers excellent image quality, and AMD has finally delivered a mid-range GPU with decent ray-tracing performance. RDNA 4 - AMD Levels the Playing Field Below is a summary of AMD's new RDNA 4 architecture, applicable to all models. AMD's new RDNA 4 architecture presents a massive improvement over the chiplet design that we saw with RDNA 3. Returning to the monolithic design of RDNA and RDNA 2 might sound like a regression, especially when AMD CPUs have gone in the other direction, but this isn't the case. In a nutshell, RDNA 4 is built for 2025. This GPU architecture embraces ray-tracing performance as a key pillar, lays the groundwork for neural rendering, and supercharges AI performance for the new FSR 4 and complex AI workloads. Throw in a revamped media engine for creators and streamers and support for next-gen DisplayPort 2.1a displays, and RDNA 4 presents a new and revitalized direction for Radeon graphics. RDNA 4's overhauled Compute Unit, which houses all of the raster, ray-tracing, and AI hardware, has seen several enhancements over RDNA 3 and is one of the reasons why the Radeon RX 9070 and Radeon RX 9070 XT deliver impressive performance gains. The RDNA 4 Compute Unit, or CU, features an enhanced memory subsystem, improved scalar units (for raw raster), dynamic register allocation to reduce latency and bottlenecks, and increased efficiency. The improvements also mean that Radeon RX 9000 Series GPUs can ship with much higher clock speeds than their RDNA 3 counterparts, hitting close to 3 GHz in several XT models, with OC variants pushing 3.1+ GHz. The show's star, at least in terms of the massive improvement over what has come before, has to be the arrival of RDNA 4's 3rd Generation Ray-Tracing Accelerators. AMD is aware that game developers across PC and console are embracing ray tracing, which presents a realistic depiction of lighting and related effects like shadows and reflections. The only problem is that real-time ray tracing is complex, requiring the right blend of raw performance and innovative technologies to enhance efficiency and deliver a playable experience. One area RDNA 4's RT Accelerator delivers where RDNA 3's don't is the arrival of "Oriented Bounding Boxes," an innovative method of handling ray-tracing Bounding Volume Hierarchy (BVH) data. Think of it as efficiently tracing rays through an environment and geometry with a lower memory cost and less hardware. RDNA 4's RT Accelerator also adds a second intersection engine to double the performance of specific raytracing workloads and calculations. The results can be seen in titles with heavy ray-tracing like Alan Wake 2 and Cyberpunk 2077, where the Radeon RX 9070 XT delivers a significant 30+% performance improvement over the previous gen flagship - the Radeon RX 7900 XTX. A card with 50% more RT Accelerators than the Radeon RX 9070 XT. The Radeon RX 9070 also delivers ray-tracing performance faster than the Radeon RX 7900 XTX. RDNA 4 also fully embraces AI, with new AI accelerators that support FP8 while delivering double the FP16 and four times the INT8 performance compared to RDNA 3's AI accelerators. For gamers, this means the new AI-powered FSR 4 leverages AI to deliver a massive improvement in image quality over FSR 3. However, AMD's custom AI model, which was trained on powerful AMD Instinct hardware, is FP8-based, so it is exclusive to RDNA 4 hardware. AMD is also looking to introduce its answer to NVIDIA's DLSS Ray Reconstruction for Path Tracing with Neural Supersampling and Denoising. However, it's currently only a tech demo, which means Path Tracing, although possible on RDNA 4, does not look anywhere near as good as it does running on a GeForce RTX GPU. With improved raw performance and a massive boost to ray-tracing and AI performance, RDNA 4 presents an enormous leap forward over RDNA 3. However, catching up to GeForce RTX in these areas and offering a viable DLSS alternative does mean that early adopters will need to wait for game support and AMD to deliver its Path Tracing solution. With 30+ FSR 4-ready titles on day one and over 100 to arrive by the end of the year, there will still be a massive deficit compared to DLSS 3 and DLSS 4 as we head into 2026. Falling behind for over five years in these areas means there's a lot of catching up to do, and this will take time. This brings us to RDNA 4's enhanced Media Engine, which offers creators and streamers a big improvement to H.264, HEVC, and AV1 encoding and decoding. NVIDIA's lead in this area has meant that few creators use Radeon hardware. With RDNA 4, AMD is looking to close the gap and offer a viable alternative - especially regarding image quality using popular settings in apps like OBS. Here's a look at the specs for the new Radeon RX 9070 XT and Radeon RX 9070 compared to the previous generation's Radeon RX 7800 XT and Radeon RX 7900 XT. The Radeon RX 9070 and Radeon RX 9070 XT share the same 357mm die with a whopping 53.9 billion transistors. The baseline Radeon RX 9070 is a cutdown version, with 56 Compute Units instead of 64 - a 14% reduction. With lower boost clock speeds and a lower power rating of 220W, it is the more efficient GPU of the two as it's able to do more with less, with the Radeon RX 9070 XT using its additional Compute Units and higher clock speeds to push performance into the realm of the GeForce RTX 5070 Ti and the GeForce RTX 4080 SUPER. On paper, it's also more efficient than NVIDIA's GeForce RTX 5070 - which features a power rating of 250W. With significant architectural improvements, enhanced Ray Accelerators, and supercharged AI hardware and performance compared to RDNA 3, RDNA 4 feels like a Radeon reboot - especially when you factor in the arrival of FSR 4 and AMD's promise of neural rendering and AI denoising in the future. AMD has always delivered when it comes to raw raster performance, and RDNA 4 is no different. However, with ray-tracing performance and AI rendering now catching up to NVIDIA in a real way, AMD has delivered a level of competition that we haven't seen from the company in a long time. With 16GB of VRAM, the Radeon RX 9070 is also better equipped to handle 4K gaming and more intense workloads than the GeForce RTX 5070 - which has decided to stick with 12GB. Even though AMD's decision to stick with GDDR6 is slower than the GDDR7 found in the GeForce RTX 5070, capacity still rules the day for gaming with high-res textures and ray-tracing in 4K. The Radeon RX 9070 is also a capable AI GPU, with 1156 TOPs of INT4 and 7578 TOPs of INT8 performance. Powerful enough to run local LLMs, this AI performance also powers FSR 4, exclusive to RDNA 4. FSR 4 is a game changer for Radeon in 4K and 1440p, offering a substantial boost to performance without sacrificing much or anything noticeable regarding image fidelity. Granted, FSR 4 adoption is nowhere near the levels of DLSS 3, and NVIDIA's DLSS 4 is a step above FSR 4, but going from the least impressive upscaling solution to one that can compete is fantastic to see. The GIGABYTE Radeon RX 9070 GAMING OC is a relatively compact triple-fan GPU. Its cooling is powered by GIGABYTE's new Hawk Fans, which offer better performance without increasing noise. According to the company, the design was inspired by the wings of eagles, but all you need to know is that they're super practical and the same fans found in more premium AORUS cards from the company. That and under load, the GPU temperature of the GIGABYTE Radeon RX 9070 GAMING OC sits below 60 degrees Celsius, even with its generous +180 MHz boost clock speed. GIGABYTE's latest GPUs feature server-grade thermal conductive gel instead of thermal pads, which you can find on the VRAM and MOSFETs. With improved contact, this is another reason why the GIGABYTE Radeon RX 9070 GAMING OC runs nice and cool. Rounding out the internal design is a large copper plate that makes direct contact with the GPU alongside composite copper heat pipes for effective heat transfer to the heatsink. The actual look and physical design of the GPU is somewhat unique, too, with a curved surface that features what looks like a paint splatter pattern alongside a little piece of plastic that slides over the RGB lighting strip on the side - giving you the options of lighting up the GIGABYRE logo or simply basking in the glow of some nice RGB. The reinforced metal backplate features the GIGABYTE GAMING logo, AMD Radeon logo, and a vented exhaust for additional cooling power. There are several nice little touches, from the DUAL BIOS switch to the power dual 8-pin power connectors, including indicators to let you know that it's all connected up correctly. The Games and Tests PC gaming not only covers a wide range of genres and styles, from indie games with simple 2D graphics to massive 3D worlds lit by cutting-edge real-time ray tracing technology. With that, the needs and requirements of each gamer vary. High refresh rates and latency reduction become more important than flashy visuals or playing at the highest resolution possible for those who live and breathe fast-paced competitive games. For those who want to live in a cinematic world and become a key player in an expansive narrative, ray-tracing, and high-fidelity visuals are a stepping stone toward immersion. Our chosen benchmarks cover various games, engines, APIs, and technologies. For the Radeon RX 9070, all tests are run at 1080p, 1440p, and 4K and include results for performance-boosting Super Resolution technologies like AMD's new AI-powered FSR 4, alongside older FSR 3 and FSR 2 versions. However, our benchmark results are still sorted using 'raw performance' or native rendering. Here's the breakdown of games, settings, and what's being tested. As we start to see more mid-range and mainstream GPU launches in 2025, we're starting to add 1080p benchmark data to our reviews, and here we see the GIGABYTE Radeon RX 9070 GAMING OC deliver 1080p performance that is slightly faster than the GeForce RTX 5070 Founder Edition. The performance improvement varies from title to title, so check out the individual game results below for a closer look. When it comes to evaluating the GIGABYTE Radeon RX 9070 GAMING OC, the focus of this review is placed on 1440p and 4K gaming performance, so let's dig in. Average Gaming Performance - 1440p Results Looking at the average 1440p gaming performance, the GIGABYTE Radeon RX 9070 GAMING OC is roughly on par with the Radeon RX 7900 XT. However, performance sits behind the Radeon RX 7900 XT when looking at pure raster performance, with the most significant gains coming from RDNA 4's superior ray-tracing performance. Here the GIGABYTE Radeon RX 9070 GAMING OC is 5.3% faster than the GeForce RTX 5070 Founders Edition and around 14% slower than the beefier Radeon RX 9070 XT. Traditionally, ray-tracing performance has been where Radeon has fumbled or fallen behind. That's not the case here, as if you look at Cyberpunk 2077 using the RT Ultra preset in 1440p. The GIGABYTE Radeon RX 9070 GAMING OC delivers the same performance as the GeForce RTX 5070 - which is also faster than the previous-gen flagship, the Radeon RX 7900 XTX. Numbers, of course, don't paint a complete picture, so it's worth reiterating that with DLSS 4 Super Resolution and Ray Reconstruction, Cyberpunk 2077 with ray-tracing looks much better running on GeForce RTX hardware. One game where the GIGABYTE Radeon RX 9070 GAMING OC significantly outperforms the GeForce RTX 5070 is Call of Duty: Black Ops 6, where the 120 FPS average (which increases to 151 FPS with FSR 4) is a massive 39% faster. If Call of Duty is your game, the new Radeon RX 9000 Series is the way to go, as Call of Duty: Black Ops 6 also features native support for FSR 4. The GIGABYTE Radeon RX 9070 GAMING OC only falls behind the GeForce RTX 5070 in several titles, however, it has the lead more often than not. Outside of Call of Duty, performance is close, even regarding ray-tracing, so the real difference becomes features like FSR 4, DLSS, and Frame Generation. NVIDIA has the edge here, which is no surprise, which adds weight to the idea that the Radeon RX 9070 is priced higher than it should be. Average Gaming Performance - 4K Results The GIGABYTE Radeon RX 9070 GAMING OC doesn't deliver 4K gaming performance that is on par with the Radeon RX 9070 XT or GeForce RTX 5080. However, it's still surprisingly capable at this resolution - with an average frame rate of 67 FPS across all titles in our benchmark suite. This means that to get more performance, you'll need to enable FSR 4 in games that support it or tweak settings if you're after the 100 FPS experience. Here the GIGABYTE Radeon RX 9070 GAMING OC delivers performance that is roughly on par with the GeForce RTX 4070 Ti SUPER, or around 14% slower than the Radeon RX 9070 XT. There are some notable wins for the GIGABYTE Radeon RX 9070 GAMING OC compared to the GeForce RTX 5070 in 4K: Dragon Age: The Veilguard with ray-tracing runs 20% faster, F1 24 with ray-tracing runs 14% faster, and the visually impressive Horizon Forbidden West runs 19% faster. Horizon Forbidden West features FSR 4 support, so the 74 FPS quickly becomes 89 FPS with fantastic image quality. Like with 1440p gaming, NVIDIA's hardware's advantage comes via the widespread adoption of DLSS 3, DLSS 4, and AI-rendering that extends to Frame Generation and the new Multi-Frame Generation. The GIGABYTE Radeon RX 9070 GAMING OC's overall 4K gaming performance is impressive, but if you look at the chart, several GeForce RTX options have the same overall performance. As mentioned a few times in this review, the biggest issue with the Radeon RX 9070 is that the price is close enough to the Radeon RX 9070 XT, so it makes more sense to go for that. Especially if you want to enable ray-tracing, as the most significant difference between the Radeon RX 9070 and Radeon RX 9070 XT is that ray-tracing performance is up to 20% faster. Benchmarks - 3DMark Synthetic Tests 3DMark offers a suite of synthetic benchmarks built to test GPUs in various scenarios. 3DMark Steel Nomad is a cutting-edge DirectX 12 benchmark with newer, modern rendering techniques designed to push GPUs to their limit. The 'Light' version tests at 1440p, while the main Steel Nomad benchmark tests pure native 4K rendering. Port Royal is a benchmark focusing exclusively on real-time ray tracing for lighting effects like reflections, shadows, and more. The synthetic Steel Nomad Light test, which targets 1440p gaming, sees the GIGABYTE Radeon RX 9070 GAMING OC deliver a score that is on par with the GeForce RTX 5070 Founders Edition. Bump things up to the more intensive 4K Steel Nomad test, and the GIGABYTE Radeon RX 9070 GAMING OC delivers a score that is 20% higher than the GeForce RTX 5070. This isn't something you'll find in many titles, making us wonder if VRAM capacity has a more significant role in this test. The Steel Nomad Light test better reflects what we saw in real-world gaming, with non-ray-tracing performance sitting slightly below the Radeon RX 7900 XT. The 3DMark Port Royal results are interesting because they show that the GIGABYTE Radeon RX 9070 GAMING OC is 14% faster than the GeForce RTX 5070 regarding ray-tracing. The score here is even higher than what we got with the GeForce RTX 4070 Ti SUPER. In titles without Path Tracing or super-intensive RT workloads, the GIGABYTE Radeon RX 9070 GAMING OC is faster than the GeForce RTX 5070, however, the lead isn't as pronounced as we see here. Still, the results here show RDNA 4's massive RT gains over RDNA 3. With the arrival of the Radeon RX 900 Series and the new Radeon RX 9070, AMD's new FSR 4 represents a massive improvement over FSR 3.1, FSR 3, and FSR 2. The shift to a custom and powerful AI model for upscaling shows that ML or AI is the definite way to maintain image quality that is on par or even superior to native rendering. As FSR 4 was explicitly designed for RDNA 4 and trained on powerful AMD hardware, it is exclusive to the Radeon RX 9000 Series due to the advanced AI hardware requirements. The good news is the games with FSR 3.1 are automatically upgraded to FSR 4 via AMD's Adrenalin Software, with a nice overlay showing a green FSR 4 tick when booting up a compatible game. FSR 4 was available in a few games in our benchmark suite. However, when looking closely at image quality compared to DLSS and previous versions of AMD's FSR, we focused on two titles-Horizon Forbidden West and Warhammer 40K: Space Marine 2. Without beating around the bush, the difference that FSR 4's new AI model brings to image quality is night and day compared to previous versions. The image quality, especially in 4K, is now at a level that the 'Quality' preset is something you would enable whenever it's available for a free boost to performance. When paired with AMD Frame Generation and the company's improved latency reduction technology, you can push performance, even in 4K, to triple-digit territory on the GIGABYTE Radeon RX 9070 GAMING OC. Granted, the Frame Generation quality is not at the level of NVIDIA's, but the results show a concerted effort from AMD to level the playing field. You're always going to sacrifice image quality and latency with Frame Generation. Still, the 136 FPS in Horizon Forbidden West with FSR 4 with Frame Gen was something we preferred over native rendering. And really, that's a big win for Radeon. Path Tracing Performance - 1440p Path Tracing takes real-time ray-tracing and applies the concept of ray-traced effects to anything and everything - global illumination, shadows, reflections, indirect lighting, and more. With multiple bounces, it's a hardware-intensive and cutting-edge look at the future of PC gaming that is only possible thanks to AI tools and technologies. With RDNA 4 dramatically improving ray-tracing performance compared to RDNA 3, GPUs like the Radeon RX 9070 and Radeon RX 9070 XT are capable of rendering stunning Path Traced visuals. However, it's more proof of concept than something practical. Path Tracing is possible on the GIGABYTE Radeon RX 9070 GAMING OC, however, it's more proof of concept as image quality is nowhere near as crisp and detailed as the DLSS 4 image quality you get. Also, this is one area where the Radeon RX 9070 falls behind the GeForce RTX 5070, so if Path Tracing is something you're interested in, GeForce RTX GPUs are the way to go. At least for now, as we know, AMD is working on its version of Ray Reconstruction and other neural rendering techniques that have been explicitly designed to improve path tracing performance and image quality. Indiana Jones and the Great Circle is a title that recently added support for AMD's FSR and Frame Generation technologies; however, as it's FSR 3 and not FSR 3.1, there's no option to enable FSR 4. Still, with FSR and AMD's Frame Generation, you can hit 100 FPS on the GIGABYTE Radeon RX 9070 GAMING OC - which is fantastic. A mid-range Radeon GPU that delivers decent Path Tracing performance is a game changer for the brand. Again, without the same level of AI-enhanced rendering as GeForce RTX, it's more proof of concept, but RDNA 4 lays the foundation upon which a future generation can build. Temperature and Power Efficiency The GIGABYTE Radeon RX 9070 GAMING OC has a boost clock speed of 2700 MHz, notably higher than the reference spec of 2520 MHz. With temperatures topping out at only 57 degrees in our stress test, this leaves room for further overclocking. However, the 220W power rating of the Radeon RX 9070 might not deliver as much OC potential as the beefier Radeon RX 9070 XT's 300-340W limit. That said, the Radeon RX 9070 is the more efficient GPU of the two, with OC Radeon RX 9070 XT models drawing up to 40-50% more power than the Radeon RX 9070. Final Thoughts The Radeon RX 9070 main competition might be NVIDIA's similarly priced GeForce RTX 5070, but its real competitor is its older sibling - the Radeon RX 9070 XT. For $50 more, you get a notable bump in performance, with improved ray-tracing and even more impressive 4K gaming. When stacked against the GeForce RTX 5070 reference spec (the GIGABYTE Radeon RX 9070 GAMING OC is an overclocked model), you're looking at faster performance in most titles - including games with ray-tracing. Yes, RDNA 4 is a big win for AMD, thanks to the improvements you'll get with RT and the arrival of the AI-powered FSR 4 upscaler. However, with DLSS 4 and hundreds of games released over the past couple of years, including DLSS support, FSR 4 still has a long way to go to catch up. And with DLSS Ray Reconstruction, games with ray-tracing look better running on GeForce RTX hardware, so the software advantage still favors Team Green. With this in mind, we're not sure why AMD decided to price the Radeon RX 9070 the same as the GeForce RTX 5070, other than to place the focus squarely on the Radeon RX 9070 XT. Put it this way: If AMD only released the Radeon RX 9070, RDNA would feel like a giant leap forward for Radeon capabilities followed by a big step back. Without market or mind share, AMD's biggest drawcard right now is pricing, especially when you look at the current state of the PC gaming market. The Radeon RX 9070 XT offers GeForce RTX 5070 Ti performance for $150 less - that's a win. On average, the Radeon RX 9070 is a little bit faster than the GeForce RTX 5070 for the same price. That's not that exciting. With a price starting from $449, the Radeon RX 9070 would be an easy recommendation, offering a new level of performance for a sub-$500 GPU - a game change along the lines of the Radeon RX 9070 XT. The GIGABYTE Radeon RX 9070 GAMING OC is still a great GPU for PC gaming thanks to AMD's new RDNA 4 architecture - it's simply priced too high.

AMD Radeon RX 9070 Review - IGN

The AMD Radeon RX 9070 arrives in a weird time for graphics cards. We're fresh off the launch of Nvidia's latest generation of cards, which places AMD's new $549 card in direct competition with the disappointing GeForce RTX 5070. That's a competition that AMD has no problem winning right now, which would usually make the Radeon RX 9070 the graphics card to buy for 1440p gaming. Of course it's a little more complicated than that, and AMD has nothing to blame but itself. The Radeon RX 9070 is only $50 cheaper than the excellent Radeon RX 9070 XT. That price difference makes sense in sheer mathematical terms - the 9070 is about 8% slower and 9% cheaper than the 9070 XT - but it's hard to justify not just spending the extra 50 bucks to get even better performance. Either way, when it comes down to choosing between two AMD graphics cards, things are looking pretty good for Team Red. Just like the AMD Radeon RX 9070 XT, the RX 9070 is based on the new RDNA 4 graphics architecture. This new graphics tech brings huge improvements to performance, to the point where the 9070 drastically outperforms the Radeon RX 7900 GRE from last generation, despite having 30% fewer compute units. The Radeon RX 9070 has 56 Compute Units, each with 64 Streaming Multiprocessors, or SMs, making for a total amount of 3,584 shaders. Each compute unit also has one Ray Accelerator and two AI Accelerators, making for 56 and 112 total, respectively. While the SMs do a bulk of the heavy lifting, the improvements to the Ray and AI Accelerators finally make this AMD graphics card able to hold its own in games with ray tracing. Plus, the AI Accelerators have improved enough that AMD was able to introduce FidelityFX Super Resolution (FSR) 4, bringing AI upscaling to AMD graphics cards for the first time. Just like the 9070 XT, the RX 9070 is paired with 16GB of GDDR6 VRAM on a 256-bit bus. That's pretty much the exact same memory configuration as the 7900 GRE, and should be enough for 1440p gaming for years. However, it would have been nice to see AMD adopt GDDR7, like Nvidia did, but that probably would have brought up the price. AMD recommends at least a 550W power supply to run the RX 9070, as it has a power budget of 220W. In my testing, though, power consumption peaked at 249W, which is a bit over its budget. That's not enough to make it require better cooling, per se, but I'd recommend going with at least a 600W PSU, just to be on the safe side. As for cooling, though, it's important to keep in mind that AMD is not releasing a reference design of the RX 9070, for the first time in a few generations. That means every version of the Radeon RX 9070 will be manufactured by a third-party board manufacturer. AMD sent me the Gigabyte Radeon RX 9070 Gaming OC 16G. It's a beefy triple-slot card with a slight factory overclock, so keep that in mind. Since DLSS blew up in 2018, AI upscaling has been one of the best ways to get better performance without losing too much in the way of image quality. This used to be a solution that was essentially exclusive to Nvidia graphics cards, but the times have changed and FSR 4 brings AI upscaling to AMD GPUs for the first time. Just like any other similar solution, FSR 4 takes previous frames, along with in-game data, and runs them through an AI model to accurately upscale a lower resolution image back up to your native resolution. This is slightly different from the Temporal upscaling found in previous versions of FSR 3, which didn't have an AI algorithm to tighten up the details, resulting in artifacts like ghosting. The downside is that because of the performance costs of running an AI model, FSR 4 has a slight performance loss compared to FSR 3. For instance, in Call of Duty: Black Ops 6 at 1440p on the Extreme preset, FSR 3 gets me a frame rate of 165 fps, but switching it over to FSR 4 drops that to 159. Likewise, in Monster Hunter Wilds, the Radeon RX 9070 got 81 fps at 4K max settings with ray tracing enabled, but went down to 76 fps when I switched to FSR 4. But because there's a toggle in the Adrenalin software that essentially lets you 'opt-in' to FSR 4, you can choose whether you want the better image quality of FSR 4 or the slightly better performance of FSR 3. I mostly play single player games, so I'd go for FSR 4 every day of the week, but you may prefer sticking to FSR 3 for fast-paced online games like Marvel Rivals. At $549, the AMD Radeon RX 9070 goes head-to-head with the Nvidia GeForce RTX 5070, and comes out ahead more often than not. At 1440p, this mid-range GPU is on average 12% faster than the RTX 5070, with a 22% lead over its AMD predecessor, the RX 7900 GRE, which launched for the same $549 in 2024. That's a significant improvement, especially with the 9070 having 30% fewer cores. However, keep in mind that AMD sent me a factory overclocked version of the RX 9070. The exact specs of the Gigabyte Radeon RX 9070 Gaming OC haven't been released at the time of writing, but GPU-Z reports a boost clock of 2,700Mhz, which is about a 7% clock speed boost. That won't quite be a 7% performance boost, but it should increase frame rates by about 4-5%. I tested every graphics card on their current public drivers at the time of writing. That means all Nvidia cards were tested on Game Ready driver 572.60, and all AMD graphics cards on Adrenalin 24.12.1. The AMD Radeon RX 9070 and 9070 XT were tested on review drivers provided by AMD, as was the Nvidia GeForce RTX 5070 with Nvidia's review drivers. The 9070 gets off to a strong start in 3DMark, which gives more of an impression of potential performance, rather than real-world gaming frame rates. In the Speed Way test, which has ray tracing enabled, the 9070 gets 5,828 points, compared to 5,845 points from the RTX 5070, which is essentially a tie. But in Steel Nomad which doesn't have ray tracing enabled, the 9070 beats Nvidia's card 6,050 to 5,034 - a 20% difference at the same price. Call of Duty: Black Ops 6 is literally the game AMD used to preview the 9070 at CES 2025 - I even snuck in an early benchmark in at AMD's booth - so it's no surprise that it soars here. At 1440p with FSR 3 set to Balanced, the 9070 gets 165 fps, compared to 131 from the 5070 and 143 from the 7900 GRE. That's a 26% and 15% lead, respectively. The result in Cyberpunk 2077 is wild, as it's a game that's always favored Nvidia hardware, especially with ray tracing enabled. However, at 1440p with the Ray Tracing Ultra preset, the AMD Radeon RX 9070 outperforms the RTX 5070 by 3%. That's not much of a lead, but any AMD lead in this game is huge. Especially at the same price. When I test Metro Exodus, I run it without any kind of upscaling, because it only supports DLSS, and not FSR or XeSS. So instead it's just raw ray tracing performance. But even here, the Radeon RX 9070 wins the day, getting an average of 71 fps, compared to 64 fps from the RTX 5070. That's an 11% lead in another game that throws Nvidia's logo at you as you load into it. Red Dead Redemption 2 gives the RX 9070 another huge lead, this time using Vulkan. At 1440p with everything maxed out, the 9070 gets 142 fps, compared to just 115 from the RTX 5070, making for an incredible 23% performance lead in favor of Team Red. Though, keep in mind that the 7900 GRE also gets 113 fps with the same settings, so this could be another case of the engine favoring AMD hardware. In Total War: Warhammer 3, the RX 9070 only has a huge lead at 4K, likely due to the higher frame buffer. Moving down to 1440p, the RTX 5070 catches up, getting 134 fps to the 9070's 135. That's within the margin of error. In Assassin's Creed Mirage, the RX 9070 gets 193 fps at 1440p with the Ultra preset and FSR set to Balanced. Nvidia has a hard time keeping up, only getting 163 fps in the same test. That's an 18% lead for the AMD card, though this is another game that has traditionally favored AMD GPUs. Black Myth Wukong, however, is another game that loves Nvidia hardware, and it ends up being a wash between the Radeon 9070 and the RTX 5070. At 1440p, the RX 9070 gets 67 fps to Nvidia's 66, and that's with the hard-to-run Cinematic preset. Forza Horizon 5 is getting up there in years, but it's one of those games where high framerates are rewarded. At 1440p, the Radeon RX 9070 averages 185 fps, compared to 168 from the 5070 and 152 from the RX 7900 GRE, making for a 12% and 25% difference, respectively. Coming out hot on the heels of the GeForce RTX 5070 only works in AMD's favor. The Radeon RX 9070 and the GeForce RTX 5070 both cost $549, so the fact that AMD is able to pull ahead in so many of the tests is incredibly impressive. What's more, the Radeon RX 9070 will better stand the test of time, thanks to having 16GB of VRAM, even if it is a little slower than the RTX 5070's GDDR7. Even if these two graphics cards were perfectly tied - which they're not - AMD having 33% more VRAM would have made the RX 9070 the better value. Combine that with better performance, and it's a no-brainer.

AMD Radeon RX 9070 XT Review - IGN

For the last couple of generations, AMD has been desperately trying to compete with Nvidia at the high end. However, now, with the AMD Radeon RX 9070 XT, Team Red has ceded the ultra-high-end to the RTX 5090 and is instead focused on making the best graphics card for the majority of gamers - a goal it absolutely achieves. The AMD Radeon RX 9070 XT is a $599 graphics card that trades blows with the $749 GeForce RTX 5070 Ti. That alone sets it up as one of the best GPUs on the market today - but AMD further sweetens the deal with FSR 4, finally bringing AI upscaling to an AMD graphics card for the first time. Simply put, this is the graphics card to get for 4K gaming, especially if you don't have $1,999 to mindlessly blow on the RTX 5090. The AMD Radeon RX 9070 XT is built on the RDNA 4 architecture, and while it brings improvements to its shader cores, the real stars of the show are the new RT and AI Accelerators. The AI Accelerators, in particular, are the driving force behind FidelityFX Super Resolution 4, or FSR 4, which brings AI upscaling to AMD graphics cards for the first time. This new AI upscaling doesn't necessarily improve framerates over games using the older FSR 3.1 model, but it greatly improves accuracy - meaning better image quality. Luckily, if you care more about the framerates, there's a nice little toggle in the Adrenalin software that lets you turn off FSR 4 entirely. But it's more than just AI upscaling. AMD has also made huge improvements to its shader cores, resulting in better performance on a per-core level. So, despite the Radeon RX 9070 XT having 64 Compute Units, compared to 84 in the previous-generation Radeon RX 7900 XT, AMD was able to deliver a sizable generational leap - and with a much lower launch price this time around. Each of these Compute Units has 64 Streaming Multiprocessors, or SMs, making for a total of 4,096, along with 64 ray accelerators and 128 AI accelerators. However, the Radeon RX 9070 XT also has less memory than the RX 7900 XT, coming in at 16GB of GDDR6 on a 256-bit bus, rather than 20GB of GDDR6 on a wider 320-bit bus. That's a hit to both capacity and bandwidth, but it's still enough to get by in most games at 4K. Still, given that AMD is still on GDDR6, it would have been nice to not get this downgrade at all. While the new architecture is more efficient, the RX 9070 XT has a higher power budget than the 7900 XT that came before, though not by much. The 9070 XT needs 304W of power, compared to 300W with the older card. It's an odd increase, especially because in my testing, I found that the 7900 XT still consumed more power, peaking at 314W, compared to 306W from the 9070 XT. This is a pretty standard power budget in a modern graphics card, so cooling it won't be too challenging. Unlike any other generation, though, AMD is not launching a reference design for the Radeon RX 9070 XT, which means you'll have to rely on third-party manufacturers for this GPU. I received the Powercolor Radeon RX 9070 XT Reaper for review, with a surprisingly narrow triple-fan design. Despite its small footprint, though, it still kept the temperature at 72°C throughout my test suite. AMD is still sticking with standard power connectors too, with the Radeon RX 9070 XT needing two 8-pin PCI-E power connectors, with no meltable adapters in sight. This makes it an easy upgrade for pretty much anyone, provided you have the 700W power supply AMD recommends. As for ports, you get three DisplayPort 2.1a and one HDMI 2.1b, which is exactly what you would expect from a modern graphics card. I do wish AMD would add a USB-C port, though, just for more flexibility. For years now, AMD has needed an AI upscaling solution that could go toe-to-toe with DLSS. Because while the performance was certainly there with the existing versions of FidelityFX Super Resolution, it has been plagued with ghosting and fuzziness from the word go. AMD addresses that with the Radeon RX 9070 XT, with the AI-powered FSR 4. Just like DLSS, FSR 4 utilizes the AI accelerators in the Compute Units to analyze previous frames, along with other data from the game engine, in order to accurately upscale a lower-resolution image up to your native resolution. The result does look better than FSR 3, which used a Temporal upscaling solution instead of AI, though it does come with a bit of a performance hit. In Call of Duty: Black Ops 6 at 4K Extreme settings with FSR 3.1 set to "Performance", the AMD Radeon RX 9070 XT was able to get 134 fps. However, swap that over to FSR 4, and that number goes down to about 121 fps, a 10% performance loss, albeit with better image quality, especially in things like grass or in-game text. Likewise, in Monster Hunter Wilds, the 9070 XT was able to get 94 fps at 4K max settings with both FSR 3 and ray tracing enabled, but dropped to 78 fps when using FSR 4 - a 20% drop in performance. That's a bummer, I know, but really it makes sense. By its very nature, an AI upscaling solution is a more expensive workload than the temporal solution AMD was using in the past. I even checked in with AMD and I was told the performance drop was totally expected behavior, and that AMD is hoping the improved image quality will make up for the performance drop. That's of course going to vary person to person, but it's worth it if you primarily play single-player games where image quality is more important than raw frame rate. Luckily, it's not like FSR 3.1 is going away any time soon. FSR 4 is an opt-in feature: you can easily go into the Adrenalin software and toggle FSR 4 off. In fact, it was turned off by default on my review sample, though that could have been because of early drivers. AMD really came out swinging with the Radeon RX 9070 XT. At $599, it undercuts the Nvidia Geforce RTX 5070 Ti by a whopping 21%, but ends up being, on average, 2% faster. There are games where the RTX 5070 Ti beats the Radeon RX 9070 XT, but the fact that these two graphics cards are trading blows at all is a win for AMD. Across my whole test suite, I found that the RX 9070 XT was about 17% faster than the RX 7900 XT - which launched at $899 2 years ago - and 2% faster than the new $749 RTX 5070 Ti. But where this card really shines, especially considering its cost, is at 4K. AMD maintains the same lead at that resolution, making the 9070 XT the ultimate entry-level 4K graphics card, even with ray tracing enabled. I tested all graphics cards on the most recent drivers available. That means all Nvidia cards were tested with Game Ready Driver 572.60, except for the RTX 5070, which is still on review drivers at the time of writing. All AMD cards were tested on Adrenalin 24.12.1, except for the Radeon RX 9070 XT and RX 9070, which were tested on pre-release drivers provided by AMD. While 3DMark isn't a game you can actually play, it still gives a good look at how the potential of each graphics card compares to one another. It's rare that it reflects real world performance, but it generally gives you a good idea what to expect. And, well, the 9070 XT beats the 7900 XT by 18% in Speed Way, even though it's 18% behind the RTX 5070 Ti. However, swap to the Steel Nomad benchmark, and the story shifts altogether. The performance jump over the Radeon RX 7900 XT jumps to 26%, and the 9070 XT even beats the more expensive RTX 5070 Ti by 7%. In Call of Duty: Black Ops 6, the Radeon RX 9070 XT pulls ahead of the GeForce RTX 5070 Ti by 15%. However, it's clear that this game favors AMD hardware, with the Radeon RX 7900 XT only losing to the new graphics card by 6%. Cyberpunk 2077 has always performed best on Nvidia cards, so it's not a surprise that the 5070 Ti beats the Radeon RX 9070 XT, but by less than in previous generations. The Radeon RX 9070 XT gets 71 fps at 4K with the Ray Tracing Ultra preset with FSR 3 set to performance mode. The RTX 5070 Ti with the same settings, but DLSS set to performance mode, gets 75 fps. That's only a 5% lead, despite the huge price gap. Metro Exodus is another game with heavy ray tracing, though this one is tested with no upscaling at all. At 4K, the Radeon RX 9070 XT is able to manage 47 fps, which is within the margin of error of the RTX 5070 Ti's 48 fps. The last-generation Radeon RX 7900 XT really struggles here, only getting 38 fps, making for a 24% jump. While it's old, Red Dead Redemption 2 is a great showcase of Vulkan performance, and it scales extremely well on the 9070 XT. In everyone's favorite cowboy-em-up, the Radeon RX 9070 XT gets 125 fps with everything cranked, with the RTX 5070 Ti trailing behind at 110 fps. The 7900 XT trails a bit further behind at 106 fps. Unfortunately, the winning streak can't go on forever, with the Radeon RX 9070 XT falling 13% behind the RTX 5070 Ti in Total War: Warhammer 3. What's worse is with 76 fps the 9070 XT also doesn't pull that far ahead of the 7900 XT's 71 fps. In Assassins Creed Mirage, the RX 9070 XT gets back on the right foot, with its 163 fps outperforming the RTX 5070 Ti's 146 fps by 12% and the 7900 XT's 150 fps by 9%. This is another game that has traditionally favored AMD GPUs, but it's still a huge jump over the competition. The most surprising RX 9070 XT win is in Black Myth Wukong, where it gets 70 fps at 4K with the Cinematic Preset and FSR set to 40%. With those same settings and DLSS, the RTX 5070 Ti gets 65 fps, making for an 8% lead in Team Red's favor. Black Myth Wukong has a lot of intense ray tracing effects, something AMD hasn't beat Nvidia at before. It's incredible how much the Ray Accelerators have improved over RDNA 3, too - the Radeon RX 7900 XT only gets 60 fps, despite having more Compute Units. Forza Horizon 5 is also getting up there in years, but the Radeon RX 9070 XT still beats the competition, getting 158 fps over the 5070 Ti's 151 fps. That's only a 5% improvement, but it demonstrates just how aggressive the 9070 XT is. Due to how it was announced under the radar at CES 2025, it felt like AMD was holding the Radeon RX 9070 XT as some kind of secret weapon against Nvidia's Blackwell graphics cards, and I'm glad it did. At $599, the Radeon RX 9070 XT seems like a sort of return to sanity for the graphics card market. Sure, it's not quite as fast as the RTX 5080 or the RTX 5090, but those cards are overkill for most people - and cost at least $400+ more. I know I'm not the only one that feels like the last great flagship graphics card was the GTX 1080 Ti when it launched for $699 in 2017. That was the fastest consumer card of its time, and while the 9070 XT can't quite make that claim, it still feels like the first worthy flagship we've seen since then.

AMD Radeon RX 9070 and 9070 XT review: Hitting NVIDIA where it hurts

AMD has finally done it. For years it's tried to undercut NVIDIA with slightly cheaper, but less capable, video cards like the Radeon 6700 XT and 7900 XT. And sure, it's still following that same strategy with the new Radeon 9070 and 9070 XT. This time around, though, AMD has produced far more capable hardware, especially when it comes to 4K and ray tracing performance. And there's hope that it could finally catch up to NVIDIA's DLSS AI upscaling with its new FidelityFX Super Resolution 4 (FSR4) technology. These aren't perfect video cards, to be clear. But for $549 and $599, the Radeon 9070 and 9070 XT are far more compelling than AMD's previous lineup. That's particularly true since the 9070 XT is $150 less than NVIDIA's RTX 5070 Ti, and in our testing it's sometimes faster than that card. AMD still has to prove that it can catch up to NVIDIA's DLSS, which has had several years to steadily improve its AI upscaling capabilities. In particular, AMD needs to match the performance of NVIDIA's multi-frame generation in DLSS 4, which has led to some surprisingly high fps figures while testing the RTX 50-series GPUs. AMD's Fluid Motion Frames technology is a start, but it's not nearly enough at this point. The Radeon RX 9070 and 9070 XT are AMD's first RDNA 4 GPUs, a new platform sporting more powerful and efficient compute units, third-generation ray tracing accelerators and second-generation AI accelerators. There isn't a huge technical difference between the two cards: The RX 9070 features 56 compute units, 56 ray accelerators and 112 AI accelerators, whereas the 9070 XT has eight more compute accelerators and 16 additional AI processors. The XT model is clocked slightly higher, as you'd expect, and it also draws more power (304W vs 220W). Notably, both cards also ship with 16GB of GDDR6 VRAM, whereas NVIDIA skimped a bit and only put 12GB of RAM in the $549 RTX 5070. As I was comparing our XFX-built review units, though, it was practically impossible to tell the RX 9070 and 9070 XT apart without reading the fine print on their labels. This is fairly typical for AMD cards with XT variants, and it makes sense since they're technically so similar. The XFX Swift GPUs I tested came in an elegant frost white case, with three large fans and surprisingly massive heatsinks. They're both 3.5-slot cards, so they also take up more room than the RTX 5090, which is a dense two-slot beast. Thanks to AMD's efficient RDNA design, you also don't need to rely on complex power dongles to drive these cards. They both take two PSU connections directly, and XFX recommends 800-watt power supplies at a minimum. That's significantly higher than AMD's minimum 650W spec though, so it could be that XFX is just trying to play it safe with its massive cooling setup. (If you're planning to overclock these cards, though, you'll likely want more than a mere 650W PSU.) I knew AMD was onto something special when I learned the Radeon RX 9070 averaged 60 fps in Cyberpunk 2077 in 4K Ray Tracing Overdrive mode. Sure, it needed FSR 3 upscaling to get there, but it was still better than the 50 fps I saw on the Radeon 7900XT a few years ago. The game occasionally dipped below 60 fps, but it was still very playable. 1440p was much smoother overall, averaging 117 fps. The Radeon 9070 XT, meanwhile, averaged 68 fps in 4K and 130 fps in 1440p. Sure, these numbers are far behind the raw figures from NVIDIA's RTX 5070, which used the magic of multi-frame generation to deliver 115 fps in 4K with ray tracing and maxed out graphics. That card also hit 205 fps in 1440p. But as impressed as I was by the 5070, most of those frames were just meant to deliver the illusion of smoothness. During actual gameplay, I had a hard time seeing much slowdown with either Radeon card in 4K, and the difference between the NVIDIA card was practically erased in 1440p. Remember, fps figures aren't the entire story, even though NVIDIA wants you to believe otherwise. I kept that idea in mind as I tested FSR 4, which surprised me when it averaged around 20 fps less than FSR 3 in Call of Duty: Warzone on both GPUs. The Radeon 9070 XT reached 250 fps in 4K with FSR 3, cranked up graphics and frame generation, whereas it hit 229 fps in FSR 4. AMD tells me that's no error, it's to be expected since FSR 4's AI upscaling is more focused on delivering higher quality graphics rather than a pure frame boost. I couldn't really see a huge difference while dodging bullets in CoD, but I did notice that finer textures like chain-link fences looked a bit clearer in FSR 4. (I also saw a few random glitched textures, something that was common in DLSS upscaled games early on. They weren't a major problem, but AMD clearly has to refine its upscaling model further.) At the moment, enabling FSR 4 is a bit counter-intuitive as well. You have to turn it on in AMD's driver software, and then flip on FSR 3.1 in a compatible game. Warzone also required a reboot to fully enable the feature, but the game didn't prompt me to do so. And if you want frame generation, that's another option that has to be toggled on outside of the game. Hopefully this process will be smoothed out over time, along with wider availability for FSR 4. In addition to Black Ops 6, it's supported in the newly launched FragPunk, Civilization 7, Marvel Rivals and a handful of PlayStation 5 ports like the Spider-Man games. But it's nowhere to be found in Avowed or Dragon Age: The Veilguard, where you'll be stuck with FSR 3. For games that don't work with FSR at all, AMD's Adrenaline software also has a "HYPR-RX" mode that enables features like Radeon Super Resolution upscaling (a separate driver-level technology) and AMD's Fluid Motion Frames generation. Altogether, they led to me seeing 200 fps in Forza Horizon 5 using the RX 9070 XT in 4K with maxed out graphics settings, up from 85 fps natively. But again, those are just fps figures - AMD points out Radeon Super Resolution may not look as clear as FSR alternatives. (I didn't notice any weirdness in Forza, but I may have been distracted by the beautiful racing vistas in Mexico.) When it comes to benchmarks, the Radeon RX 9070 and 9070 XT hold their own against NVIDIA's 5070 and 5070 Ti in most of 3DMark's tests. In the Steel Nomad benchmark, the 9070 scored 1,100 points higher than the 5070, and the 9070 XT beat out the more expensive 5070 Ti by almost 3,000 points in Timespy Extreme. I was also surprised to see the RX 9070 scoring almost 2,000 points higher than the RTX 5070 in the Port Royal ray tracing benchmark. Previously, ray tracing of any kind was AMD's Achilles heel. Both of our XFX cards also stayed surprisingly cool during benchmarks and extended gaming sessions. They never went beyond 65C under load, and they typically idled below 40C. And since they never got very hot, I could barely hear their fans spinning up. On paper, both of AMD's new GPUs are compelling alternatives to NVIDIA's midrange cards for 1440p gamers who occasionally dabble in 4K. The RX 9070 XT is particularly interesting, since it's noticeably faster and still comes in $150 less than the RTX 5070 Ti. But we're also dealing with a chaotic time in the PC gaming world, where GPU stock can disappear quickly and prices can rocket up quickly. If you can nab either Radeon card at their listed prices, they'll be good deals. But it's not worth overpaying by too much for now. It also remains to be seen how the Trump administration's combative tariffs will affect pricing for PC hardware and electronics. Prices could easily jump by 20 percent or more to cover those costs. While NVIDIA's DLSS 4 technology is more mature and leads to higher interpolated frame rates, there's also a good argument for going with AMD's cards since they have 16GB of VRAM. They'll be better suited to handling larger textures in games down the line, and there's also the potential for FSR 4 to improve as well. It's clear now why AMD was focused on upgrading its mid-range Radeon cards first. There's not much point competing with NVIDIA at the extreme high end, like it did with Radeon RX 7900 XT and XTX. It makes more sense to focus on cards people can actually buy. The Radeon 9070 and 9070 XT also solve many of the problems I've had with AMD's GPUs in the past. They can go toe-to-toe with NVIDIA's cards, they have better ray tracing support and finally, they have AI upscaling. It remains to be seen if AMD will actually build on the promise of those features, but these cards are a hopeful start.

NVIDIA GeForce RTX 5070 Founders Edition Review - DLSS 4 is Still a Game Changer in 1440p

12GB of VRAM instead of 16GB feels like a misstepOnly slightly faster than the GeForce RTX 4070 SUPERNo, the RTX 5070 doesn't deliver RTX 4090 levels of performance As the Steam Hardware Survey results have shown for months, the GeForce RTX 4070 and the GeForce RTX 4070 SUPER are two of the most popular GPUs among PC gamers. The older Ampere-era GeForce RTX 3070 is right up there, showcasing that the mid-range 70-class offerings from NVIDIA often lead to the best value - while giving PC gamers access to cutting-edge technologies like ray-tracing and the hardware-intensive Full Ray Tracing. The GeForce RTX 5070 offers a decent but not mind-blowing upgrade over the GeForce RTX 4070, with only a minor upgrade over the GeForce RTX 4070 SUPER refresh from 2024. In a way, it's the least exciting release in the GeForce RTX 50 Series so far in that - outside of access to the new DLSS 4 Multi Frame Generation feature for boosting performance - it feels more like a GeForce RTX 4070 SUPER 2. More refresh than mid-generation game-changer, the GeForce RTX 5070 is still a great 1440p option - especially when you factor in the mind-blowing DLSS 4. A quick look around the NVIDIA GeForce RTX 5070 Founders Edition DLSS 4's new transformer AI model for Super Resolution and Ray Reconstruction is arguably one of the most impressive aspects of the GeForce RTX 50 Series launch. It delivers improved image quality and stability over native rendering - especially in motion. With the GeForce RTX 5070 Founders Edition, a GPU that is every bit the 1440p beast as the RTX 4070 SUPER, we spent quite a bit of time gaming with DLSS 4 in 1440p to see how the new and improved upscaler and denoiser performed when not gaming in 4K. Regarding upscaling technologies, even the most basic spatial solution can look 'okay' in 4K. However, technologies like DLSS make the most sense with GPUs other than the GeForce RTX 5090, and seeing DLSS 4 in action in 1440p (with the Quality and Balanced presets) on the GeForce RTX 5070 raises the bar for AI-powered rendering. In fact, in titles with DLSS 4 support (native or via NVIDIA's DLSS Override feature), you're looking at a 'free' double digit performance improvement from software alone. With DLSS 4 the GeForce RTX 5070 is what we'd consider an entry-level GPU for Path Tracing, where it delivers playable and impressive-looking visuals at 1440p. However, performance here is roughly the same as what you'd find with the GeForce RTX 4070 SUPER - with the option to enable Multi Frame Generation for increased smoothness. The Path Tracing gains for the GeForce RTX 5070 (and ray-tracing for that matter) sit behind the raw or rasterized performance improvements - a surprising result, as beefier RT performance would have made the GeForce RTX 5070 stand out. Thankfully, the $549 MSRP is lower than the $599 launch price of the GeForce RTX 4070 and GeForce RTX 4070 SUPER - which is excellent. However, for those eyeing the GeForce RTX 5070 there is another layer to the GPU in the form of AMD's new RDNA 4 offerings - the Radeon RX 9070 and Radeon RX 9070 XT. Our reviews of AMD's latest mid-range duo will go live in the coming days, so stay tuned. In the meantime, let's dig into the GeForce RTX 5070 - in stunning (and compact) Founders Edition form. RTX Blackwell - NVIDIA's Gaming Architecture for the AI Era Below is a summary of NVIDIA's GeForce RTX 50 Series and RTX Blackwell architecture, applicable to all models. NVIDIA describes 'Neural Rendering,' which includes all previous versions of DLSS and the brand-new DLSS 4, as the 'next era for computer graphics.' They're not alone; the Lead System Architect for the PlayStation 5 Pro console, Mark Cerny, recently said that ray-tracing is the future of games and that AI will play an integral role in making that happen. DOOM: The Dark Ages developer id Software shared a similar sentiment, adding that the arrival of DLSS was an 'inflection point' for PC game visuals and performance and on par with the arrival of dedicated GPUs and programmable shaders. With the arrival of the Blackwell generation and the GeForce RTX 50 Series, AI is now being used to accelerate programmable shaders with the brand-new RTX Neural Shaders. Yes, these are actual neural networks that use live game data, and the power of Tensor Cores to do everything from compress textures, render lifelike materials with a level of detail impossible to match using traditional rendering methods, and even use AI to partially trace rays and then infer "an infinite amount of rays and bounces for a more accurate representation of indirect lighting in the game scene." RTX Mega Geometry is incredible in its own right; it essentially increases a scene's geometry detail and complexity (triangles or polygons) by up to 100x. 100 times the detail, it's hard to wrap your head around - but the added benefit in a game like Alan Wake 2 is dramatically improving the performance of the game's Full Ray Tracing or Path Tracing mode. With DLSS 4 and RTX Neural Shaders, NVIDIA's GeForce RTX 50 Series and RTX Blackwell architecture (which includes the same AI optimizations as data center Blackwell) can be viewed as the turning point for PC gaming - the moment when AI becomes integral to everything from designing a game to programming and then finally rendering it on a 4K display to play. DLSS 4 includes more goodies than NVIDIA's highly touted new Multi Frame Generation technology, but let's start there. DLSS 3's version of Frame Generation has evolved with DLSS 4, powered by Blackwell hardware and software, and an innovative use of AI to generate frames 40% faster while using 30% less VRAM. Switching to a new model also means that Frame Generation and Multi-Frame Generation could soon come to GeForce RTX 20, 30, and RTX 40 Series owners. DLSS 4 benefits all GeForce RTX gamers. With the 5th Generation of Tensor Cores in the GeForce RTX 50 Series delivering 2.5X more AI performance, NVIDIA's latest GeForce RTX 50 Series GPUs can execute five complex AI models - covering Super Resolution, Ray Reconstruction, and Multi Frame Generation in a couple of milliseconds. Part of the reason it happens so quickly is the addition of hardware Flip Metering, which shifts frame pacing to the Blackwell display engine - the result is frame rates of up to 4K 240 FPS and higher without stuttering issues. With up to 15 of every 16 pixels generated by AI, the result is up to 8X the performance when compared to native rendering or rasterized performance. DLSS Super Resolution and Ray Reconstruction are also switching to a new 'Transformer' model, with over double the parameters and four times the compute requirement. This is one of the most exciting aspects of the GeForce RTX 50 Series, as it pushes DLSS into a new realm of image quality and performance. The best part is that it will work on all GeForce RTX GPUs; however, there will be a performance hit compared to running it on an RTX 50 Series GPU. Already available in games, DLSS 4's Transformer model is another DLSS 2.0-like moment for the technology, and the results speak for themselves. Even better, DLSS 4 is integrated into the NVIDIA App with a new 'DLSS Override' feature that allows users to experience the latest tech without waiting for a patch or game update. DLSS 4 is built to be backward compatible, with 75 games and apps supported (so far). It doesn't stop there, as the new AI Management Processor (AMP) allows AI models to share the GPU with graphics workloads. As a result, expect to see digital humans in games alongside AI assistants like NVIDIA's Project G-Assist becoming more prevalent in the coming years. This filters down to the creator side, with AI assistants for streamers, who will also benefit from the GeForce RTX 50 Series' expanded creator features. RTX Blackwell introduces 4:2:2 chroma-sampled video encoding and decoding. The ninth-generation NVENC encoder also improves AV1 and HEVC quality. RTX Blackwell is a game changer for creators and editors, especially with the new low-voltage and cutting-edge GDDR7 memory that dramatically improves memory bandwidth and speed. Here's a look at the specs for the GeForce RTX 5070 compared to the previous generation's GeForce RTX 4070, GeForce RTX 4070 SUPER, and GeForce RTX 5070 Ti. With the new GeForce RTX 50 Series and the arrival of the new RTX Blackwell architecture, one of the most surprising aspects of this new GeForce GPU generation was that it was arriving on a similar custom TSMC 4N process as the GeForce RTX 40 Series. Without the usual efficiency gains and room for more hardware that comes with a process node shrink, it's up to RTX Blackwell's architectural improvements - from neural rendering to next-gen RT Cores and improved AI hardware - to deliver. And with the GeForce RTX 5070 we probably see the most prominent example of this, with the GPU giving a clear picture of what RTX Blackwell can do. And that is, a lot more with less. Without a process node shrink, the GeForce RTX 50 Series has been labeled by some as a GeForce RTX 40 Series refresh. We alluded to this in the introduction of this review by calling the GeForce RTX 5070 a GeForce RTX 4070 SUPER 2. It is a statement that is more of a description of the RTX 5070's overall performance, not its hardware. Looking at CUDA, Tensor, and RT Cores, the GeForce RTX 5070 only has 4% more GPU hardware than the GeForce RTX 4070 it succeeds. A pretty minor upgrade, for sure, but when stacked up against the GeForce RTX 4070 SUPER, the GeForce RTX 5070 features 14% fewer CUDA, Tensor, and RT Cores. With the same VRAM capacity of 12GB, the only real notable upgrade comes with an increase to memory bandwidth - 672 GB/sec on the RTX 5070, which is 33% higher than the RTX 4070 and RTX 4070 SUPER. Add all of this together and it's enough to make the GeForce RTX 5070 20% faster than the GeForce RTX 4070, and anywhere between 4-6% faster than the GeForce RTX 4070 SUPER. The good news is that like the GeForce RTX 5070 Ti and GeForce RTX 5080, the GeForce RTX 5070 is an excellent card for overclocking, with the ability to push boost clock speeds above 3 GHz for even more performance. Stay tuned for our thoughts on this side of the GPU as we look closer at OC models from MSI and ASUS in the coming days. The GeForce RTX 5070 does deliver when it comes to creator-friendly features, future proofing, and AI performance - the latter delivering double the AI TOPS over the GeForce RTX 4070 thanks to FP4 support. Throw in DisplayPort 2.1, PCIe Gen5, and NVIDIA's latest NVENC hardware for video encoding and software improvements with things like RTX Broadcast, and the GeForce RTX 5070 is an impressive all-rounder. And with native support for RTX Neural Shaders and Multi Frame Generation, it's ready to power the biggest games of 2025, and beyond. However, when it's all said and done, we feel that the GeForce RTX 5070 would have been the hands-down winner at $549 if it included 16GB of GDDR7 memory, similar to the RTX 5070 Ti. NVIDIA's refreshed Founders Edition design for the GeForce RTX 50 Series is beautiful, modern, sleek, and stylish, making it feel more like a futuristic piece of technology than your everyday gaming GPU. In GeForce RTX 5070 form you're essentially getting the GeForce RTX 5080 Founders Edition card put through a Honey, I Shrunk the Kids machine set to 50%. It's small and compact in a way we rarely see outside of pure budget gaming options, making it an excellent choice for slotting into a tiny build. Like its larger sibling, the GeForce RTX 5070 Founders Edition sports a custom, small PCB, with two fans pulling air through the heat pipes and dense fin-stack. The build quality is exceptional and on par with the GeForce RTX 5080 and RTX 5090 Founders Edition cards, with an all-metal premium feel, two-slot thickness, and high-quality components. The new Double Flow Through design is an aesthetic winner, and like the RTX Blackwell architecture, it scales extremely well, the smaller it gets. It will be interesting to see if NVIDIA releases a Founders Edition model for the GeForce RTX 5060 or RTX 5060 Ti. With the GeForce RTX 5070's power rating of 250W, a 40% reduction from the GeForce RTX 5080, you might wonder if the design is efficient at cooling the GPU. The answer is yes. However, the GeForce RTX 5070 Founders Edition will probably sit below partner and OC cards regarding thermal performance. The Games and Tests PC gaming not only covers a wide range of genres and styles, from indie games with simple 2D graphics to massive 3D worlds lit by cutting-edge real-time ray tracing technology. With that, the needs and requirements of each gamer vary. High refresh rates and latency reduction become more important than flashy visuals or playing at the highest resolution possible for those who live and breathe fast-paced competitive games. For those who want to live in a cinematic world and become a key player in an expansive narrative, ray-tracing, and high-fidelity visuals are a stepping stone toward immersion. Our chosen benchmarks cover various games, engines, APIs, and technologies. For the GeForce RTX 5070 Ti, all tests are run at 4K and 1440p and include results for performance-boosting Super Resolution technologies like NVIDIA DLSS 4 - including Frame Generation and the new Multi Frame Generation. In many ways, DLSS numbers are more important in 2025 than native rendering - a title with ray tracing isn't meant to be played without Super Resolution. Also, DLSS technologies like Ray Reconstruction and the new RTX Mega Geometry dramatically improve visual fidelity and detail compared to native rendering. However, our benchmark results are still sorted using 'raw performance' or native rendering. Here's the breakdown of games, settings, and what's being tested. As we start seeing more mid-range and mainstream GPU releases, we will add more 1080p results to showcase how the various new GPUs from NVIDIA, AMD, and Intel perform. Looking at average performance, the GeForce RTX 5070 Founders Edition is 15.3% faster than the GeForce RTX 4070 at this resolution while only 11% slower than the GeForce RTX 5070 Ti. In 2025, 1080p is the realm of more budget or mainstream GPUs, with the GeForce RTX 5070 Ti making the most sense at 1440p. However, for competitive gamers, the GeForce RTX 5070 absolutely delivers this resolution even with Ultra quality visual settings. 127 FPS in Call of Duty: Black Ops 6, a whopping 344 FPS in Counter-Strike 2, and 140 FPS in Marvel Rivals. Average Gaming Performance - 1440p Results The GeForce RTX 5070 is an excellent GPU for 1440p gaming. In this resolution, DLSS adds an extra level of performance while also improving image fidelity in titles that support DLSS 4's new AI transformer model. With DLSS, using the Quality preset, the 133 FPS average becomes 135 FPS, to deliver 20% faster performance than the previous generation's GeForce RTX 4070. Looking back even further, the GeForce RTX 5070 is 45% faster than the GeForce RTX 3070 Ti at this resolution. Unfortunately we didn't have time to fully re-test the GeForce RTX 3070 for this review, but it's safe to say that the GeForce RTX 5070 is a significant upgrade over this popular Ampere-era card. However, looking at the GeForce RTX 4070 SUPER and the GeForce RTX 5070 is only 3.6% faster - on average. A result that warrants looking into the individual game results. In titles like Black Myth: Wukong, Counter-Strike 2, Cyberpunk 2077 (without ray-tracing), and Total War: Warhammer III, and the GeForce RTX 5070 is on par or faster than the GeForce RTX 4070 Ti. However, switch to Call of Duty: Black Ops 6 or DOOM Eternal and the GeForce RTX 5070 is slower than the GeForce RTX 4070 SUPER - which is disappointing. DLSS 4 is a definite game changer at this resolution, especially for titles with ray-tracing. Cyberpunk 2077, with the RT Ultra preset, sees 45 FPS become 74 FPS. This is a massive 64% improvement in performance that arrives via AI hardware and software. The game also looks significantly better thanks to DLSS Super Resolution and Ray Reconstruction. Average Gaming Performance - 4K Results With a 63 FPS average that increases to 84 FPS when you factor in DLSS 'Quality' upscaling, it's safe to say that the GeForce RTX 5070 is a GPU that can be used for 4K gaming. At this resolution, the lead over the GeForce RTX 4070 increases to 23.5%, with the lead over the GeForce RTX 4070 SUPER increasing to 6.8%. It's also 53.7% faster than the GeForce RTX 3070 Ti. 4K performance is roughly on par with the GeForce RTX 4070 Ti and Radeon RX 7900 XT, which does decide to stick with 12GB of VRAM over bumping it up to 16GB, looks like a mistake. The higher the resolution, the more critical VRAM capacity and memory bandwidth become. So, having the increased speed that comes with GDDR7 but not the capacity makes it feel like something's missing. 12GB is not a deal breaker, it only means that in some titles you'll need to play around with settings and things like texture pool sizes - which is par for the course as the GeForce RTX 5070 is not a 4K 'max settings' GPU. Thankfully, NVIDIA's updated Frame Generation now uses significantly less VRAM than the previous version and upcoming RTX Neural Shaders leveraging AI for texture compression will open the door for more hardware intensive ray-tracing on the GeForce RTX 5070. The GeForce RTX 5070 Ti is the more potent 4K gaming GPU, which is to be expected. Benchmarks - 3DMark Synthetic Tests 3DMark offers a suite of synthetic benchmarks built to test GPUs in various scenarios. 3DMark Steel Nomad is a cutting-edge DirectX 12 benchmark with newer, modern rendering techniques designed to push GPUs to their limit. The 'Light' version tests at 1440p, while the main Steel Nomad benchmark tests pure native 4K rendering. Port Royal is a benchmark focusing exclusively on real-time ray tracing for lighting effects like reflections, shadows, and more. Built for modern gaming, the 3DMark Steel Nomad Light synthetic benchmark results are a relatively accurate representation of what you'll find regarding GeForce RTX 5070 performance. The score for this 1440p test, which doesn't include upscaling or ray-tracing, sees RTX 5070 performance sit slightly above the GeForce RTX 4070 SUPER to join the likes of the GeForce RTX 4070 Ti and the Radeon RX 7900 XT. 3DMark Port Royal is a pure ray-tracing benchmark and here the results reflect what we see in several titles with ray-tracing - there's not a lot separating the GeForce RTX 4070 SUPER, GeForce RTX 4070 Ti, and the new GeForce RTX 5070. It's also a reminder that even though you're looking at ray-tracing performance that is up to 35% faster than the GeForce RTX 4070, it's not a massive leap forward over the GeForce RTX 4070 SUPER - which is something you see first-hand when you enable Path Tracing or Full Ray Tracing. DLSS 4, Frame Generation, and the new Multi-Frame Generation DLSS 4 and Multi Frame Generation are impressive bits of technology, thanks mainly to the overall improvements to performance and latency on the Frame Generation side and the new 'Transformer' model for Super Resolution and Ray Reconstruction. We used the DLSS 'Quality' mode preset for these benchmarks, often delivering better-than-native image quality. Paired with the right game, settings, and Frame Generation, and the new Multi Frame Generation can be a game changer. Exclusive to the new GeForce RTX 50 Series, Multi Frame Generation gives RTX 5070 owners the option to choose between 2X, 3X, or 4X - which refers to one or three AI-generated frames that will be rendered on top of the standard DLSS frame. As long as the input frame rate is ideally above 70 FPS, then the minimal increase in latency is barely noticeable, thanks to excellent frame-pacing and NVIDIA's Reflex technology doing its magic. Cyberpunk 2077 with RT Ultra enabled, the 126 FPS with Frame Generation and the 218 FPS of Multi Frame Generation look excellent, smooth, and responsive. One of the benefits of Frame Generation and Multi Frame Generation is that the increased smoothness is immediately noticeable as it improves motion clarity. However, this benefit is only felt when you've got a display that can match the FPS number you see in refresh rate. For most displays this would mean toning down the 317 FPS of Dragon Age: The Veilguard to Frame Generation 2X or 3X. This might be sacrilege, but Frame Generation works surprisingly well in Marvel Rivals as the increased smoothness can make it easier to play and notice things on the battlefield. Path Tracing Performance - 1440p Path Tracing, or Full Ray Tracing, arrived with the GeForce RTX 40 Series and DLSS 3 and is leveling up with the GeForce RTX 50 Series and DLSS 4. It's only possible thanks to AI technologies like DLSS Super Resolution, Ray Reconstruction, and RTX Neural Shader technology like RTX Mega Geometry. It's designed specifically for these technologies, and we're only including native or rasterized performance to highlight just how intensive it is on a GPU like the GeForce RTX 5070. In fact, outside of the massive increase in performance, these games also look notably worse without DLSS 4. It's great to see Path Tracing running on the GeForce RTX 5070, however, the performance here - outside of Multi Frame Generation - is on par with what you can achieve on the GeForce RTX 4070 SUPER. Playable? Definitely. But there's not a lot of headroom, and you might need to tweak settings to get the best Frame Generation and Multi Frame Generation results. DLSS 4 is the real star and the secret weapon alongside new technologies like RTX Mega Geometry. Going from 36 FPS to 73 FPS in Alan Wake 2 is not only a massive 103% increase to performance - but the new transformer model, alongside RTX Mega Geometry, delivers a more detailed image. It looks stunning, and it's the sort of visual fidelity and performance currently impossible to achieve on any gaming hardware that doesn't feature 'GeForce RTX' in the naming. Indiana Jones and the Great Circle is probably our favorite example of Path Tracing or Full Ray Tracing as the game is built on a custom version of id Tech. The same engine that is powering the upcoming DOOM: The Dark Ages. You can run Path Tracing on the GeForce RTX 5070 and achieve a playable frame rate without DLSS. However, as the game now supports DLSS 4, the 57 FPS to 87 FPS jump is accompanied by a light switch or light bulb-style moment where the visuals dramatically improve to reveal one of the best-looking games ever made. This title still looks fantastic without Path Tracing, but if you were to do a side-by-side comparison it would be like looking at the same game running on a PlayStation 4 versus the PlayStation 5 Pro. Temperature and Power Efficiency NVIDIA's GeForce RTX 5070 Founders Edition is a quiet GPU; however, it runs more remarkably than some of the GeForce RTX 50 Series partner models we've been testing in recent weeks. If the GeForce RTX 5070 Ti and RTX 5080 are anything to go by, we fully expect to see overclocked GeForce RTX 5070 GPUs push the boost clock to over 3 GHz for a 5% or so increase to performance and run cooler than the Founders Edition variant. That said, NVIDIA's design is super compact and stylish, and there's enough headroom to overclock and boost performance - however, you might want to check out other models if you want to overclock the RTX 5070. Final Thoughts DLSS has come into its own with the arrival of the GeForce RTX 50 Series, and with widespread game support, it's reason enough to pick up a GeForce RTX GPU - which says a lot. 20% faster raw performance than the GeForce RTX 4070 for 1440p and 4K gaming is a decent uplift, but not groundbreaking. It's enough to offer fantastic 1440p gaming performance, but the incredible DLSS 4 is the icing. DLSS 4 delivers a free double-digit boost to performance that remarkably makes games look better, even when enabled in 1440p. That said, the GeForce RTX 4070 SUPER's existence makes the GeForce RTX 5070 feel less exciting, which shouldn't have been the case. The GeForce RTX 5070 features 14% fewer CUDA, Tensor, and RT Cores than last year's SUPER refresh; however, the 6.8% improvement in performance in 4K is not all that noticeable. And yes, in some games the GeForce RTX 4070 SUPER is faster - which is strange. Path Tracing performance is virtually the same, making the GeForce RTX 5070 an entry-level Path Tracing GPU (regarding image fidelity and resolution) and not the game changer we hoped to see. Of all the GeForce RTX 50 Series GPUs released so far, the GeForce RTX 5070 will be the one that sells the most and the quickest to skyrocket up the Steam Hardware charts. With stiff competition from AMD (stay tuned for our RDNA 4 reviews), we're hoping that this is one release that offers several MSRP models as the performance you get makes the most sense when viewed from the lens of it costing $549. At this price, it's the best value offering in the GeForce RTX 50 Series lineup, and it has an excellent GPU to power your 1440p gaming for the next few years.

Nvidia GeForce RTX 5070 review: No, it's not "4090 performance at $549

"4090 performance at $549." That's what Nvidia CEO Jensen Huang said of the GeForce RTX 5070 when he announced the card at CES in January. Thanks to AI, this new midrange GPU would be able to match the frame rates of what had been the fastest consumer GPU that had previously existed, for around one-third the price. Let's dispel that notion up front. No, the GeForce RTX 5070 is not as fast as an RTX 4090, not without some very creative comparing of non-comparable numbers. Per usual for the 50-series, Nvidia is leaning on its AI-generated interpolated frames for the bulk of its claimed performance improvements. In terms of actual rendering speed, the 5070 isn't even as fast as a 4080 or a 4070 Ti. It's barely faster than last year's 4070 Super, and with disproportionately higher power usage. The RTX 5070 Founders Edition is a lot smaller than the 5090/5080 design. Andrew Cunningham The RTX 5070 Founders Edition is a lot smaller than the 5090/5080 design. Andrew Cunningham Like the 5090/5080, the 5070 switches to a slightly angled 12-pin power connector. Andrew Cunningham Like the 5090/5080, the 5070 switches to a slightly angled 12-pin power connector. Andrew Cunningham The RTX 5070 Founders Edition is a lot smaller than the 5090/5080 design. Andrew Cunningham Like the 5090/5080, the 5070 switches to a slightly angled 12-pin power connector. Andrew Cunningham For all that, it's still not necessarily a bad card -- it's a little faster than the $599 4070 Super at the same $549 price that Nvidia used for the RTX 4070. It still represents Nvidia's minimum viable 4K GPU, once you factor in DLSS upscaling, and it's well suited for people with 1440p monitors. But it's a hard card to get excited about, and unlike every other 50-series GPU released so far, it actually has to compete against something -- AMD's new Radeon RX 9070 series. I can't talk in detail about how these cards all stack up until later this week, but suffice it to say that Nvidia could use something a bit more impressive than the RTX 5070 at this price. GeForce RTX 5070 specs RTX 5070 Ti RTX 4070 Ti Super RTX 5070 RTX 4070 Super RTX 4070 CUDA Cores 8,960 8,448 6,144 7,168 5,888 Boost Clock 2,452 MHz 2,610 MHz 2,512 MHz 2,475 MHz 2,475 MHz Memory Bus Width 256-bit 256-bit 192-bit 192-bit 192-bit Memory Bandwidth 896 GB/s 672 GB/s 672 GB/s 504 GB/s 504 GB/s Memory size 16GB GDDR7 16GB GDDR6X 12GB GDDR7 12GB GDDR6X 12GB GDDR6X TGP 300 W 285 W 250 W 220 W 200 W The 5070's CUDA core count falls right in between the RTX 4070 and the 4070 Super's, where the 5080 and 5070 Ti both included small increases from their previous generation counterparts. To boost speeds over the 4070 Super, this means Nvidia is leaning on architectural improvements in Blackwell, the memory bandwidth increase from GDDR7 (a fairly significant 33 percent increase), and increased GPU clock speeds (the memory switch and clock speed boosts likely explain the higher power consumption). The 5070 gives you 12GB of memory, which still clears that crucial "more than 8GB" bar that you'll want for higher-than-1080p resolutions and ultra-quality textures, though as time goes on it may be more limiting at 4K if that's a resolution you're hoping to hit with this card. The number that sticks out in a bad way is the power consumption -- a maximum of 250 W under load, 30 W more than the old 4070 Super. The 5070 series is still far more power-efficient than older cards like the RTX 3070 or 3080, but efficiency is no longer as much of an advantage for the 5070 as it was for the 4070 series. We reviewed the Founders Edition of the card provided by Nvidia, and it uses a different two-slot design than the much larger, much more powerful RTX 5090 and 5080 cards. It's the same physical size as the 4070 and 4060 Founders Edition cards, but with an updated power connector that's set at an angle instead of sticking out directly from the top of the card. Nvidia has updated the cooler design, too, moving both fans to the same side of the card rather than placing one on each side. Whether it's because of the updated cooler or the higher power consumption or both, the 5070 actually runs hotter than any other card in our current round of testing, including (by a few degrees) the 5090. The card runs a bit loud under load -- that's not a scientific measurement, but I noticed it more than I did when testing either the 5090 Founders Edition or the 4070-series Founders Editions. It's something to keep an eye on for any smaller or lower-end 5070 cards from partners that use smaller double-fan cooler designs, though we doubt it will be an issue for the triple-fan monstrosities. Testbed notes Gaming testbed CPU AMD Ryzen 7 9800X3D (provided by AMD) Motherboard Asus ROG Crosshair X670E Hero (provided by AMD) RAM 32GB (2x16GB) G.Skill Trident Z5 Neo RGB series (provided by AMD), running at DDR5-6000 SSD 1TB Samsung 990 Pro/1TB Crucial P3 Plus Power supply Thermaltake Toughpower GF A3 1050 W CPU cooler 360 mm MSI MAG CoreLiquid I360 Case Montech XR ATX Mid-tower with three 120 mm cooling fans installed and side panel removed OS Windows 11 24H2 with Core Isolation on, Memory Integrity off Drivers Nvidia RTX 5090: Beta driver 571.86 Nvidia RTX 5080: Beta driver 572.12 Nvidia RTX 5070 Ti: Beta driver 572.43 Nvidia RTX 5070: Beta driver 572.50 Other Nvidia cards: Game Ready Driver 566.36 AMD cards: Adrenalin 24.12.1 Our testbed is the same as it's been since we updated it for the RTX 5090 review -- a Ryzen 7 9800X3D is the heart of the build to minimize the likelihood that CPU performance will cap any of our GPUs' frame rates. We've tested the 5070 at 4K and 1440p. Cards in this $500 to $600 price range tend to be a better fit for a 1440p high-refresh-rate gaming monitor, especially if you're hoping to play without DLSS or other upscaling technologies available. But if you play older or lighter games, if you don't mind upscalers, and if you can turn down a few settings, hitting 60 frames per second at 4K is often within the realm of possibility. It's certainly the least I would spend if I was hoping to play on a 4K monitor. Performance and power 4K game benchmarks. 4K game benchmarks. The 5070 usually represents a mid-single-digit performance increase over the 4070 Super, despite having fewer CUDA cores. It's more-or-less 20 percent faster than the RTX 4070 -- a comparison Nvidia would probably prefer, since the 5070 is technically replacing the 4070 at $549 and not the 4070 Super at $599. But the launch MSRPs for the 50-series have been mostly imaginary so far, due to a combination of supply, demand, and people buying cards just so they can resell them for more money. We'd assume the same for the RTX 5070 until proven otherwise. There were a couple of exceptions in our suite that performed strangely no matter how many times we re-tested them or how many times we tried clearing our shader cache or reinstalling drivers, and one game that benefitted from an outsized increase. Cyberpunk 2077 in Ultra mode with ray-tracing disabled showed a larger improvement than the other games we tested -- 24 percent faster than the 4070 Super, and 72 percent faster than the 4070. It was an outlier, but some games may benefit more than others. Horizon Zero Dawn Remastered and Cyberpunk 2077 in Overdrive mode with DLSS turned off all gave abnormally low or otherwise strange numbers -- we've presented those here for the sake of transparency, and we've communicated our findings to Nvidia, but we'd take these with a grain of salt when comparing results. Horizon seems to have some kind of issue with DLSS frame generation regardless of GPU, with low frame rates and odd visual artifacts across all the GeForce cards we've tested with recent drivers. 4K benchmarks with DLSS/FSR upscaling and single frame-generation enabled. 4K benchmarks with DLSS/FSR upscaling and single frame-generation enabled. On the AMD side, we can't compare the 5070 directly to the RX 9070 series yet, but it's worth noting that the 5070 outperforms the old RX 7900 GRE by a small but consistent amount in our benchmarks without ray-tracing effects (the gap in ray-traced games is larger because of AMD's typical performance deficit in these games). AMD says that the $549 RX 9070 and $599 9070 XT should be 21 percent and 42 percent faster (respectively) than the RX 7900 GRE on average. That could spell trouble for the 5070, but it remains to be seen whether AMD has fixed its ray-tracing performance problems and how much the cards' power efficiency has improved. 1440p benchmarks. 1440p benchmarks. We ran our 1440p benchmarks on the GeForce RTX 3070, too, to give a bit more context for upgraders. The 5070 is usually between 40 and 70 percent faster than the 3070 here, a card that will find itself more and more limited by its 8GB of RAM as time goes on. The 5070 is still a good upgrade from the 3070, it's just not all that much better of an upgrade than what you could get from a 4070 Super a year ago. As usual, the older your card is, the better the upgrade will feel, even if the generation-over-generation improvement is small. In exchange for five or six percent better performance, the 5070 uses 13.5 percent more power under load than the 4070 Super did. Nvidia was so far ahead of AMD and the RTX 30-series on efficiency with the RTX 40-series GPUs that one relatively inefficient upgrade cycle doesn't really wreck things, but we do generally like to see power consumption either increase proportionally with performance, or for performance to increase more than power consumption does. Improving efficiency was always going to be tough for the 50-series, since the Blackwell GPU dies are manufactured using a 4nm TSMC process similar to what Nvidia used for the Ada Lovelace GPU dies for the 40-series. But the 5070 ends up slightly worse on this metric than the other 50-series cards. Frame generation comparisons and caveats When you turn frame generation on, you can see the source of Jensen Huang's "4090 performance for $549" performance claim. The 5070 with DLSS Multi-Frame Generation (MFG) enabled in 4x mode has a similar average frame rate in Cyberpunk 2077 compared to the RTX 4090 with the old Frame Generation mode enabled (now labeled "2x" in games that support MFG). That's because the 5070 is capable of generating three AI-interpolated frames for every rendered frame, rather than one. But the 5070 (and, when they arrive, the 5060 series) demonstrates the limits of FG and MFG more than the other 50-series cards do. With MFG enabled, Cyberpunk 2077 can hit over 80 frames per second on average with the Overdrive preset enabled. That sounds pretty good! Until you consider that the base frame rate (with DLSS upscaling enabled, but not any frame generation) is closer to 20 frames per second. With a base frame rate this low, user input will feel sluggish, and visual artifacts are plainly visible when objects are in motion. DLSS MFG can still have its uses -- the 5070 can push some pretty high frame rates at 1440p, and MFG could help you take better advantage of a 240 Hz or 360 Hz 1440p monitor than you could without MFG enabled. But it remains a relatively niche technology that's useful under certain circumstances, not the broad cure-all for mediocre performance that Nvidia's charts imply it is. It could be better, or at least cheaper The GeForce RTX 4070 and 4070 Super were some of the best values in Nvidia's lineup last generation, offering the same performance as high-end 30-series cards and outstanding power efficiency for a little less money than you'd spend on an RTX 3080 or 3090. Now that it's here, it's hard to classify the RTX 5070 as anything other than a disappointment. It's just barely, by the skin of its figurative teeth, faster than the 4070 Super for the same $549 MSRP as the regular 4070. This technically makes it an improvement, along with support for multi-frame generation. But this nearly imperceptible performance improvement comes with a 13.5 percent increase in power consumption under load, which seems like a bad trade-off any way you cut it. The RTX 5070 feels like the kind of product you make when you're not particularly worried about what your competition is doing. The entire 50-series has sort of felt like that so far, between the astronomically high price of the RTX 5090 and the so-so performance improvements for the 5080 and 5070 Ti. But it's different for the RTX 5070 because AMD actually has an answer for it coming soon. AMD's RX 7600, 7700, and 7800 GPUs may not have been smash-hit sales successes for its last generation, but stiffer competition at and under $600 did help pull Nvidia's pricing down to earth a bit. Soon we'll know whether the RX 9070 series will exert a similar pull -- or if it can actually manage to wrest some market share back from Nvidia after years of trending in the wrong direction. The good Great 1440p performance and can stretch to 4K with the right settings. Same price as the old RX 4070 and a little cheaper than the 4070 Super (in theory). Compact design that should fit in just about any case. Reasonably power-efficient when compared to older RTX 30-series or Radeon cards. DLSS Multi-Frame Generation is a situationally useful technology. Wide ecosystem support for DLSS, CUDA, and other Nvidia software features. The bad Higher power consumption for not that much better performance. A few odd performance results that will hopefully be ironed out by driver and/or game updates. Founders Edition runs a bit hot. 12GB of RAM feels stingy. The ugly Nvidia's grandiose apples-to-oranges performance comparisons. Availability remains a big question mark.

Nvidia GeForce RTX 5070 Review

Nvidia has been having a hard time with its latest line of graphics cards. While the RTX 5070 and all of its more expensive brethren are fine, they have failed to deliver the kind of generational leaps you would expect after two years. The RTX 5070 is the worst example of this yet. Because while every RTX 50-series card, from the RTX 5070 Ti to the RTX 5080 has a small lead over their Super-variant counterparts from the last generation, the RTX 5070 is straight up tied with the 4070 Super - and is even slower in a lot of games. So while at $549 the Nvidia GeForce RTX 5070 is technically $50 cheaper than the RTX 4070 Super was at launch - if launch prices hold up at least - you're essentially just getting a minor price cut on an existing card. That's not an exciting proposition, even when Nvidia throws in DLSS 4 multi-frame generation to help niche users with high-refresh monitors. Specs and Features Just like the RTX 5070 Ti before it, the RTX 5070 is built on the Blackwell architecture. Before the RTX 5090 launched in January, Blackwell was typically known as the GPU tech behind the supercomputers powering cloud-based AI applications. So, while there is a bit of improvement to the CUDA cores, Nvidia's real focus was on its new Tensor cores, which power DLSS 4. The RTX 5070 itself features 6,144 CUDA cores, spread across 48 Streaming Multiprocessors, or SM. Each of these SMs also has an RT Core and four Tensor Cores, making for 48, and 192, respectively. Therein lies the problem though. While this is essentially a similar configuration to the RTX 4070 - which had 46 SMs, the RTX 4070 Super, which replaced it in 2024, has 56 SMs. So while each SM in a Blackwell GPU is a bit more powerful, the architecture doesn't provide enough of a bump to make up for the sheer difference in silicon. The RTX 5070 also has the same amount of VRAM as its predecessor, clocking in at 12GB. However, unlike its predecessor, it's upgraded to GDDR7, which has higher bandwidth, even though the RTX 5070 is on the same 192-bit memory bus as the RTX 4070 and the 4070 Super. The faster memory does help in games that are demanding on memory, as it's able to supply data to the GPU faster. However, with the AMD Radeon RX 9070 launching soon for the same price with 16GB of VRAM, it would have been nice to see Nvidia provide a bit more frame buffer. Continuing in the tradition of Blackwell graphics cards needing more power than their predecessors, the Nvidia GeForce RTX 5070 requires 250W of power, with Nvidia recommending a 650W power supply to run the thing. That's a slight increase over the 220W of the RTX 4070 Super or the 200W of the RTX 4070, but its still low enough to only need two 8-pin PCIe cables to power it, even if you'll have to plug them into the included 12V-6x6 adapter. There is a new bit of silicon in the RTX 5070, though: an AI Management Processor, or AMP. This core, which sits on its own outside of the SMs, intelligently assigns work to different parts of the GPU, a task that has traditionally been handled by the CPU. This is the secret sauce that allows Nvidia to push its frame generation tech to generate multiple frames off of a single rendered frame, as it now has a local core that paces out the frames to avoid artifacting or uneven frame times. It also allows DLSS to have its biggest fundamental change since it was first implemented in the RTX 2080 in 2018. Now, instead of the Convolutional Neural Network (CNN) that DLSS ran on originally, DLSS can now run on a Transformer Model. This doesn't make it any faster per se, but it does greatly improve accuracy, making for a cleaner upscaled image with less blur or ghosting artifacts. DLSS 4 What Nvidia is hoping will really sell the RTX 5070, along with the rest of the RTX 5000 lineup, is DLSS 4. Now, DLSS has been around for years and has grown to be the standard upscaling solution. And while DLSS 4 does improve the upscaling with a Transformer model that's much more accurate, the real push this time around is Multi-Frame Generation, or MFG. Frame generation debuted with the RTX 4090 a couple years ago. The idea originally was that your GPU would render two frames, and then generate an AI-interpolated frame in between the frames, in order to boost your frame rate. This approach drastically improves frame rates, but because the GPU is essentially holding a rendered frame hostage until the generated frame is produced, it could increase latency slightly - usually by a few milliseconds. To compensate for the increased latency, Nvidia Reflex is required in order to enable any kind of frame generation. Essentially, Reflex zeroes out your render queue, and synchronizes your CPU and GPU so that frames are rendered just as they're needed, rather than having a queue of frames waiting for your GPU to render them. This can dramatically reduce latency, making the added latency in Frame Generation less noticeable. With the RTX 5070, however, Nvidia now supports Multi-Frame Generation, so instead of having one AI-generated frame based on each rendered frame, the GPU can generate up to 3 frames at a time. Obviously, this makes your frame rate go up significantly, but it does add a bit more latency to the equation. In Cyberpunk 2077, at 1440p and with the Ray Tracing Overdrive preset, the Nvidia RTX 5070 gets about 70 fps with an average PC latency of 31ms. That's not bad, and is completely playable, but what if you want even more frames? Well, with 2x frame gen - what's supported on the last-generation hardware - the framerate nearly doubles, up to 116 fps, and with 36ms of latency. Frame gen directly adds 5ms of latency, but that's not much at all. With Frame Gen set to 4x, however, that framerate goes up to 200 fps, nearly tripling the original frame rate. Though, the latency gets a bit worse, going up to 38ms of average latency. Keep in mind, though, that MFG isn't there to turn an unplayable frame rate into a smooth experience. If you're playing a game that's only getting 20 frames per second, frame generation will make the frame rate go up, but you'll still be subjected to the high response times that come with low frame rates. MFG also needs more frames in order to accurately predict how objects in the scene are moving from frame to frame, and a lower frame rate means there are bigger differences between each rendered frame. This makes it harder for the algorithm to accurately generate images, and will introduce errors and artifacts into the scene. So, don't turn on frame gen unless you're already getting something like 60 fps already. Multi-Frame Generation, then, makes for a slightly less responsive gaming experience, but with more frames being sent to your display. If you have a monitor with a high refresh rate, this results in better visual smoothness, even if you're not getting the boosts to responsiveness you'd get from faster native rendering. This makes it hard to recommend for anyone playing fast-paced shooters like Black Ops 6 or Marvel Rivals, but is absolutely worth it if you play single-player games like Cyberpunk 2077 or Indiana Jones and the Great Circle. Performance When a new generation of graphics cards come out, it's reasonable to expect them to have a big improvement over its predecessor. However, while the Nvidia GeForce RTX 5070 has a decent uplift over the RTX 4070, it falls on its face as soon as the RTX 4070 Super enters the conversation. The RTX 4070 Super hit the market about a year after the original RTX 4070, replacing it at its $599 price, and delivering huge improvements in performance, especially for a mid-generation refresh. That makes it a hard act to follow, and unfortunately, the RTX 5070 does not pull it off. Across all tests at all resolutions, the RTX 5070 was on average just 4% faster than the RTX 4070 Super, even while it was 20% faster than the original RTX 4070. When I dial in the results to only include 1440p, that number does improve a bit to a 5% lead over the 4070 Super. However, take the synthetic 3DMark tests out of the equation, and the 1440p improvement is just 3%. What makes it even worse is that there's one game where the RTX 5070 soars over the 4070 Super: Total War: Warhammer 3. Take that one game out of the equation, and the average improvement over the RTX 4070 Super is 0%. That's embarrassing, but it actually feels worse when I look at it game-by-game. I tested every graphics card on their current public drivers. That means all Nvidia cards were tested on Game Ready Driver 572.60, and all AMD cards were tested on Adrenalin 24.12.1. The Nvidia GeForce RTX 5070 itself was tested on a review driver provided by Nvidia. 3DMark doesn't necessarily reflect real-world gaming performance, but it does give a good glimpse at what the potential of a graphics card could be. In Speed Way, the RTX 5070 gets 5,845 points, compared to 5,190 from the RTX 4070 Super and 4,498 from the RTX 4070. That's a small 12% lead over the RTX 4070 Super, but it's a wider lead against the original 4070. However, the leads narrow in the next couple tests. In Steel Nomad, the RTX 5070 only leads the 4070 Super by 10%, with only an 8% lead in Port Royal. The problems start to show up in Call of Duty: Black Ops 6. Rather than getting a performance lead over its predecessor, the RTX 5070 is 5% slower than the RTX 4070 Super, while maintaining just a 7% lead over the RTX 4070 at 1440p. At 131 fps, it's far from slow, but it's still slower than what came before. In Cyberpunk 2077 the losses continue at 1440p. The RTX 5070 gets a respectable 90 fps on the Ray Tracing Ultra preset with DLSS set to balanced, but the RTX 4070 Super gets 96 fps, making for a 6% performance loss gen-on-gen. What's more, the RTX 5070 is actually slower than the AMD Radeon RX 7900 GRE, which gets 143 fps in the same test, with FSR set to balanced. That doesn't bode well, with the Radeon RX 9070 around the corner. I tested Metro Exodus: Enhanced Edition with no upscaling, because it only supports DLSS, which means I can't get comparable results from AMD graphics cards. But even here, the RTX 5070 loses 4% performance over the RTX 4070 Super, even if it's 19% faster than the RTX 4070. Luckily, the RTX 5070 stays over 60 fps, which is saying something. The RTX 5070 gets 115 fps in Red Dead Redemption 2 at 1440p, which is within the margin of error with the 4070 Super's 116 fps. But that's still a dead-even result with a one-year gap and a new architecture. Total War: Warhammer 3 is a bit of an outlier, though. The RTX 5070 gets 134 fps in this game, which is a 22% lead over the RTX 4070 Super's 110 fps. That's just a single game, but it does perhaps suggest the RTX 5070 has a bit of room to grow in the future. This is likely due to the improved memory bandwidth. Assassins Creed Mirage is another wash, with the RTX 5070 getting 163 fps to the RTX 4070 Super's 165 fps. That's within margin of error, but it's still a 1% loss for the new graphics card. This game also shows the lowest improvement over the RTX 4070, which gets 151 fps: that means the RTX 5070 only has an 8% lead here. In Black Myth: Wukong, the RTX 5070 pulls ahead of the RTX 4070 Super, but just barely. It gets 66 fps at 1440p, compared to 64 fps from the RTX 4070 Super and 55 from the original 4070. But that's still just a 3% lead over the RTX 4070 Super. Luckily, Forza Horizon 5 looks much better for the RTX 5070. In this game, the RTX 5070 gets 168 fps at 1440p with the Extreme preset and DLSS set to Balanced. With the same settings, the RTX 4070 Super gets 149 fps, which makes for a 13% lead in favor of the 5070. That's more in line with the potential shown by 3DMark, even if it's not the most exciting generational improvement. It would have been bad enough if the RTX 5070 had single-digit improvements over the RTX 4070 Super, but with so many games showing the new GPU being slower than its predecessor, it becomes really hard to see why anyone should buy it. The only reason you should get this is if you have an older graphics card and you don't have the cash to spend on an RTX 5070 Ti. And even then, you're better off getting an RTX 4070 Super - if you can find one. Jackie Thomas is the Hardware and Buying Guides Editor at IGN and the PC components queen. You can follow her @Jackiecobra

No, the RTX 5070 isn't an RTX 4090 rival, but I'd still call it a solid 4K GPU for under $600

Why you can trust GamesRadar+ Our experts review games, movies and tech over countless hours, so you can choose the best for you. Find out more about our reviews policy. The Nvidia GeForce RTX 5070 doesn't truly pack RTX 4090 performance, at least not without AI shenanigans. Does that stop the GPU from potentially being a mid-range champion this generation? Absolutely not, but it heavily relies on DLSS and Multi-Frame Generation to get anywhere near close to last-gen flagship frame rates, and that might not be every player's bag. Rather than being a RTX 4090 rival, the GeForce RTX 5070 will furnish your eyeballs with approachable mid-range visuals for $549 / £539. At least, it should do if you can find any RTX 5070 stock at MSRP, and if the rest of Nvidia's best graphics card contenders this generation are anything to go by, that could be painful. The moment you step into the wasteland of inflated AIB partner pricing, the UHD door practically slams shut on most mid-range player faces, including mine. If the majority of RTX 5070 options can stay under $600, I'm confident that it'll serve as the graphics card most players should pick up for solid mid-range performance. Whether it'll actually get to sit pretty on that particular throne really hinges on whether AMD ends up providing more reasons for middle ground players to buy an RX 9070 series instead. I'll happily watch both scrap it out if it means 4K gaming is technically getting cheaper, and I'd say things are starting to look up for the mid-range scene. Kitted out with a Blackwell GB205 GPU, and 12GB VRAM, the RTX 5070 isn't drastically different to the RTX 4070 Super. In fact, more of the improvements come in the form of faster GDDR7 memory modules, next-gen cores, and faster boost clock speeds. On paper, the card's core count is actually lower, but that's largely the result of optimized architecture rather than reduced specs. Custom cards like the Asus RTX 5070 Prime OC boast faster clock speeds, so that's worth keeping note of if you're not gunning for a Founders Edition. Nvidia's FE models are usually the first to sell out, and since there will be factory overclocked cards theoretically available at MSRP, you'll at least end up with something a little faster and with the cooling setup to pull off further overclocking. In short, the RTX 5070 is pretty distant from its beefier Nvidia GeForce RTX 5070 Ti in terms of specs. In truth, that GPU feels closer to a GeForce RTX 5080 variant, and the 70-class model is firmly a mid-range graphics card. It is a pity that the card is rocking 12GB VRAM rather than 16GB again this generation, and while the extra memory might not be required with DLSS on, it would help futureproof the card. For what feels like years now, I've been asking myself why GPUs need to be almost the size of a phat PS2. I get that the new age of UHD gaming requires chonky coolers to keep temperatures at acceptable levels, but I've still been longing for a new card that's around the same size as my trusty old GTX 970. Well, we're finally getting that kind of design back with the GeForce RTX 5070. I've been testing two models - Nvidia's Founder's Edition and the Asus Prime OC, and the latter is kitted out with a smaller two-slot shroud that effectively follows the same blueprint as the GeForce RTX 5080. Its sleek, two-tone grey body isn't going to remotely dominate your PC case, and I can see this card making its way into plenty of small form-factor builds over the next two years. Just like with the RTX 5080, the new RTX 5070 FE design is possible due to Nvidia's new split PCB approach that allows for more efficient airflow and heatsink placement. The brand has used this to pack a dual-fan solution, large fins on top, and ventilation around back. You won't have to worry too much about thermal caveats, but if Founders Edition cards sell out instantly, you'll end up with a custom option at (hopefully) MSRP anyway. On that note, the Asus RTX 5070 Prime OC isn't quite on the same page as the Founder's Edition, as it sticks with a familiar tri-fan setup that prioritizes cooling. Custom 70-class designs, thankfully, aren't as absurd as the Nvidia GeForce RTX 5070 Ti since that card uses the same shroud as many RTX 4090 models. Plus, Asus' take is one of the nicer AIB designs out there, as its curves and subtle black body help prevent the card from looking too industrial. If you're waiting to hear if the RTX 5070 ditches Nvidia's new 12-pin power connector, I've got bad news. You'll still have to use the dreaded dongle to hook the GPU up to your PSU, but it at least scales things back to two 8-pin connections, which should create less cable bulk inside your case. The Founders Edition also uses an angled socket to steer wires out the way, but the Asus Prime OC does still have everything firing outwards. Again, it'd be nice to see custom brands embrace some of this generation's more helpful design queues, but that'll perhaps come with time. DLSS 4 pretty much sits at the core of the entire RTX 50-series range, but the feature arguably does its heaviest lifting in the mid-range ring. In case it wasn't already obvious, the Nvidia GeForce RTX 5070 embodies the idea of using tools like Super Sampling and Multi-Frame Generation by default, and it's the reason why Nvidia felt it could get up on stage at CES and say the card packs "RTX 4090 performance". Don't worry, I'll be getting to that bold claim, but I'll outline exactly what DLSS 4 as a package is first. In a nutshell, it's effectively Nvidia's umbrella term for its suite of AI upscaling tools for GeForce GPUs. If you've already got an RTX 20-series card, you'll likely have come across Super Sampling, a setting that dynamically scales resolutions to boost fps. RTX 4000 users can also tap into Frame Generation abilities that, in simple terms, fill in frame rate gaps using AI, with the end goal being faster visuals without a noticeable loss in quality. With every generation, Nvidia seemingly builds upon DLSS with a new series-exclusive feature. I'm not remotely a fan of locking software features to newer generations, but in the case of the RTX 50-series, you're getting enhanced "Multi-Frame Generation" abilities that can make three frames for every single rendered image. Switching the option on means you're theoretically boosting fps by 4x, and that ultimately helps the RTX 5070 outpace even RTX 4090 in terms of AI-enhanced performance. Super Sampling and Frame Generation is anything but infallible, and both can produce artifacting, latency issues, and other visual hiccups in specific scenarios. Nvidia clearly spends a lot of time looking at its formula in an attempt to kick these quirks to the curb, and DLSS 4 makes headway by adopting a new "Transformer" AI model and additional features like Ray Reconstruction for enhanced ray tracing. The result is instant enhancements in compatible games that are going from strength to strength, and that's going to help PC builds hopefully resist the generational winds for longer than just a few years. I get it - the idea of AI creeping into PC gaming more than gives me the ick too. However, DLSS as a toolkit did emerge before ChatGPT and the likes sank their claws into global life. Helping lighten the load for GPUs like the RTX 5070 is beneficial, especially since system requirements keep getting more outrageous by the second, but I get why some are trying to stick to native settings where possible. I don't completely subscribe to the "fake frames" argument making the rounds since I don't think the situation is as clear cut, but I will be keeping a close eye out for caveats tied to the tech as more games with support roll out. The other main draw of buying cards like the RTX 5070 in 2025 is the Nvidia App. The green team's new hub has effectively replaced the old GeForce Experience software included with previous generations, and it focuses more on quick optimization and providing easier access to settings buried in the Control Panel. There's even voltage, power, and temperature sliders to play with in there too, but if you're using a card like the Asus Prime OC, you'll want to go with the brand's own GPU Tweak III software for proper overclocking. The RTX 5070 has a lot on its shoulders thanks to Nvidia. Claims that the $549 GPU can provide the same performance as a $1,500 flagship raised my eyebrows into outer space during CES, and I immediately assumed the involvement of AI was a given. I'm pretty sure the green team is going to receive a bit of backlash post launch, but now that I've benchmarked the card, I'm pretty sure it'll live up to many players' mid-range expectations. Let's start by taking a look at comparative RTX 4090 performance. Across my test suite of games, the RTX 5070 naturally fell short on the native track. In Cyberpunk 2077, you're looking at an average of 48fps using the ultra preset at 4K, whereas the Lovelace flagship can easily hit around 76fps. We're talking about a card with a 250W TDP and a much lower core count versus a 450W monster, so forgive me if I'm not remotely surprised. That native trend continues across all my benchmarked games, so where's the RTX 4090 performance Nvidia promised? You guessed it - the green team's talking exclusively about DLSS enhanced frame rates. For what it's worth, Multi-Frame Generation does help the mid-ranger outpace the flagship, as I was able to hit around 215fps with x4 MFG enabled while the 4090 couldn't quite reach the 200fps mark using x2. Nvidia claims that 80% of users play games on RTX graphics cards with DLSS on. If you're one of those players, you'll effectively be hitting RTX 4090 frame rates using the RTX 5070 across your Steam library. I'm planning to test even more games using the mid-range card throughout the generation, but in my usual go-to benchmark test subjects, the card pulls big moves with ray tracing off. Things are much tighter when you turn fancy lighting on, and the 4090 actually still beat the 5070 by 5fps in that instance, but I ended up with over 10% higher frame rates in Hitman 3 and over 20% in Dragon Age: The Veilguard. Personally, I think comparing the RTX 5070's native performance to the last-gen Nvidia GeForce RTX 4070 Super makes far more sense. Using Cyberpunk 2077 and Hitman 3 as benchmark examples, the Blackwell card manages to provide a modest native uplift that makes the GPU feel less generational at 4K. I'm talking three frames of a difference in CP2077 with 4K Ultra Ray Tracing enabled, but the gap is far wider at 1440p. In CD Projekt Red's RPG, I was able to comfortably hit 104fps, whereas the RTX 4070 Super taps out at 87fps. Hitman ran at 174fps versus 140fps on the Lovelace 70-class card, and since you won't be hitting that 60fps sweet spot at 4K on either card without DLSS assistance, the RTX 5070 makes more sense as a 1440p option at $549. If the last-gen model manages to show up for significantly less, it'll end up being the better value option, but MSRP versions are practically nonexistent right now. Since the RTX 5070 properly hits 4K with DLSS 4 enabled, I want to talk about what that will actually look like on your gaming monitor. For the most part, the visuals look pretty clean, and I couldn't find many instances of AI-related quirks, glitchy textures, or misinterpreted images. On occasion, I'd see some finer details look slightly strange, most notably a bush in the Obsidian RPG Avowed that had a glowing marker during the opening, but even that was momentary. I also saw some signage in Cyberpunk 2077 briefly look a little garbled when speeding down Night City in a commandeered vehicle, which is something that also happened with the more expensive RTX 5080. From my experience, most of the visual hiccups are tied to Super Sampling rather than Frame Generation. I've yet to see any drawbacks to using x4 MFG specifically compared to x2, and Nvidia Reflex does a nifty job of keeping latency issues at bay in shooters like Marvel Rivals. Keep in mind that in most games, you can switch MFG and SS on and off independently from each other, so if you ever do find visual weirdness tied to one of the options, tweaking settings can help. Let's take a moment to talk about the Asus RTX 5070 Prime OC I've also been testing, as its faster clock speeds are somewhat promising. Using just the base factory overclock, I was able to grab a few extra frames here and there, and I was pretty impressed with Hitman 3 running at 302fps with DLSS 4 and MFG on. The three-fan setup means you'll be able to play around more with higher wattages and squeeze every drop out of Nvidia's GPU, and if you don't mind a slightly bulkier card, that could result in some nice performance gains. Onto thermals, and the highest recorded temperature I have for the RTX 5070 Founders Edition is 71°C. That was under full load in Dragon Age: The Veilguard with ray tracing and ultra settings maxed out, so it's safe to say this smaller card can keep itself pretty cool. For the most part, the reference model stayed within the mid-60s, but if that's not icy enough, then the fact the Asus Prime OC variant peaked at 67°C should please you. Keep in mind that's with the factory overclock enabled, so pushing things beyond the norm shouldn't get too toasty. It's not an RTX 4090 alternative, but the Nvidia GeForce RTX 5070 is a solid mid-range graphics card that'll help more players hit 4K for under $600. You'll have to make peace with the fact that DLSS 4 and Multi-Frame Generation are almost mandatory when it comes to UHD performance, but you're still getting pretty nice results for $50 less than the RTX 4070 Super at launch. Nvidia is clearly trying to shift the GPU scene away from solely focusing on native benchmarks. That's exactly why it thinks the RTX 5070 boasts RTX 4090 levels of performance in the first place, and if the entire market ends up embracing AI upscaling as a default setting, marketing graphics cards in that way makes sense. While I'm not seeing any glaring issues with using Mult-Frame Generation right now, there's no guarantee new quirks won't emerge in the future, so I'm hoping releases without DLSS support won't suffer too much over the next two years. AMD isn't leaning quite as hard into AI with the Radeon RX 9070 series from the looks of things, as the company is sticking with standard Frame Generation and actively talking more about native performance. If that mid-range rival manages to pull off higher frame rates without assistance for under $600, Nvidia could find itself in a spot of bother, so we'll see what happens when I put the card through the same benchmarks. Of course, all of that also hinges on whether the RTX 5070 will be available at MSRP. If all the $549 / £539 stock sells out instantly, it could end up being side-lined. Ideally, retailers will have plenty of both Nvidia and AMD's next gen options available throughout the year, as this still truly help keep things competitive within the mid-range market. For a week, I used the RTX 5070 Founders Edition and Asus Prime OC in my main gaming PC. During that time, I used Cyberpunk 2077, Hitman 3, Dragon Age: The Veilguard, Total War: Warhammer 3, and Marvel Rivals to benchmark the GPU and record average frame rates. This enabled me to compare fps against both the RTX 4090 and RTX 4070 Super to assess whether the new model can provide better native and DLSS 4-enhanced results and whether there are any visual caveats tied to the latter.

I'd hate Nvidia's GeForce RTX 5070 if it didn't have DLSS 4

When the RTX 5080 launched in January, it became clear that the fate of Nvidia's RTX 50-series lies in DLSS 4's hands. The graphics card offered a bare minimum performance upgrade over its predecessor, instead relying on the magic of DLSS 4's new Multiple Frame Generation feature to drive frame rates forward in supported games. The $550 GeForce RTX 5070 takes that to the extreme. Nvidia's new graphics card delivers virtually identical performance to the RTX 4070 Super, a year after the 4070 Super launched. It has the same memory configuration as its predecessor. It costs just $50 less. The RTX 5070 is essentially a 4070 Super with DLSS 4. This isn't an upgrade; it's stagnation. We've spent the past week benchmarking Nvidia's cute lil' RTX 5070 Founders Edition in a variety of games and workloads. Here's what you need to know before buying Nvidia's latest GPU. Again, the RTX 5070 delivers nearly identical performance to the 4070 Super. In some games the RTX 5070 is a little faster; in some games, surprisingly, the 4070 Super is a little faster; and in others they're basically in a dead heat. That's profoundly disappointing - but just because the RTX 5070 is an atrocious generational upgrade doesn't mean it's necessarily a bad graphics card. It sails past the 100 frames-per-second mark in many of our gaming benchmarks, even with all graphics settings cranked to the maximum at 1440p resolution. In our most strenuous tests - Black Myth Wukong and Cyberpunk 2077's RT Overdrive mode with path-traced lighting - the RTX 5070 manages to hit the hallowed 60fps mark demanded by PC gaming enthusiasts. The 12GB of GDDR7 memory feels skimpy for a $500+ graphics card in 2025 and may limit performance in future memory-heavy games, however. If you already have a gaming PC you can simply drop the RTX 5070 into, it'll deliver a vastly better gaming experience than the Sony PlayStation 5 Pro for $150 less. That value proposition goes away if you need to build the rest of the rig around it, however. You could definitely play a lot of games at 4K resolution with the RTX 5070, especially in games that support DLSS. But don't buy the RTX 5070 for 4K gaming. This card is built for 1440p gaming. Nvidia gave the RTX 5070 the same memory configuration as the RTX 4070 Super; 12GB of memory over a 192-bit bus, though the memory has been upgraded from GDDR6X to GDDR7. Here's a quote from our RTX 4070 non-Super review in 2023 that's just as applicable to the RTX 5070: "Nvidia's decision to equip the RTX 4070 with a 192-bit bus and 12GB of memory prevent us from being able to recommend it for long-term 4K gaming, especially with memory requirements only rising in modern games." Nvidia failed to respond to this criticism and it's only more urgent two years later. Slapping just 12GB of RAM on a $550 graphics card feels insulting - especially since Intel's $250 Arc B580 offers the same capacity for less than half the price. AMD's challenger, the Radeon RX 9070, comes with 16GB of RAM. Nvidia needs to do better here...but the uninspiring memory configuration should hold up fine for 1440p gaming in most scenarios. I've said it in every RTX 50-series review thus far, and I'll say it again for the RTX 5070: DLSS 4's new Multi Frame Generation feature - which inserts up to three AI-generated frames between every two "traditional" frames, to send frame rates and visual smoothness absolutely soaring - is truly transformative. It can make even a clunky game like Star Wars Outlaws feel as sublime as the legendary Doom 2016, though the overall experience is a bit hard to measure with normal tools. The RTX 5070 is the ideal "vehicle" for MFG in a lot of ways. Since MFG's inserted AI frames don't respond to your inputs, it can add latency compared to running a game at the same frame rate without MFG. If you can get a game's baseline performance to 60fps or so before flipping on MFG, it generally feels fine (and looks stunning) in single-player games. As shown above, the RTX 5070 hits 60fps at 1440p even in our most strenuous benchmarks. That means you can crank the eye candy to the max, flip on MFG, and enjoy a mind-bendingly smooth experience in the 75+ games that support DLSS 4, no additional headaches or hassles required. PCWorld video guru Adam Patrick Murray spent days playing more than 20 DLSS 4 games to get a feel for the new technology. You can see his thoughts in the video above; he used an RTX 5070 Ti for testing, but the same takeaways apply to the 5070 as well. Nvidia won't like this tidbit, but there's another crucial aspect to know about the GeForce RTX 5070's launch: AMD's direct competitor is launching just a day later. The RTX 5070 launches March 5, tomorrow. AMD's $599 Radeon RX 9070 XT and $549 Radeon RX 9070 launch March 6. Independent reviews of the Radeon 9070 series have yet to be published, but AMD outfitted both GPUs with 16GB of GDDR6 memory and says it's targeting "4K gaming at a 1440p price." The company also promises large improvements in ray tracing performance, and finally invested in beefed-up AI accelerators to power FSR 4, which will be available in 30 games at launch. Given the RTX 5070's negligible performance upgrade over the 4070 Super, smart buyers may want to see what AMD is brewing before investing $549 on Nvidia's card. AMD partners have been shipping Radeon RX 9070 stock to retailers for months now so you may actually be able to grab one before the GPUs sell out, too. If you're coming from an RTX 3070 - or anything older or weaker - the RTX 5070 will be a tangible upgrade. You'll feel the leap forward in performance, and the extra memory capacity. Even though the RTX 5070's mere 12GB of memory disappoints in a $500+ graphics card in 2025, the older RTX 3070 only had 8GB, and that can feel awfully tight at 1440p resolution. The RTX 3070 already needs to make visual compromises to meet the memory demands of many modern games. The RTX 5070 doesn't, though the 12GB capacity is nowhere near as future-proof as the 16GB found on virtually all other graphics cards in this price range. The limited capacity means you won't want this GPU for 4K gaming. While the RTX 5070 is one of the worst generational GPU "upgrades" in memory, it's still a very good 1440p graphics card. Performance soars over 100fps in many games even with graphics settings cranked to the max, and crosses 60fps even in our most strenuous tests with ray tracing enabled. That gives the RTX 5070 enough firepower to flip on DLSS 4's jaw-dropping Multi-Frame Gen tech, which unlocks new levels of visual smoothness that must be seen to believed. It's available in 75+ games. If you don't plan on utilizing DLSS 4, well, you're truly missing out. It's great. But the RTX 5070 loses a lot of its luster when it's not churning out AI frames. In non-DLSS games, the RTX 4070 Super has offered identical performance levels for over a year now. Bottom line? Nvidia's GeForce RTX 5070 is a stagnant "upgrade" from a hardware point of view, and it's skimpy with RAM considering the price. But yes, DLSS 4 software gives the RTX 5070 superpowers otherwise unachievable - at least in the games that actively support it. I'm deeply disappointed that Nvidia didn't move the needle in performance or memory capacity, and barely nudged the price down in return. But this is nonetheless a good 1440p graphics card. Given the stagnation, and given that AMD's Radeon RX 9070 launches just a day after the RTX 5070 but with 16GB of memory, I'd strongly suggest waiting for reviews of that graphics card before making a purchase. Weigh all your options before plunking down your $550.

ASUS TUF GeForce RTX 5070 OC review

Introduction and Analysis It is time for a closer look at ASUS's new GeForce RTX 5070 TUF OC Edition 12GB. Along the way, it packs handy features like dual BIOS, 0dB technology, and a sturdy design, which make it ready to adapt to various performance needs. Under the hood, NVIDIA Blackwell architecture drives the GeForce RTX 50 Series GPUs, offering next-level AI power, fifth-generation Tensor Cores for smooth DLSS 4, and fourth-generation Ray Tracing Cores for lifelike lighting effects. Priced around 549 dollars for the base MSRP models, the powerhouse comes with the GB203 GPU, packing 6,144 CUDA cores to handle everything from immersive gaming sessions to demanding AI tasks. It starts with a 2,160 MHz base clock and typically boosts to 2,510 MHz under normal conditions, while a special OC version pushes up to 2,572 MHz with a higher power limiter for extra performance. The 12GB of GDDR7 memory runs at 28 Gbps on a 256-bit interface, ensuring plenty of bandwidth for top-tier gaming and detailed 3D rendering. Right off the bat, it is clear that NVIDIA is aiming to strike the perfect balance between raw power and sensible power consumption. Built on the 5 nm manufacturing process, this card features the GB205-300-A1 variant, showcasing about 31,000 million transistors. The die measures around 263 square millimeters, which indicates a careful design that focuses on maintaining a compact footprint without sacrificing capabilities. Clock speeds, the default core frequency runs at 2165 MHz. Under heavy loads, it boosts up to approximately 2510 MHz, giving users the extra headroom they need during gaming sessions or demanding workstation tasks. The card carries 12 GB of GDDR7 memory connected via a 192-bit bus interface, delivering a noteworthy 672.0 GB per second of bandwidth thanks to its effective memory speed of 28 Gbps. These specs hint at plenty of bandwidth for high-resolution textures, complex real-time effects, and advanced calculations that modern software requires. Looking at estimated performance figures, the GeForce RTX 5070 is capable of pushing up to 160.6 GPixel per second in pixel fill rate and around 30.84 TFLOPS in FP32 throughput. These numbers underscore its potential not just for intense gaming sessions in 4K or VR but also for professional tasks such as 3D modeling, video editing, and machine learning workflows. By focusing on powerful yet energy-conscious design, NVIDIA positions the RTX 5070 as a versatile solution for a broad spectrum of users. SpecificationRTX 5090RTX 5080RTX 5070 TiRTX 5070BlackwellGB202-300 GB203-400GB203-300-A1GB205-300-A1GPU SMs170 (192 Full)84 (84 Full)70 (84 Full)50 (50 SM Full)GPU Cores21,760 10,7528,9606,144Clock Speeds2010/24102300/26172300/24502160/2510Memory Capacity32 GB GDDR716 GB GDDR716 GB GDDR712 GB GDDR7Memory Bus512-bit 256-bit256-bit192-bitMemory Speed28 Gbps30 Gbps28 Gbps28 GbpsBandwidth1,792 GB/s960 GB/s896 GB/s672 GB/sTotal Board Power (TBP)575W360W300W250W Power Interface1x 12V-2x6 (16-Pin)1x 12V-2x6 (16-Pin)1x 12V-2x6 (16-Pin)1x 12VHPWR (16-Pin)Launch Date 2025January 30, 2025January 30, 2025February 20, 2025March 6th, 2025Price$1999$999$749$549 ASUS TUF GeForce RTX 5070 OC 12GB In the world of PC gaming, the 5070 the GPU is packed with 6144 shading units, 192 texture mapping units, and 64 render output pipelines. These core components lay the groundwork for rendering sharp, detailed graphics at high frame rates. Further boosting visual fidelity is the inclusion of 48 RT Cores, which specialize in handling real-time ray tracing. Meanwhile, the 192 Tensor Cores give machine learning tasks a significant performance boost, speeding up processes like AI-based image upscaling and other neural network operations. A key element in the GeForce RTX 5070's design is its cache architecture. Each Streaming Multiprocessor (SM) contains 128 KB of L1 cache, while the entire GPU benefits from a large 40 MB L2 cache. This hierarchical approach to caching helps reduce latency and improves overall efficiency when dealing with complex workloads. On the software front, the GeForce RTX 5070 supports DirectX 12 Ultimate. This ensures it stays compatible with the latest gaming titles and graphical features, such as variable rate shading and advanced lighting techniques. Power considerations are also at the forefront of NVIDIA's design. The RTX 5070 is rated at 250 W, drawing energy through a single 16-pin power connector. Thanks to its dual-slot form factor, it generally fits inside most standard PC cases without too much trouble. A PCIe 5.0 x16 interface supplies the card with the bandwidth it needs, matching the capabilities of the new Blackwell 2.0 architecture. NVIDIA recommends a 600 W power supply, which allows for ample headroom during peak performance moments. ASUS has equipped this card with three DisplayPort 2.1b ports and two HDMI 2.1b ports, which is perfect for anyone running multiple monitors or aiming for resolutions of up to 8K at 120Hz. As it typically draws around 250W, ASUS recommends using a 750W or higher power supply. The cooling fans have a claw texture and circular arc design that increases airflow while remaining quiet, providing a balance between performance and comfort.