Nvidia RTX 5090 vs RTX 4090: A Leap in AI-Powered Gaming Performance

Nvidia GeForce RTX 5090 vs RTX 4090: 5 biggest differences

Nvidia's RTX 5090 is easily the best gaming experience I've ever witnessed -- both in general and at CES 2025. But it's easy to get whipped up in incredible graphical fidelity without asking one question: What is actually different from the last generation RTX 4090? Turns out there are a lot of differences, and it can be easy to get lost in all the numbers of CUDA cores and memory bandwidth. So with hands-on experience with these GPUs, let me tell you about what the key differences are. The RTX 4090 started at $1,599 -- I only say "started" because there was a high likelihood that the actual price you paid ballooned given the demand for this beast. Well, the 5090 is going one step further with a cost of $1,999. To clarify, this is for the Founders Edition GPU only, so be wary that as other companies like Asus and MSI bolt on a ton more cooling techniques and RGB the cost may go even higher. There are plenty of spec differences, but the biggest changes to the Blackwell architecture come in the huge bumps in AI performance potential, the ability to multitask way more graphics tasks and perform unseen levels of ray tracing. Top it all off with the thermal management capability to run at a blistering 575 watts over the 450 W of the RTX 4090, and while this could be a problem for your electric bill, it will deliver the absolute best gaming experience. Taking a look at the table up above, it's admirable that Nvidia is bringing a good amount of its enhanced AI-driven DLSS features to older graphics cards. But make no mistake about it, you're getting the ultimate experience on 5090 alongside the RTX neural features too. The multi-frame generation here means that not only can the trained AI running on your GPU predict the next frame, it can predict the next 3. This drastically improves framerate over RTX 4090, but it's not the only improvement. This is not possible on the RTX 4090, and the end result is massive gains in frames per second by up to 2X. In the RTX 5090, you're getting new Neural Rendering features too -- things that use an on-board AI trained with the game itself to boost fidelity and detail. Behind the scenes of any game, there's a neural network running away to push a few key things. And on top of that, responsiveness has been drastically improved too with Nvidia Reflex 2. By putting some horsepower towards synchronizing the GPU and CPU rendering of a game, Nvidia is able to drastically reduce the latency of all of this AI work. In my time playing with the RTX 5090, I felt literally zero latency between my inputs and what happens on screen. One thing that has become abundantly clear whenever you look at the recommended specs of the latest PC games is that the amount of video memory required to save critical graphics instructions for demanding titles is getting higher. There's a lot to render, and keeping these visual cues in the background ready to show up at a moment's notice is important to how gorgeous these games look. Not only does the VRAM jump up from 24GB to 32GB in the RTX 5090, but it gets a lot faster too with GDDR7 memory. You're getting both more of it and a faster speed at 32 Gbps. So it should come as absolutely no surprise that the newer graphics card is better than the older one. That's the nature of this industry we're in. But the gains we're seeing are not so natural. Double the framerate, vastly improved fidelity and support for the new wave of 4K OLED 240Hz monitors means a new standard in gaming has been reached. Nvidia has put all of its chips on AI driving this industry forward, and with the RTX 5090 giving you neural nets that are capable of crazy detail with unnoticeable latency, I'm inclined to agree with them.

Nvidia RTX 5090 vs. RTX 4090: an early spec comparison

Table of Contents Table of Contents Pricing and availability Specs Performance The RTX 5090 will be the best, but by how much? Nvidia's next-generation kingpin, the RTX 5090, is now official, and it's as mad as the predictions predicted -- at least, according to Nvidia. We haven't had a chance to test this card ourselves yet, but it's looking set to be an absolute monster, even when compared to its last-generation counterpart. The RTX 4090 is no slouch, though. Until now, it was the fastest graphics card the world had ever seen, and it will remain a fantastic gaming GPU for many years to come. So, how do these two behemoth graphics cards compare? Let's take a look. Recommended Videos Pricing and availability The Nvidia RTX 5090 was announced at CES 2025, with availability starting on January 30. Its recommended retail price is an eye-watering $2,000, however -- more like the GPU-shortage pricing for the RTX 4090 than that card's debut price. When that card debuted in October 2022, its launch price was a comparatively affordable $1,600. Get your weekly teardown of the tech behind PC gaming ReSpec Subscribe Check your inbox! Privacy Policy Today you can find Nvidia Founders Edition 4090s for that same price, with third-party options ranging from $1,500 up to $2,000 for those with the most advanced cooling solutions. Nvidia has defended the price, of course, but we won't know for sure how much extra value it offers until we're able to see it more in action. Specs If we went by the specifications alone, the RTX 5090 should hold a commanding lead over its predecessor in just about any scenario, but all those additional cores and faster memory cells come at the cost of power. Nvidia RTX 5090 Nvidia RTX 4090 CUDA Cores 21,760 16,384 RT Cores Unknown quantity, 4th generation 128, 3rd generation Tensor Cores Unknown quantity, 5th generation 512, 4th generation Maximum clock 2.41GHz 2.5GHz Memory size 32GB GDDR7 24GB GDDR6X Memory bus 512-bit 384-bit Memory speed 28Gbps 21Gbps Memory bandwidth 1,792GBps 1,008GBps TBP 575W 450W The RTX 5090 is a big generational uplift in a number of ways. It's got 33% more CUDA cores, which should lead to a strong improvement in general rasterization performance over the 4090, as well as further mitigate the FPS drop caused by demanding visual effects, like ray tracing. I'd love to tell you that the RTX 5090 has a big increase in the number of RT cores, but unfortunately Nvidia's marketing uses a new metric for measuring those. While they are based on a new, fourth-generation design, we've been given a measurement of 318 TFLOPS. That doesn't give us much to go on for now, but we can assume a solid improvement in overall RT performance considering the general increases we've seen from previous Nvidia generations. The same goes for Nvidia's Tensor cores. They're a new fifth-generation design, with an arbitrary 3,352 "AI TOPS" performance rating, which doesn't give us the ability to compare it to the previous generation. They do allow for multi-frame generation capabilities, though, which we'll get to a little later. On the memory front, we can be much more clear about the big boost Nvidia has given its latest GPU. The RTX 5090 sports a new generation of GDDR7 video memory, and 50% more of it, with 32GB. It's much faster, at 28Gbps, delivering a near 80% improvement in overall memory bandwidth. That's massive overkill for gaming, perhaps betraying this card's design as more of a Titan-esque card, than a flagship gaming GPU -- even if it is likely to be the fastest card of its generation. All of that comes at the cost of power, though, as this flagship Nvidia card once again requires far more than its predecessor. Nvidia claims you'll need 575W of power for the GPU alone. That's more power than some modest gaming PCs need in their entirety. Performance Nvidia made some grandiose claims about the performance of all its RTX 50-series cards, and while they may be some technical truth to them, it's not as clear cut as Team Green would have you believe. Until we test these cards ourselves, though, we can't say for sure how they will perform, so we'll take a look at Nvidia's claims and benchmark numbers, and temper our expectations with a dose of skepticism. GeForce RTX 5090 / RTX 4090 Comparison | Cyberpunk 2077 As you can see, there are some scenarios where the RTX 5090 can perform at up to twice the speed of the RTX 4090, delivering incredible frame rates in even the most demanding of games with full ray tracing. Impressive stuff! Those numbers are echoed in similarly AAA games, too, according to Nvidia's charts. There are no FPS numbers, though, and a big caveat at the bottom. There, Nvidia lets us know that in the games that use DLSS in its testing, it used DLSS 3.5 for the RTX 4090 and DLSS 4 frame generation x 4 with the RTX 5090. Multi-frame generation uses AI to generate multiple frames per rendered frame, leading to much greater increases in frames per second, without the traditional performance required to do it from the main GPU. That is likely to lead to some latency issues, even with the RTX 5090's new Nvidia Reflex technology. While that might leave us guessing about the true performance of this new GPU and how it compares to the RTX 4090, there are a few additional hints on this chart about what we might really be looking at. In games that don't support DLSS 4, like Plague Tale: Requiem, the performance difference between the two cards is much less impressive -- more like 50% faster, rather than 100% or more. And in Far Cry 6, with no DLSS used at all, we get what may well the the true performance advantage of the RTX 5090: something much closer to a traditional 30% generational improvement. It would make sense that the RTX 5090 is at its best in RT and DLSS-supporting games, where it can leverage its RT and Tensor cores to aid the GPU. But most games don't have ray tracing, and almost no games support DLSS 4 yet, so it may be sometime before the RTX 5090's true capabilities can be realized. Even then, not everyone likes the look or feel of frame generation, and it can be a downright hindrance in competitive games. The RTX 5090 will be the best, but by how much? The RTX 5090 isn't out yet, so if you want the best graphics card money can buy right now, the RTX 4090 is still it. However, once the 5090 debuts at the end of January, the 5090 is almost certainly going to be the new king of the hill. By how much though? That's still a question to be answered. Nvidia pushed the limits of credulity with some of its marketing terms and style at CES 2025. It was very clear it didn't expect any real competition at this level of graphics card, even from AMD. But the RTX 5090 is not going to be twice as fast as the RTX 4090 outside of very specific scenarios, and will likely be around 30% to 50% faster depending on the game. However, as DLSS performance continues to improve and more games adopt multi-frame generation, its performance advantage over the RTX 4090 (and any other GPU out there) will continue to grow. For now, we'd recommend waiting to see what this new card can do once it's available for third-party testing. It's going to be impressive, but that power draw looms large over it and the price puts it outside the realm of realistic purchases for the vast majority of gamers.

Nvidia GeForce RTX 5090 versus RTX 4090 -- How does the new halo GPU compare with its predecessor?

Nvidia is betting heavily on AI and new features, even moreso than with the 40-series. The Blackwell RTX 50-series GPUs mark the end of the more than two long years of waiting since the RTX 40-series Ada Lovelace GPUs launched in late 2022. Nvidia announced its upcoming GeForce RTX 50-series cards during the CES 2025 keynote, providing the specifications, pricing, and even a preview of performance. Big claims were made, with new technologies like DLSS 4 playing a major role in those claims. As the new halo part, the RTX 5090 takes over from the RTX 4090, boasting more memory, more compute, more features, and more power. It's not yet available, but there's a lot going on that's worth dissecting before cards go on sale. Will the RTX 5090 be one of the best graphics cards when it arrives? If by "best" you mean "fastest" then yes, there's little doubt it will surpass it's predecessor. Will it be twice as fast? Depending on how you want to measure performance, maybe, but that's putting a lot of trust in AI techniques that aren't the same as traditional rendering. Let's dig into the specifications and features that we know about to discuss how the old and new kings of the GPU world stack up. Let's talk raw specs first. The RTX 5090 has 170 Blackwell Streaming Multiprocessors (SMs), compared to 128 SMs on the 4090. That's a 33% increase in GPU cores -- and the number of CUDA cores, tensor cores, RT cores, texture units, etc. is directly tied to the SM counts, so that's basically a 33% increase overall. Clock speeds also play a role, however, and the 4090 has a 2520 MHz boost clock compared to (based on calculations and Nvidia's official specs) a 2407 MHz boost clock. That means for raw compute, the 5090 'only' offers a 27% improvement over the 4090. However, that's assuming no other architectural differences exist, which almost certainly isn't a good assumption. Memory capacity, speed, and bandwidth are all higher with the RTX 5090, thanks to GDDR7 as well as a bigger, beefier chip. The RTX 5090 has 33% more VRAM than the 4090, clocked 33% higher, for a net 78% improvement in raw bandwidth. We don't know the L2 cache size or if there are any other changes that could impact bandwidth, and both of those are important considerations. Still, that's a big increase in raw memory bandwidth. Nvidia is betting big on AI with the RTX 50-series, and that's where we see some of the biggest changes. The RTX 4090 has 661 TFLOPS of FP16 tensor compute (with sparsity), and 1321 TOPS (teraops) of INT8 tensor compute (again with sparsity). That's far more than AMD's RX 7900 XTX that only offers 123 TFLOPS / TOPS of FP16 / INT8 compute (without sparsity). But it still pales in comparison to the RTX 5090. We're not certain on the FP16 figure, but assuming Nvidia follows the same ratios as the prior generation, the RTX 5090 will deliver up to 1676 TFLOPS of tensor FP16 compute, and double that for 3352 TOPS of tensor INT8 compute (both with sparsity). That's a 154% increase (2.54X) in AI computational performance with the new generation. And Nvidia intends to put the AI potential to good use. As we've discussed elsewhere, Nvidia DLSS 4 will leverage the new features in Blackwell to power its AI algorithms. Multi frame generation will "predict the future" and generate up to three additional frames from one rendered (and potentially upscaled) frame. Because it's using frame projection rather than interpolation, the latency penalty shouldn't be all that different from what we've seen already with DLSS 3 frame generation, but the additional frames will make everything look smoother. How does that actually feel? We haven't had a chance to test it ourselves, so we'll withhold any final judgement, but we're quite skeptical. It will probably work decently, but one rendered frame based on user input followed by three AI generated frames with no new user input won't have the same feel as a game where every frame takes any new user input and gets fully rendered. There are other changes coming as well, however, some exclusive to the Blackwell RTX 5090 and others that will work with older RTX cards. RTX Neural Materials appears to use AI compression and learning to reduce memory requirements for the textures and material descriptions used in games by about a third. However, the hardware pipeline needs to be able to use AI alongside the shaders to have this work, so it will be another 50-series exclusive. DLSS Transformer upscaling on the other hand uses a newly trained network built off of AI transformers, rather than the convolutional neural network (CNN) used with earlier DLSS upscaling algorithms. Transformers have been at the heart of the AI revolution, power things like ChatGPT, DALL-E, and other AI content generators. The sample videos Nvidia has shown of old versus new DLSS upscaling look very impressive, and we're eager to try it out in person. What's more, the new DLSS Transformer algorithm apparently runs faster than the older CNN version, and it will be available for all RTX GPUs. Nvidia's own performance preview, where it suggests RTX 5090 can be up to twice as fast as the RTX 4090, also show a couple of games where there's no DLSS 4 or even DLSS 3 in one instance to muddy the waters. Looking at the Far Cry 6 results, it appears the 5090 will offer about 27% more performance than the 4090 in games where the new AI features aren't part of the equation. In A Plague Tale: Requiem, the gap increases to about 43% (yes, I'm counting pixels!). While in the games that use DLSS 4 MFG (versus DLSS 3 FG), Nvidia shows a 2.3X-2.45X improvement. Does that mean the RTX 5090 is or isn't worth the higher price? We think it will largely depend on what you're doing. There will almost certainly be a lot of people and companies that are interested in AI who will jump at the chance to pay $1,999 for an RTX 5090. Those same groups have been buying RTX 4090 cards for the past couple of years. In generative AI testing, the 5090 also showed a massive 2X jump in performance using Flux.dev. But if you're mostly playing games, and you don't love frame generation? It's probably not a bad idea to sit back and wait to see how things develop for a bit. Maybe DLSS 4 in actual use will look and feel great. Or maybe pulling up to 575W of power through the new 16-pin connector will result in Meltgate Part 2. But however you slice it, two grand is a lot of money to spend on a gaming GPU -- and you'll definitely want the rest of your PC to be up to the task, as powering the RTX 5090 and providing a steady stream of game updates will need a very potent PC.

RTX 5090 vs RTX 4090: How does Nvidia's next-gen Blackwell GPU compare?

Now that the RTX 5090 is official, many of you are likely wondering how it compares to the RTX 4090. After all, Nvidia's current gen GPU has been sitting pretty on the GPU throne since 2022, and even top end AMD cards haven't managed to usurp its position. However, its extremely likely that it'll finally lose its crown to the green team's new Blackwell contender, especially since it's packing new AI tricks. Even the phrase RTX 5090 vs RTX 4090 sounds a bit vicious, but Blackwell is a successor rather than a rival. The new Nvidia card is specifically designed to replace the Lovelace flagship, a tradition that has taken place every generation since the Nvidia 256 back in 1999. However, that doesn't mean the fresh GPU will universally better, as while it boasts elevated specs on paper, there's more to this match than brute performance. To properly put things into perspective, I'm going to run you through RTX 5090 vs RTX 4090 in terms of price, specs, and features. That way, you'll know whether you should buy the RTX 5090, consider another model, or even try and grab the last-gen Lovelace card if prices happen to drop in the months ahead. The RTX 5090 is set to arrive on January 30 for $1,999, whereas the Founder's Edition RTX 4090 retails for $1,499. At face value, you're paying $500 more this generation for Nvidia's flagship compared to 2022, but the increase can be attributed to inflationary changes and technical enhancements. Of course, these MSRPs don't apply across the board, and custom RTX 5090 will likely come with a much higher asking price. At the same time, there's a chance RTX 4090 stock will start to receive a price cut, meaning you'll pay even less for the former flagship. Whether or not those discounts are worthwhile fully depends on how the RTX 4070 and RTX 4080 perform in benchmarks, as CEO Jensen Huang says the former boasts 4090 levels of performance for $549. If you're looking for a GPU that'll provide uncompromised performance throughout the generation ahead, the RTX 5090 will do so for $1,999. The same can't be said for the RTX 4090 since the new flagship will boost fps further at 4K, but if you can get one for cheaper than something like the RTX 5080, it could be a winner in terms of price. Again, that hinges on how mid-range Blackwell GPUs perform, as the Lovelace card could be end up worse value. For many, specs will be the main point of comparison between the RTX 5090 and RTX 4090. That makes sense given that the Blackwell GPU will use a shiny new architecture paired with 32GB of faster GDDR7 VRAM and enhanced AI abilities, but it's worth noting that on-paper figures don't always translate performance-wise. I'd advise waiting for my RTX 5090 review if you want a true idea of how the above specs compare to the RTX 4090. I'll be sure to return here with some comparative benchmarks after that too, but even before we get to that point, the Blackwell graphics card is packing serious specs under the hood that will impact frame rates. For starters, the RTX 5090 has 21,760 CUDA cores versus 16,384 on the RTX 4090. Those extra units go a long way in improving rendering performance and generally speeding things up, which should result in a boost in capacities over and above the 4090. RTX 5090 boost clock speeds are technically lower, coming in at 2.41 GHz versus 2.52 GHz with the RTX 4090. That might lead you to believe the new card is somehow slower, but an increased CUDA core count and enhanced architecture give Blackwell an edge regardless. The 32GB GDDR7 VRAM in the RTX 5090 is not only an increase over the 4090's 24GB GDDR6X offering, but the modules are up to 33% faster and 20% more power efficient. The upgrade will help Nvidia's card better deal with heavy rasterization and ray tracing workloads, and should help playing games at 8K become more viable. As for power, the RTX 5090 comes with a 575W TDP, and Nvidia recommends using a 1,000W PSU for your entire system. That's a big leap over the 4090's 450W requirements, so you might end up needing to upgrade your power supply if you go next-gen. AI upscaling is absolutely the secret sauce when it comes to both the RTX 5090 and the RTX 4090, and the former will launch with brand new DLSS 4 tricks. Not only will the toolkit help Blackwell GPUs boost fps even further using new "Multi Frame Generation" abilities, but it'll also use something called "Transformer." Nvidia's Bryan Catanzaro provides a full rundown of DLSS 4 in the video above, but Transformer is the same model used by the likes of ChatGPT, Flux, and Google Gemini. It effectively aids elements like temporal stability and anti-ghosting to provide a more seamless upscaling and Frame Generation experience, meaning you're less likely to notice any visual hiccups that break immersion. During the Nvidia CES 2025 keynote, Jensen Huang also claimed that Blackwell GPUs can generate 33 million pixels using just 2 million pixels worth of data and new AI abilities. The event included a "real-time computer" graphics example that showed off hyper-detailed textures and ray tracing on every pixel. The AI basically predicts visual data that hasn't been rendered yet, in turn increasing performance by lightening GPU workload. Okay, so how does that impact games? Well, Nvidia uses Cyberpunk 2077 in a DLSS 4 demonstration in one of its examples. In a side-by-side clip, the feature boosts 4K frame rates to 245fps versus 142fps using DLSS 3.5. That's the version available to the RTX 4090, so you can count it as a reason to go for a Blackwell card over Lovelace. If you're into shooters, you'll be pleased to hear the RTX 5090 will use Reflex 2 - the latest version of Nvidia's latency-busting tech. It'll come armed with a feature dubbed "frame wrap" that seemingly improves results dynamically using mouse input information and calculating camera positions. Using The Finals for a visual example, the GPU giant claims thew new tech will reduce latency from 27ms using OG Reflex compared to just 15ms with the new version, adding up to an overall 75% reduction compared to native results. On paper, the RTX 5090 beats the RTX 4090 in terms of performance and features, but I'll want to get hands-on with Blackwell before recommending it over Lovelace. Once I've got benchmarks and frame rates in hand, I'll be able outline whether it's worth spending $500 more on the latest model or trying to get the last-gen card for cheaper. It's important to stress that the RTX 5090 and 4090 aren't equivalents, and it's more likely that lower-class cards will outpace the previous flagship. Nvidia says the RTX 5070 will boast 4090 levels of performance, and while that might be thanks to AI upscaling and DLSS, it still means it's more of a rival to the previous frontrunner. Of course, I'll also be looking to see how AMD's eventual RDNA 4 challenger holds up against the RTX 5090. However, seeing as the red team glossed over its graphics options during its own keynote, and has only publicly mentioned a lower-spec Radeon RX 9070 option so far, we might have to wait a little bit.

Here's when it's worth paying $2,000 for the RTX 5090

We waited for it, made memes about it, and ranted about it, and now it's here -- Nvidia has launched the new fastest consumer GPU in the world, the RTX 5090. Unlike the other SKUs, like the RTX 5080, whose MSRP took almost all of us by surprise, the leaks about the RTX 5090 were bang-on. It's a $2,000 monster that is so far ahead of the RTX 5080 that it's ridiculous. This is the first time Nvidia has put a $1,000 price gulf between its 80 and 90 class cards. It goes without saying that most people will be crazy to even consider the RTX 5090 -- $2,000 can get you a beast of a 4K gaming PC, even with the latest RTX 50 series cards. Why would you spend it all on a single graphics card? Well, there might be a handful of scenarios where buying the RTX 5090 would be sensible, maybe even a good deal. Related 5 reasons I don't plan to upgrade to Nvidia's RTX 5000 series It'll be exciting to cover what Nvidia's RTX 5000 series brings in January, but I don't plan to upgrade. Posts 6 You want a phenomenal upgrade over your RTX 4090 Playing king of the hill comes at a cost For most gamers who own a mid-range or budget graphics card, old or new, almost any of Nvidia's 50 series cards can count as a good upgrade. Price-to-performance considerations aside, upgrading from, say, an RTX 3060 or RTX 4070 to the RTX 5070 (which is supposedly as fast as the RTX 4090) will be a massive upgrade. However, for the enthusiast audience that bought the RTX 4090 -- for anywhere between $1,600 to $2,500 -- only the very best can suffice. While I don't really buy the argument that the $549 RTX 5070 is enough to outclass the RTX 4090 -- it's mostly AI frames anyway -- the point is moot when considering RTX 4090 owners. Someone who needs to have the absolute best GPU on their system doesn't get into the nitty-gritty of whether they should upgrade; they feel compelled to. After all, once you experience flagship gaming performance, anything less feels like a downgrade. The RTX 4090 was already miles ahead of the RTX 3090 (and even the 3090 Ti), but the RTX 5090 seems to have unlocked a whole new frontier of performance. You might balk at its price, as you should, but there's no contesting the multiple initial impressions by creators showing how ridiculously fast it is compared to the RTX 4090. For enthusiasts who can afford to buy the Halo product once again, the price of the RTX 5090 is incidental -- the performance, real or AI-generated, is all there is. Related 5 reasons I'm never buying a high-end GPU again and you shouldn't either High-end graphics cards don't justify the premium. I can save hundreds of dollars and still get a superb gaming experience with a mid-range card. Posts You just can't settle for 16GB of VRAM It sucks, but it is what it is Whatever game-changing performance Nvidia may have cooked up with the Blackwell architecture and AI trickery, the same memo didn't reach the VRAM department. The lack of VRAM on Nvidia GPUs is a beaten horse that's flatter than the RTX 4060 Ti's improvement over its predecessor. However, the fact remains that the company refuses to provide sufficient VRAM even on its expensive offerings. The RTX 5080, a $999 graphics card, has the same VRAM as the previous-gen RTX 4080 Super and the $749 RTX 5070 Ti. Yes, the memory is now GDDR7 and the architectural improvements count for a lot, but if you feel that getting 16GB of VRAM after spending a grand is a rip-off, you'll be forced to make a colossal jump to the RTX 5090 and its 32GB of VRAM. If you can't handle the anxiety of running out of VRAM and can afford the Blackwell flagship, splurging on the RTX 5090 can seem like a reluctant but obvious decision. I just hope that the RTX 6090 doesn't make the RTX 5090 buyers regret their decision -- get in touch with your RTX 4090 friends for more details. Related How much VRAM do gamers need? Unfortunately, having enough VRAM on your GPU is a luxury today Posts You need all the AI compute you can get Top of the TOPS charts Close Nvidia made no efforts to conceal its AI push in its CES keynote. Even the slide showing the prices of each of the newly announced SKUs proudly displayed the AI TOPS of each card. More than ever, the company focuses heavily on AI computing and probably wants consumers to do the same. Professionals who want the latest and greatest AI accelerator for their LLMs, deep learning projects, and more without spending thousands more on professional-grade GPUs might seriously consider the RTX 5090. It's a big improvement over the already blazing-fast RTX 4090, with 8GB more VRAM and only a 25% premium. In professional circles, that might even be considered a bargain. Moreover, an RTX 4090 in the current market, for better or worse, already costs upwards of $2,000. Buying the RTX 5090 for AI workloads makes the most sense from a consumer standpoint. Enthusiast gamers will get it just for the kicks, yes, but it isn't a gaming GPU, no matter what Nvidia says. In a way, I welcome the enormous price and performance gulf between the RTX 5090 and RTX 5080. This way, the majority of consumers will consider the RTX 5080 the "flagship gaming GPU," with the RTX 5090 being an anomaly that's insanely fast and insanely priced, only making sense for professionals. Of course, as I mentioned before, gamers who can afford them and need nothing but the best will keep buying top-tier cards every generation, but more than ever before, 80-class cards will be more than enough for high-end gaming. Related Nvidia and AMD are in their AI era, and gamers can do nothing but watch The AI wave has turned into a tsunami for gamers, as Nvidia and AMD turn all their energies to data center GPUs Posts The RTX 5090 costs a bomb, but it's worth it for some With every GPU generation, Nvidia keeps increasing the prices across the board, but the most eye-watering prices are seen on its flagship cards. The RTX 4090 was received as well as you'd expect, and the RTX 5090 is going much the same way. However, for a handful of consumers who demand nothing but the best, can't compromise on any aspect of the card, or need extra performance for professional workloads, the RTX 5090 can be value for money (imagine that).

Nvidia RTX 5090 vs 4090: What are the Upgrades?

After months of rumors and leaks, Nvidia finally launched the Blackwell GPU for consumers. The new Nvidia GeForce RTX 5090 GPU delivers multi-fold improvements over its predecessor, RTX 4090. It packs faster memory, brings DLSS 4, houses a large number of CUDA cores, and much more. So if you want to find out the difference between the two Nvidia GPUs, go through our detailed comparison between Nvidia RTX 5090 and 4090. First off, the Nvidia RTX 5090 is developed on the latest Blackwell architecture whereas the RTX 4090 is developed on the Ada Lovelace architecture. At this point, the Ada architecture is over two years old. In addition, RTX 5090 is manufactured on TSMC's 4nm (N4P) process node, which results in 92 billion transistors on the die. On the other hand, RTX 4090 is fabricated on TSMC's 5nm (4N) process node and packs around 76.3 billion transistors. As for CUDA cores, RTX 5090 is the first Nvidia consumer GPU that features over 20,000 CUDA cores (21,760, to be precise). RTX 4090 houses 16,386 CUDA cores. Here, thanks to the Blackwell architecture and improved 4nm node, RTX 5090 packs 32% more transistors. Overall, Nvidia says RTX 5090 delivers close to 2x better performance than its predecessor, RTX 4090. In AI workloads, the performance difference is nearly 3x. And all of this is possible at a lower frequency which is surprising. The RTX 5090 card has a peak frequency of 2.41GHz, slightly lower than the RTX 4090's 2.52GHz peak clock speed. Moving to performance, Nvidia has disclosed some numbers comparing the RTX 5090 and 4090. The Shader cores on RTX 5090 can deliver up to 125 TFLOPS whereas the RTX 4090 can output up to 83 TFLOPS. Here, RTX 5090 leads by 50%. The 5th-gen Tensor cores on RTX 5090 can perform up to a whopping 3,352 trillion operations per second (TOPS) on AI workloads. The 4th-gen Tensor cores on RTX 4090 can process up to 1,321 AI TOPS, which means RTX 5090 is around 2.5x faster than RTX 4090 on AI tasks. Next, the 4th-gen Ray Tracing cores on RTX 5090 can deliver graphics performance up to 318 TFLOPS. And the 3rd-gen RT cores on RTX 4090 deliver up to 191 TFLOPS. In Ray Tracing performance, the RTX 5090 leads by around 66%. In gaming benchmarks that Nvidia shared, the company says RTX 5090 offers 2x better performance than RTX 4090 in titles like Cyberpunk 2077, Black Myth: Wukong, and Alan Wake 2. In a demo that Nvidia showcased, RTX 5090 with DLSS 4 produced 238 FPS in Cyberpunk 2077. The RTX 4090 with DLSS 3.5 could only reach up to 109 FPS. As for Generative AI workloads, RTX 5090 again doubles the AI performance when running the Flux.dev image generation model locally. Coming to DLSS 4, it brings a groundbreaking Multi Frame Generation feature which is only available on the RTX 50 series cards. However, other improved DLSS 4 features like Frame Generation, Ray Reconstruction, Super Resolution, etc. are available on RTX 40, 30, and 20 series cards. The highlight of DLSS 4 is indeed Multi Frame Generation which uses Transformer-based neural networks to predict up to three frames in advance. So besides raw performance improvements, DLSS 4 with Multi Frame Generation is only exclusive to RTX 50 series cards. You must keep this in mind if you want to upgrade from RTX 4090. The Nvidia RTX 5090 is the first consumer GPU to feature the ultra-fast GDDR7 memory. It packs 32GB of VRAM and operates on a 512-bit memory bus, allowing the memory to reach a crazy bandwidth of 1.8 TBps (1,792 GBps). On the other hand, RTX 4090's memory bandwidth is limited to 1,008 GBps. You get 24GB of GDDR6X memory on RTX 4090, which relies on 384-bit memory bus. As for power consumption, the Nvidia RTX 5090 GPU has a TDP of 575W which is 125W more than the RTX 4090's 450W TDP. Nvidia recommends a PSU of up to 1000W for the RTX 5090. That said, many expected that the RTX 5090 would be much larger, but it shares the same dimensions as the RTX 4090. And unlike the RTX 4090 which was a 3-slot card, the Founder Edition RTX 5090 is a two-slot GPU which means you can fit it into a smaller case. The RTX 4090 GPU debuted at $1,599 and the RTX 5090 is priced at $1,999. Many were expecting that Nvidia would price the Blackwell GPU around $2,500, but the company chose a reasonable price. With a $400 increase in price, you get a much faster GPU that delivers strong performance on all fronts, be it gaming, AI workloads, or creative rendering. In addition, with RTX 5090, you get DLSS 4 with the Multi Frame Generation feature and have access to 32GB of ultra-fast GDDR7 memory. Of course, we need to wait for in-depth benchmark testing, but the Nvidia RTX 5090 GPU looks like a powerhouse. If you already have the RTX 4090, you may not want to upgrade instantly, but if you are using an RTX 30 series card, the RTX 5090 is an easy recommendation.

RTX 5070 vs RTX 4090: Is Nvidia's Bold Claim True?

The Ray Tracing cores on RTX 4090 can deliver up to 191 TFLOPS whereas the RTX 5070 can output up to 94 TFLOPS. At CES 2025, Nvidia boldly claimed that the affordable new RTX 5070 GPU delivers RTX 4090 levels of performance. In case you are unaware, RTX 5070 is priced at one-third of what RTX 4090 costs. So the claim by Nvidia CEO Jensen Huang is pretty bonkers. In order to dissect the claim, we have compared the Nvidia RTX 5070 with RTX 4090 to draw a definitive conclusion. On that note, let's begin. Starting with the architecture, the new Nvidia RTX 5070 GPU is based on the cutting-edge "Blackwell" architecture, the same as the RTX 5090. The older RTX 4090 GPU is developed on the "Ada Lovelace" architecture which is over two years old now. RTX 5070 is manufactured on TMSC's improved 4nm (N4P) process node whereas RTX 4090 is fabricated on TSMC's 5nm (4N) node. Currently, there is no information on RTX 5070's transistor density, but RTX 4090 packs close to 76.3 billion transistors. So far, RTX 5070 looks more advanced on paper, but as we move towards the raw hardware numbers, things start to change. In terms of streaming multiprocessors (SMs), RTX 5070 has only 48 SMs whereas RTX 4090 packs 128 SMs. Next, RTX 5070 houses 6,144 CUDA cores while the beefy RTX 4090 has 16,384 such cores, almost 3x more than 5070. RTX 5070 comes with 192 5th-gen Tensor cores, but RTX 4090 packs 512 4th-gen Tensor cores. Finally, RTX 5070 has 48 Ray Tracing cores while RTX 4090 gets 128 RT cores. It's clear that in terms of raw horsepower, the older RTX 4090 GPU is 2x-3x more capable than the RTX 5070. During the presentation at CES 2025, Nvidia CEO Jensen Huang kept repeating "impossible without AI". So with AI-powered DLSS 4 features, Nvidia is claiming that RTX 5070 can match the performance of flagship GPUs like the RTX 4090. However, that's all AI-driven rendering. In terms of raw specs, the RTX 5070 doesn't come close to matching the RTX 4090. Let's now look at the performance numbers to understand the difference between RTX 5070 and RTX 4090. The Shader cores on RTX 5070 can deliver 31 TFLOPS of performance at FP32 precision. The RTX 4090 can output up to 83 TFLOPS, again nearly three times more than the RTX 5070. In terms of AI operations, the 5th-gen Tensor cores on RTX 5070 can perform up to 988 TOPS whereas the 4th-gen Tensor cores on RTX 4090 can process 1,321 TOPS. Here, the credit must be given to the new 5th-gen Tensor cores on RTX 5070 as it uses only 192 cores (vs 512 on RTX 4090) to come within the striking distance of RTX 4090. However, in Ray Tracing performance, RTX 4090 again takes a substantial lead by delivering 191 TFLOPS whereas RTX 5070 does only 94 TFLOPS. The numbers speak for themselves, but Nvidia is relying heavily on AI-based optimizations to claim that RTX 5070 matches RTX 4090's performance. Indeed, Nvidia has introduced DLSS 4 this time on all RTX 50 series cards. It brings a groundbreaking Multi Frame Generation feature that "predicts the future" by generating three additional frames and multiplying frame rates by 8x over traditional rendering. It uses a special AI model, powered by the 5th-gen Tensor cores on RTX 50 series GPUs, to generate additional frames. Apart from that, DLSS 4 replaced a CNN-based model with a Transformer-based neural network for improved upscaling/Super Resolution and Ray Reconstruction. And this feature is also coming to older RTX 40, 30, and 20 series cards. While Nvidia has not disclosed benchmark numbers comparing the RTX 5070 with the RTX 4090, it seems with all the AI-powered features available with DLSS 4, the RTX 5070 may come closer to RTX 4090, in terms of frame rate and resolution improvements. Of course, we will wait for an independent review before drawing a conclusive judgment. Now, let's talk about memory bandwidth. Nvidia RTX 5070 operates on a 192-bit memory bus and features 12GB of ultra-fast GDDR7 VRAM. It clocks a memory bandwidth of 672 GBps. On the other hand, RTX 4090 comes with 24GB of GDDR6X memory, and operates on a 384-bit memory bus, achieving a bandwidth of 1,008 GBps. Put simply, RTX 4090 has almost 2x faster memory, compared to RTX 5070. And having just 12GB of memory may become a bottleneck in the future for high-resolution games. In terms of power consumption, the RTX 5070 is indeed more power-efficient. It has a TDP of 250W whereas the RTX 4090 can draw power up to 450W. Since RTX 5070 has much lower cores across various units, it uses less power, compared to RTX 4090. The Nvidia GeForce RTX 5070 has an affordable price tag of $549 while the flagship RTX 4090 costs $1,599. After all, the RTX 4090 brings 2 to 3 times more SMs, CUDA, Tensor, and Ray Tracing cores, delivering significantly higher performance. It also has access to a larger and faster pool of memory. To sum up, there is simply no comparison between the RTX 5070 and RTX 4090, as far as the raw horsepower is concerned. That said, as Nvidia claims, with AI-powered DLSS 4 features such as Multi Frame Generation, and Transformer-based upscaling/ray reconstruction, you might be able to enjoy some games, comparable to that of an RTX 4090 GPU. But keep in mind that game developers must adopt Nvidia's new AI features in order to leverage the full potential of DLSS 4. All in all, gamers should wait for in-depth gaming benchmarks before jumping the gun.

I just played games on Nvidia RTX 5090, and the hype is real -- here's why

Follow our CES 2025 live blog for all the biggest tech and gadget news straight from Las Vegas. And be sure to follow Tom's Guide on TikTok for the coolest videos live from the show. The Nvidia GeForce RTX 5090 is a massive GPU that just got announced at CES 2025, and it rightfully deserves its size and stature by bringing monstrous gameplay performance and AI power to usher in the next generation of gaming graphics -- something that I got to witness for myself. In my hands-on, I'll go into detail about what you get with this $1,999 graphics card that launches on January 30, alongside what software capabilities (many of which are AI-driven) the new hardware unlocks, and exactly what this means in gameplay. Put simply, I'm about to show you why Nvidia's RTX 50-series GPUs were an easy choice to win our best of CES 2025 award! Let's break down the RTX 5090 -- easily the most powerful gaming graphics card you can buy today with effectively double the performance of the outgoing 4090. Inside, everything has been upgraded thanks to Nvidia's new Blackwell Architecture: Combine all this with 32GB of GDDR7 video memory running at a ridiculous total bandwidth of 1,792 GB/sec, alongside a whopping 3,352 TOPS of AI power and 21,760 CUDA Cores (a core designed specifically to run parallel computing tasks effectively), RTX 5090 aims to hit that new 4K 240 frames per second standard found in gaming monitors like the new Asus ROG Swift OLED PG27UCDM. And after playing two incredibly demanding games, I can confirm that Nvidia isn't mucking about here. The claims are real and my mind has been truly blown. Black Myth: Wukong is a truly gorgeous-looking game that can be a pain for any GPU to render because of the massive demands for on-board graphics card memory. This is no trouble for RTX 5090, and as you can see, I was comfortably getting well over 200 FPS with DLSS on and everything turned up to max settings in 4K. This was the only demo that had a frame counter present. From the visual fidelity to the non-existent latency that comes with using DLSS, this is such a step up for PC gaming. And speaking of DLSS -- to those who are out of the loop, Deep Learning Super Sampling essentially uses AI to interpret and make predictions on what is going to happen in the next frame of gameplay based on what is happening in the current frame. That ensures vastly smoother gameplay and faster frame rates, but sometimes with the current version of DLSS 3.5, you are seeing that predictive loop of the next frame make a couple errors. The problem is called ghosting, where you may see the trace of a fast moving object stick around for a few frames more than it should. With DLSS 4, not only does it generate the next frame, it generates multiple frames ahead of time using a new AI model, which uses 30% less video memory. And on top of that, I noticed zero ghosting or artifacting. Meanwhile, this is the first time I've played Black State -- a 3rd person shooter from developer Motion Blur that tells the story of an agent that is saving the world from a cataclysmic event. Once again, this was a demo using RTX 5090, 4K max settings with DLSS 4 multi frame generation (15 out of every 16 pixels is generated entirely using this AI technique). And on top of that, the next generation Nvidia Reflex to drastically reduce latency. Pair that with the photorealistic style that the developers are going for with the visual flair, and there's no two ways around it: this is the best-looking game I've ever seen. With no frame counter on-screen, I can only go off the general vibe I'm getting off the smoothness. I'm quietly confident you're getting over 120 FPS, and with all the ray tracing, path tracing and neural rendering of said light tracing at play here, this is a jaw dropping experience. Plus, that latency feels pretty much non-existent, which contributes to those super addictive 3rd person shooter gameplay mechanics (including bullet time) with satisfyingly chunky weapons -- including a plasma gun that can literally explode your foes. Using slow motion during one of these plasma bullet impacts is a wild ride: a pinata of blood and guts that swells and fades in reaction to your shot before blowing up. But every single pixel of this is seemingly ray traced for quite the immersive light show. Nvidia RTX 5090 is truly a game changer (pardon the pun) -- a beasty card with even beastier performance gains that are going to set the stage for the next generation of graphical fidelity. CES 2025 has been a huge show for gaming, including huge improvements in integrated graphics courtesy of the AMD Ryzen AI Max chipset that is powering RTX 4060-level framerates and fidelity in the ROG Flow Z13. But Nvidia didn't just want to prove that dedicated graphics are still huge, team green wanted to completely annihilate the competition. The $1,999 RTX 5090 absolutely does this, and I cannot wait to play more!

Nvidia unveils new GeForce RTX 5090, RTX 5080, RTX 5070 Ti, and RTX 5070 graphics cards at CES 2025

After months of speculation and anticipation, Nvidia finally lifted the cover off its latest lineup of consumer graphics cards, the Nvidia GeForce RTX 5000 series, starting with the flagship RTX 5090. Nvidia CEO Jensen Huang made the announcement at a packed arena at Las Vegas's Mandalay Bay resort and casino, a headline event that topped off a day of major reveals from rivals AMD and Intel at CES 2025. However, the Nvidia RTX 5000 series graphics cards were always going to steal the show, no matter what anyone else announced, so Huang naturally had the limelight on Monday night as he unveiled our first definitive look at the RTX 5090 - amongst other new GPUs and fancy AI features. It might not be called the Nvidia Titan RTX, but the RTX 5090 might as well be, given the specs on offer and its downright scandalous MSRP of $1,999 (£1,999 / AU$4,039) - not a generational price bump from the RTX 4090, granted, but still a ludicrous amount of money for an ostensibly 'consumer' graphics card. With an astounding 92 billion transistors, next-gen Tensor Cores and Ray Tracing Cores, and more than double the AI processing speed of the 4090, the RTX 5090 will unquestionably be the most powerful consumer graphics card on the planet, and it won't even be close. Pair the GPU specs with 32GB of shiny new GDDR7 VRAM on a massive 512-bit memory bus and PCIe 5.0 x16 interface, it has an astonishing 1792 GB/s of memory bandwidth, with a memory speed upwards of 23.8 Gbps. Given its specs, not only will this graphics card absolutely blow through native 4K gaming (without upscaling) at the highest settings (including ray tracing), it's arguably the first real 8K graphics card given the amount of VRAM it has and its memory bandwidth, two key specs that allow a graphics card to process the substantially larger 8K texture files needed for gaming at that resolution. Of course, few games even support 8K resolution, much less have developers and artists effectively wasting their time on texture files so large that only a rare few will ever see them as intended. But there's no doubt that if 8K gaming ever becomes a thing, the RTX 5090 will be more than ready to meet the challenge. Of course, that doesn't really address the fact that this is no longer a gaming GPU -- not anymore, and not at this price. And if (well, when) scalpers get involved, it's going to be far worse. After all, we called the $1,199 price tag on the flagship RTX 2080 Ti 'almost obscene' in our review three GPU generations ago. With no generational price drop from the already wildly expensive RTX 4090, it's not hard to argue that the RTX 5090 is purely a professional workstation GPU, meant to process raw 4K video streams or render lengthy 3D generated sequences at Pixar or some other animation studio. As fun as it might be, this is not a graphics card meant to play Wolfenstein 3D. The Nvidia GeForce RTX 5090 is set to go on sale for $1,999 on January 30. In a move that has been telegraphed for a while, the Nvidia GeForce RTX 5080 will also be part of the first wave of Nvidia's new graphics cards to hit store shelves later this month. The RTX 5080 looks to be almost exactly half of the RTX 5090 in terms of specs. Although Nvidia has yet to release a comprehensive spec sheet for the new GPUs - a strange move, though likely one designed to avoid distracting from all the new AI features - we know that the 5080 has 1,801 AI TOPS (trillion operations per second), a little under half the RTX 5090's 3,352. It also has new GDDR7 VRAM as well, with a pool of 16GB on a 256-bit memory bus for 960 GB/s of memory bandwidth - again, basically half the specs of the 5090's VRAM. Its memory speed is a blazing fast 30 Gbps, which helps make up for the narrower memory bus. The card will go on January 21st with an MSRP of $999 (expected £999 / AU$2,019), which again matches the launch price of the previous-gen RTX 4080. The Nvidia RTX 5070 and 5070 Ti also debuted on Monday night, likely an effort from Nvidia to reassure gamers that they haven't forgotten about the midrange market - especially with its main rival AMD refocusing to target the budget and midrange space exclusively with its new GPUs. Again, we don't have the breadth of specs we'd hoped to see at this point, but we do know that the RTX 5070 Ti offers 1,406 AI TOPS while the RTX 5070 has 988 TOPS. As a rough point of comparison, Apple's new M4 chip caps out at around 38 TOPS - so a dedicated GPU is arguably still a necessity for serious AI workloads. Over on the VRAM front, the RTX 5070 Ti's memory profile is nearly identical to the RTX 5080, with 16GB GDDR7, a memory speed of 28 Gbps on a 256-bit bus, and a memory bandwidth of 896 GB/s, making it more than ready for 4K gaming. Meanwhile, the RTX 5070 has 12GB of GDDR7 at 672 GB/sec - still faster than the RTX 4070, though the same base amount of VRAM. The Nvidia GeForce RTX 5070 and 5070 Ti will go on sale in February - date to be confirmed - with respective retail prices of $549 (£549 / AU$1,509) and $749 (£749 / AU$1,109). This is actually quite pleasing to see, since the RTX 4070 retailed at $599 - meaning we finally have a generational price drop from Nvidia. In addition to the new hardware, Nvidia also showcased a selection of upgraded AI features debuting with the 'Blackwell' RTX 5000 generation of graphics cards. Chief among these, of course, was DLSS 4 - the latest update to Nvidia's resolution upscaling software, which allows for better framerates in-game by rendering the game at a lower resolution and upscaling it to a target resolution (say, 1080p to 4K) using AI. DLSS 4 will also feature 'Multi Frame Generation', an improved version of the Frame Generation tech seen in the RTX 4000 generation, which uses AI to extrapolate and produce additional frames and 'insert' them between ordinary rendered frames to boost framerate. Unfortunately for users on older GPUs, only the regular DLSS 4 upscaling will be available on older cards; Multi Frame-Gen will be exclusive to RTX 5000 cards. We're also getting Nvidia Reflex 2, a new version of the Reflex software for reducing input latency in games. Reflex 2 will feature 'Frame Warp', which aims to proactively insert generated frames by reading mouse input before it even reaches the display - this can reportedly reduce input latency by as much as 75%. RTX 5000 is also bringing AI powers to shader tech with new RTX Neural Shaders. This uses small AI networks in the GPU's programmable shader units to deliver 'film-quality' shading and lighting in-game. 'RTX Neural Faces', along with new RTX tech for hair and skin rendering and animation, promises to deliver more realistic humans than ever before. Lastly, we can expect to see more of Nvidia ACE with this generation of RTX GPUs - the improved AI capabilities of the Blackwell generation mean that projects like Nvidia's (slightly creepy) AI NPC tech can be implemented on a wider level, with ACE-powered characters planned to appear in a handful of titles including PUBG: Battlegrounds and InZOI.

Nvidia's GeForce RTX 50 series promises RTX 4090 performance for $549

Just when you thought Nvidia couldn't push its GPUs any further, we've finally got our eyes on what's next for PC gamers. Whether you're gaming on a laptop or desktop system, you'll soon have access to the RTX 50 series of graphics cards. The star of the CES show is the Nvidia GeForce RTX 5090, replacing the RTX 4090 as the new king of gaming GPUs with an all-new architecture and more powerful AI processing. Related 11 best games to play on your RTX 5000 series Whether you go for the flagship RTX 5090 or the midrange, these are the best games to show off the power of your new GPU Posts 3 What's new in the RTX 50 series Close Blackwell is the GPU architecture powering RTX 50 series GPUs like the RTX 5090 and RTX 5080. GDDR7 memory is the most apparent change from the previous generation GPUs with a bandwidth of up to 36 Gb/s. Compared to last-gen cards, we're looking at considerable gains in throughput, which is fantastic for handling higher loads on the GPU, such as ray tracing. RTX Blackwell will offer up to 4,000 AI TOPS, 380 RT TFLOPS, 125 Shader TFLOPS, and more with the RTX 5090. Deep Learning Super Sampling (DLSS) is a technology Nvidia has been perfecting since its debut in 2018 and with RTX 50 we're now up to DLSS 4. The sheer performance bump offered by Blackwell GPUs will allow technologies such as DLSS to perform better when used in conjunction with ray tracing and other more advanced graphics. As well as rendering more frames, the latest DLSS version will also attempt to create new frames before the GPU even requires them. It's essentially working the AI to predict the future and handle much of the work DLSS would have done, generating up to three frames per rendered frame. Software will play a large part here and Nvidia will likely work to continue improving DLSS and its AI tech to better utilize the improved hardware. 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 Which GPUs are launching at CES? Nvidia is following suit with the launch of the GeForce RTX 5090, showcasing just how powerful this generation of Team Green graphics processing will be. When paired with the latest processors from AMD and Intel, these GPUs will be able to stretch their wings and offer some exceptional gaming experiences at higher resolutions and fidelity settings. Price CUDA Cores TFLOPS GDDR7 Power Req Nvidia GeForce RTX 5090 $1,999 21,760 318 32GB 575W Nvidia GeForce RTX 5080 $999 10,752 171 16GB 360W Nvidia GeForce RTX 5070 Ti $749 8,960 133 16GB 300W Nvidia GeForce RTX 5070 $549 6,144 94 12GB 250W The Nvidia GeForce RTX 5090 has an impressive 21,760 CUDA cores and 32 GB of RAM. To put this into context, the RTX 4090, which was already powerful enough as it is, has 16,384 CUDA cores and 24 GB of RAM. This beast will have a TDP of 575 watts, which is more than some desktop PCs. Then there's the RTX 5080 with 16 GB of GDDR7 RAM, a slightly lower price tag, and 360 watt power requirement. Nvidia is relying on AI to push performance further and reduce prices slightly by allowing the RTX 5070 to perform as well as an RTX 4090, according to Nvidia's data. The 5070 Ti, 5080, and 5090 will perform even better, making the 4090 seem like child's play with these new advancements. It's almost bittersweet to see better performance at the same price point. Close A drawback for many will inevitably be the pricing. Nvidia has priced its higher-end GPUs out of reach for many gamers with the RTX 30 and RTX 40 series and this trend is only set to continue as the company enjoys heightened profits and increased demand for gaming and AI-heavy work. These are some impressive graphics cards, but AMD and Intel offer more reasonably priced GPUs with excellent 1080p and 1440p performance. Having the RTX 5070 available for $549 with RTX 4090 performance does make things more interesting, especially as we wait for RDNA 4 cards from AMD. If you are considering the upgrade, you may need to start considering a new power supply. Oh, and Nvidia hasn't forgotten about mobile GPUs either. Availability starts this month for the new Blackwell GPUs and we look forward to putting Nvidia's performance claims to the test.

Nvidia's GeForce RTX 5070 at $549 -- How does it stack up to the previous generation RTX 4070?

Same price, different architectures, and some big tech changes. Nvidia made a big splash with the official announcement of its upcoming GeForce RTX 50-series Blackwell GPUs during the CES 2025 keynote. And while the halo RTX 5090 certainly looks like an absolute monster, for a lot of people, it's the mainstream (-ish) RTX 5070 at $549 that will be the star of the show. The RTX 4070 has been one of the best graphics cards since it launched, and now its replacement is on the way. Nvidia claims the 5070 will offer "RTX 4090" levels of performance, at about one third the price and a bit over half the power. But how do they really stack up, and how does the 5070 compare to the existing RTX 4070? Let's find out, and we've filled in a few bits and pieces with best guess estimates for now, but most of the specifications are correct. First, let's be perfectly clear: The idea that the RTX 5070 will match the RTX 4090 in all workloads looks like some very rose-tinted glasses. It's obvious that Nvidia is going big on AI with Blackwell, and it's counting on DLSS 4 and other neural rendering techniques to make up the difference. But raw specs still matter in a lot of existing games -- barring a driver-side solution that enables higher performance without requiring patches and updates. The RTX 5070 will have 48 SMs compared to the 46 SMs on the 4070. That's not a very big change at all, and it's a far cry from the 128 SMs in the 4090. The overall FP32 graphics compute works out to 31 TFLOPS for the 5070, 29 TFLOPS on the 4070, and 83 TFLOPS for the 4090. It's extremely hard to believe that, in general, the 5070 will come anywhere near the 4090 in performance without leveraging DLSS 4 and related technologies. There's also the VRAM to consider. The 4090 has 24GB, compared to half that amount on the 4070 and 5070. There aren't too many games where 12GB is insufficient, but Indiana Jones and the Great Circle, with full RT and without upscaling, definitely exceeds 12GB at 4K. More games are likely coming that could push beyond 12GB of VRAM use at higher resolutions and settings. But this is where "RTX Neural Material" could come into play. That seems to be the enablement of Neural Texture Compression, something Nvidia discussed back in 2023, fully implemented in a game. Will it work with any game? Probably not, but we'd be pleased to see a driver-side solution that makes such a feature a reality. Without NTC or RTX Neural Materials, the 12GB will definitely keep the 5070 from matching a 4090. There's also bandwidth to consider. RTX 4090 has 21 Gbps GDDR6X on a 384-bit interface, compared to the 5070's 28~32 Gbps GDDR7 on a 192-bit interface. We know the 4090 has 1008 GB/s of bandwidth. The 5070 should end up with somewhere between 672 GB/s and 768 GB/s of bandwidth. Again, without NTC or neural materials, it's not going to keep up at higher resolutions. AI workloads like LLMs also like having lots of VRAM capacity. Quantization only gets you so far, and neural compression of LLMs isn't a thing (as far as we're aware). The RTX 4090 with 24GB of VRAM can simply load larger LLMs than the 5070, which will only match the 4070 in terms of AI model sizes. It's a different story when we look at AI computational performance. We know the RTX 50-series will have FP4 number format support, but just as important, it seems to have twice the compute per tensor core as the RTX 40-series. That's not enough compute for the 5070 to surpass the 4090, but it's 'only' about 25% slower in theoretical performance. And if something can leverage FP4 on the 5070 where the 4090 needs to use FP8, then it might run better on the 5070. But even the INT8 TOPS favors the 4090. The real kicker is of course the pricing. There are a lot of gamers that simply can't afford a $1,599 graphics card -- never mind the scarcity induced $2,000+ prices we're currently seeing on the 4090. A $549 GPU, even if it's slower in most games, is another matter entirely. Nvidia's xx70-class GPUs have traditionally been the sweet spot for mainstream gamers, and the 5070 looks like it will continue that pattern. Even if it doesn't beat the 4090, if it can consistently deliver performance close to the level of the RTX 4080, it should end up being extremely successful. But really, it all comes down to AI features and DLSS 4. We haven't tried multi-frame generation yet, and after our experiences with DLSS 3 frame generation, we're skeptical at best. It sounds like it will generate up to three frames between two rendered frames, which seems like it would only increase the latency. But DLSS 4 also appears to generate those frames more quickly, meaning the net result would be no worse latency than DLSS 3 framegen, just with additional frames. Far more promising than multi-frame generation, in our view, are the enhancements and upgrades to DLSS upscaling and ray reconstruction. Until now, DLSS has used a CNN (Convolutional Neural Network) for the AI training and inference. Now there's a new transformer-based model, which can apparently be utilized on any existing DLSS 2/3 games. Transformer models have revolutionized many areas of AI development, and the sample sequences in the above video showing CNN vs transformer DLSS look extremely promising. Nvidia has been claiming "better than native" rendering from DLSS for a while now, but the DLSS transformer model may finally deliver on those claims. If it does, that could be the killer feature that makes the 50-series worth the price of admission. Except, the transformer model also works on existing GPUs, so maybe not. As we've noted in the past, while the RTX GPUs promised ray tracing as a new technology, over time it's really been the AI features that have come to the front as the most important aspect of the RTX series. With the RTX 50-series, Nvidia yet again doubles down on AI, and the supporting DLSS software continues to outpace the RT aspect. Whether or not multi-frame generation proves to be a killer feature, if you don't already have a 40-series GPU, the 50-series including the RTX 5070 could entice you to upgrade.

Is the new RTX 5070 really as fast as Nvidia's previous flagship RTX 4090 GPU? Turns out the answer is yes. Kinda.

The Nvidia GeForce RTX 5070 was only fully revealed this week at CES 2025, but already the GPU is causing quite a stir. Though arguably the least powerful of the just unveiled 50-series, the RTX 5070 has been the subject of a number of headline-worthy claims. Alas, most folks will have to wait until the 50-series launches this February to see if it's worth the hype -- but tech journalists aren't most folks. Case in point, PCGamesN has already seen the RTX 5070 in action and has since shared its findings. So, what's the truth behind Nvidia's CEO Jen-Hsun Huang on-stage claim that the RTX 5070 will offer "RTX 4090 performance at $549"? Turns out, there's a lot to that -- but not without caveats. First, the good news: while playing Marvel Rivals, PCGamesN journalist Ben Hardwidge found that the RTX 5070 doesn't just offer comparable performance, but actually eclipses the RTX 4090, what was once Nvidia's flagship card. However, the not-really-bad-news, but more I'm-not-sure-how-to-feel-about-this-news is that we have the GPU's new Tensor cores and their support of Nvidia's just announced DLSS 4 Multi Frame Generation to thank. It's hardly a surprising move. Nvidia has been calling time on raster graphics rendering for years, with Huang himself memorably saying back in September 2024, "We can't do computer graphics anymore without artificial intelligence. We compute one pixel, we infer the other 32. I mean, it's incredible." To be clear though, the name of DLSS 4's game is most impressively frame generation. With regards to the RTX 5070, DLSS 4 uses AI and in-game data to generate up to three full frames between more traditionally rendered frames. To put it a little crudely, DLSS 4 is what's giving you that extra bang for your buck. At CES 2025, Nvidia set up two machines for trade show guests to compare -- one kitted out with the RTX 5070 and multi-frame generation, and one using the older RTX 4090 with standard frame generation. Besides that major difference, both machines were running Marvel Rivals at 4K with the highest amount of DLSS supported by their respective graphics card. The result? Hardwridge writes that the RTX 5070 consistently outperformed the older card, outputting in the region of 240 fps compared to the RTX 4090's average of 180 fps. All of that said key tech specs, such as which CPU was used for these comparison builds, remain unclear, so it's hard to say conclusively just how replicable those numbers are. It also perhaps goes without saying but the older card doesn't support DLSS 4's multi frame gen, so one could quibble about how valuable a comparison this truly is. Furthermore, even Nvidia stressed that this comparison point is a bit of an outlier, as Marvel Rivals appears to be particularly well-optimised for the RTX 5070; when tested on titles that aren't Marvel Rivals, there's apparently a much smaller performance gap between the two cards. Still, that suggests performance that once cost over a grand and a half for the RTX 4090 is now within reach for much less. Provided DLSS 4 is involved, of course.

Nvidia GeForce RTX 5090: Everything You Need to Know

The RTX 5090 GPU is based on the Blackwell architecture and features 21,760 CUDA cores. After a long wait, Nvidia finally launched the GeForce RTX 5090 GPU for consumers at CES 2025. It's the most powerful consumer GPU in the market, developed on the latest "Blackwell" architecture. Be it high-end gaming or heavy AI workloads, RTX 5090 can breeze through all. So if you want to learn more about the new Nvidia GeForce RTX 5090 GPU in detail, go through our explainer below. Nvidia announced its latest GPU microarchitecture called "Blackwell" for data centers at the GTC in March 2024. And nearly a year later, Nvidia finally launched the Blackwell-based GeForce RTX 5090 GPU for consumers. It's the first RTX 50 series GPU based on the Blackwell architecture. The RTX 4090 GPU was based on the Ada Lovelace architecture which was announced in 2022. So after over two years, Nvidia has released the successor of RTX 4090 with the Blackwell architecture. The RTX 5090 GPU brings several improvements including high-speed GDDR7 memory, a massive 21,760 CUDA cores, 512-bit memory bus, DLSS 4, and more. On top of that, RTX 5090 is manufactured on TSMC's 4nm (N4P) process node and houses 92 billion transistors. At the GTC 2024, Nvidia CEO Jensen Huang said the Blackwell architecture is designed for the Generative AI era. So with the RTX 5090, you can expect both high-end gaming and top-notch AI performance. The Nvidia GeForce RTX 5090 is a powerhouse of a GPU. It has become the first GPU to feature over 20,000 CUDA cores on a consumer GPU. In fact, it has more CUDA cores than Nvidia H100 (16,896 CUDA cores) -- a server-class GPU. As for clock speed, RTX 5090's base frequency is 2.01GHz and can boost the clock speed to 2.41GHz. As a result, the Shader cores on GeForce RTX 5090 can perform up to 125 TFLOPS; the 5th-gen Tensor cores can process up to 3,352 trillion operations per second (TOPS); and the 4th-gen Ray Tracing cores can deliver graphics performance up to 318 TFLOPS. As for real-world gaming performance, Nvidia says RTX 5090 delivers 2x better performance than RTX 4090 in gaming titles such as Cyberpunk 2077, Alan Wake 2, and Black Myth: Wukong with DLSS 4 and Full RT. In Generative AI workload, RTX 5090 again delivers twice the performance of RTX 4090. And DLSS 4 has been significantly improved to improve FPS and reduce latency in games. Nvidia says DLSS 4 comes with features such as Multi Frame Generation, enhanced Ray Reconstruction, and Super Resolution. Aided by 5th-gen Tensor cores on RTX 5090, you are going to have the best gaming experience on RTX 50-series GPU. The Nvidia GeForce RTX 5090 GPU features 32GB of ultra-fast GDDR7 memory from Micron for exceptional performance. It's aided by a 512-bit memory bus, allowing the memory to operate at a remarkable 1.8 TBps (1,792 GBps), which is 2x faster than RTX 4090. With such high-speed memory, the RTX 5090 GPU can quickly cycle through operations, delivering unmatched gaming and AI performance. The Nvidia GeForce RTX 5090 features 21,760 CUDA cores and has access to high-speed GDDR7 memory, so it's expected that it will consume more power than its predecessor. Nvidia says RTX 5090 has a TDP of 575W which is 125W more than RTX 4090. For enhanced cooling, RTX 5090 requires a large 3.5-slot cooler that can handle power draw up to 600W. Nvidia launched the GeForce RTX 5090 GPU for consumers on January 7, 2025. RTX 5090 starts at $1,999 and will start shipping in January. So are you impressed with RTX 5090's performance? Let us know your opinion in the comments below.

The Nvidia GeForce RTX 5090 will officially burst onto the scene for a eye watering $1,999 this January

Don't worry, there's a $549 RTX 5070 coming that boast RTX 4090 performance (sort of) Nvidia has made the GeForce RTX 5090 official during its CES 2025 keynote, and the next-gen graphics card will arrive this January for x. Naturally, it'll also be joined by sibling models like the RTX 5080 and RTX 5070, and while the latter would suit my personal budget, I know you're all waiting to hear more about the new GPU monarch. If, like me, you tuned into this year's CES event to hear more about the Nvidia GeForce RTX 5090, you also likely had to wait through a bunch of AI announcements. Perhaps that's your bag, but my objective was to learn about the new flagship's price and capabilities. Now that we're at the other end of the keynote, we now know the RTX 5090 for $1,999 alongside the RTX 5080 for $999, RTX 5070 Ti for $749 and the RTX 5070 for $549 this January. I'll keep you posted when I have an exact date and specific specs for each card, but CEO Jensen was pretty hyped to reveal that the latter will provide RTX 4090 levels of performance. There's no getting round that the company is now hyper focused on AI, but it also plans on using the tech to completely "revolutionize GeForce." Clad in a black snakeskin jacket, Jensen did kick things off with a quick graphics card history lesson, covering everything from Virtua Fighter to the first gaming GPU that debuted in 1999 (the Nvidia GeForce 25). By that, Jensen means that new graphics cards like the RTX Blackwell family will use new forms of DLSS to completely generate graphics from scratch. The green team boss says the tech can effectively "predict the future," which is a bit cheesy, but he means that Blackwell GPUs can create 33 million pixels using 2 million. As you can imagine, that means a lot less native computational power is required thanks to really good graphical guess work. It's that same tech that allows cheaper cards like the upcoming RTX 5070 to apparently pull off Nvidia GeForce RTX 4090 levels of performance for a quarter of the price. I'm curious to know how much of that heavy lifting is linked to AI rather than native performance, and whether there will only be limited number of games that support the new version of DLSS. Nevertheless, it's an impressive feat, and it it hammers home that the RTX 5090 will be absolutely monstrous. Right now, the keynote is still ongoing, and I'm still gathering up information of specs and specific release dates. I'll keep you all updated on that, but for know, just know that if you're looking to splash out on a new GPU, some serious powerhouses are coming this month.

Nvidia RTX 5090 Reveal Showcases 'Blackwell' Series With Triple the AI Performance

The 'Blackwell' series offers a range of options to suit different budgets, with the RTX 5080 priced at $999, the RTX 5070 Ti at $749, and the RTX 5070 at $549. Nvidia has always been a pioneer in GPU technology, and we must admit that most of our PC rigs have one RTX card. At CES 2025, Nvidia CEO Jensen Huang revealed the RTX 'Blackwell' series cards, unveiling a dazzling performance for the RTX 5090. Here is what you need about the RTX 5000 'Blackwell' series cards. The highest 'Blackwell' GPU, RTX 5090, will feature 3352 AI TOPS, while the 5080 has 1801, the 5070 Ti has 1406, and the 5070 has 988 AI TOPS. Besides AI computing, all the 5000 series cards will run on the 5th gen tensor cores. For better raytracing and visuals, the RTX 5090 will have 380 RT TFLOPS and 125 Shader TFPLOS. As for the memory bandwidth, Nvidia RTX 5090 comes with an Ultra-Fast G7 memory. The upcoming GPU will have a 1792 GB/sec memory bandwidth for the highest 'Blackwell' cards. Moreover, it will have a massive 32 GB GDDR7 VRAM, almost triple the 4090's 12 GB module. This was revealed earlier through a leaked RTX 5090 custom card. The other 'Blackwell' 5000 series cards from Nvidia will have 16 GB GDDR7 VRAM, while the RTX 5070 comes with a 12 GB variant. The new DLSS 4 pulls a powerful AI to generate multiple frames simultaneously, resulting in significantly higher frame rates. This generation of technology comes with several advancements, including RTX Neural Shaders, RTX Neural Face rendering for hyperrealistic human faces, RTX Mega Geometry for rendering vast environments, and Reflex 2 for reduced latency. Nvidia demonstrated a scene rendered at 27 frames per second (FPS) with DLSS disabled, DLSS2, and DLSS 3.5. All showed significant advancement over the others. However, DLSS 4 delivers a remarkable 247 FPS while maintaining a 34-millisecond latency. This translates to an over eightfold performance increase compared to systems lacking AI-powered predictive processing. Towards the end of the GPU presentation, the pricing for all the RTX 5000 series cards is revealed. As per the reveal, the Nvidia RTX 5090 will have a price of $1,999. The other cards from the 'Blackwell' series will be priced based on their VRAM, AI TOPS, and performance: RTX 5080 at $999, RTX 5070 Ti at $749, and RTX 5070 at $549. Along with the GPU pricing, we also learn about the laptop prices that feature 'Blackwell' series cards. The laptops will be priced based on their GPU, AI TOPS, and memory: RTX 5090 with 1,824 AI TOPS and 24GB at $2,899, RTX 5080 with 1,334 AI TOPS and 16GB at $2,199, RTX 5070 Ti with 992 AI TOPS and 12GB at $1,599, and RTX 5070 with 798 AI TOPS and 8GB at $1,299. The GeForce RTX 50 Series for laptops comes with massive improvements. It delivers twice the efficiency of its predecessor, offering increased performance while consuming only half the power. This translates to a 40% longer battery life with Black Max-Q. Furthermore, the series supports generative AI models twice as large as before, all while maintaining a remarkably slim profile with laptops as thin as 14.9 millimeters. Along with the Nvidia RTX 5090 and the whole 'Blackwell' 5000 series reveal, CEO Jensen Huang mentioned three new AI models. The new Llama Neomotron Language Foundation Models are Nano, Super, and Ultra. The 'Nano' AI model is the cost-efficient, low-latency model, best for PC and edge devices. 'Super' comes with superior accuracy, whereas the 'Ultra' is the best and works perfectly in data-center-scale operations. Are you excited to get your hands on the Nvidia RTX 5090 or any other 'Blackwell' series cards? Which part of the new-gen GPU do you like the most? Do tell us in the comments below.

Nvidia Unveils the RTX 50-Series GPUs Led By the Ginormous RTX 5090

Nvidia finally pulls the lid off the RTX 5070, 5080, and the 5090â€"one of the most powerful consumer GPUs yet. PhotoNvidia has finally pulled the curtain off its next series of graphics cards. If you somehow avoided the mountain of rumors and leaks about Nvidia’s powerful, humongous, and energy-hungry RTX 50-series GPUs, that are all that and more. The new Nvidia GeForce RTX 50-series includes all the usual suspects, though we’ll be seeing the more powerful cards first. The RTX 5070, 5070 Ti, 5080, and top-of-the-line 5090 are all present and accounted for. It’s the first time Nvidia has decided to release a “titanium†edition card alongside the main card, though there’s nothing wrong with having more options up front. They are also the most expensive cards the company has ever produced. The RTX 4090 will retail for $1,600. To take the sting off, the company also declared the RTX 5070 will go for $550. The RTX 5070 Ti will cost $750 and the 5080 will be $1,000. Nvidia CEO Jensen Huang came out to the stage of the Michelob Ultra arena in the Mandalay Bay casino and first asked the crowd "do you like my jacket?" He has a reason to act like a celebrity. The company's AI training chips have made it one of the most profitable in the world. AI was the main theme behind his entire presentation. It was one of the big reasons the new RTX video cards specs seem so outlandish. He further claimed that people who take a RTX 5070, even on a laptop, can expect performance equivalent to the RTX 4090. A big part of that, he said, is because of the company's Tensor cores and it's AI that back up the card's performance. "GeForce brought AI to the masses," he said. "Now we're bringing AI to GeForce." The top-end Blackwell cards will have 380 RT TFLOPS and 125 shader TFLOPS plus a 1.8 TB/s memory bandwidth. We'll get more word on the exact numbers of CUDA cores and VRAM soon, but it's safe to say the RTX 5090 is 50% more card than Nvidia’s previous massive, ultra-powerful GPU, the RTX 4090. Add to all that is an app new AI upscaled software suite in DLSS 4. This now includes updated ray reconstruction and super resolution models aided with AI transformer models. You’d have to have a big case and big pockets to even think about sticking an RTX 5090 in your case. Meanwhile, the RTX 5080 and RTX 5070 offer power beyond their counterparts from two years ago. The new Blackwell cards should arrive within the next few weeks. What didn't get any air was the any mention of more budget-end cards. We'll need to wait a few months for the RTX 5050, 5060, and 5060 Ti to show up.

Nvidia's RTX 5070 claims show why it's best to wait for the 50-series before buying a gaming laptop

We just wrapped up CES 2025, the Consumer Electronics Show that gave us a first look at the most exciting future technology we can expect over the course of the year. During this year's showcase, Nvidia unveiled its latest line-up of graphics cards for desktops, alongside the price ranges and expected performance of laptops that will feature their mobile versions. Yes, the Nvidia RTX 50-series has finally been revealed, making it over two years since the first 40-series GPU was unveiled in September, 2022. Back in December, I wrote a piece about how you should wait until 2025 before buying a gaming laptop, but that largely relied on how soon Nvidia would actually get the next generation of GPUs out. And now, after seeing official release dates and some details on what these GPUs are capable of, I stand by my claim: Wait until the 50-series before purchasing yourself a shiny, new gaming laptop. I predicted that we could see RTX 50-series laptops launching throughout May 2025, with release dates potentially coming sooner. Not my worst guess yet, as we'll see the RTX 5070 Ti, 5080 and 5090 laptops in March, while 5070 laptops will come in April. In just a few months, we'll see what the next generation of graphical power is capable of. I claimed in December that if you're considering buying an RTX 40-series laptop now, it's your best bet to wait, as the additional power that we expect out of the RTX 50-series would more than likely be worth the wait, and the potentially extra cost. However, Nvidia's claims about the RTX 50-series are pretty intense, to say the least. In a popular image from the company's CES 2025 press briefing, Nvidia showcases an RTX 5070 laptop that costs $1,299, with the caption "4090 performance, half the power." If it isn't clear just how bold of a statement that is, the desktop version of the RTX 5070 will launch at $549, while the desktop version of the RTX 4090 launched at $1,599. One card is nearly three times the cost of the other, yet Nvidia is claiming their mobile variants are equivalent in power. Even if we avoid talking about desktop GPUs and focus on laptops specifically, Nvidia claims that RTX 5070 laptops will start at $1,299. Comparatively, the latest RTX 4090 laptop we reviewed is the Origin EON16-X, and the model we tested cost $3,620. According to Nvidia's claim, the RTX 5070 can yield identical performance to a GPU found in gaming laptops that cost nearly three times as much. However, the boldness of these claims puts a few things into question. Nvidia's slide shows that RTX 5070 laptops will start at $1,299, but that's a generous estimate and isn't considerate towards the other aspects of a laptop. Sure, there will likely be an RTX 5070 laptop available for $1,299, but who's to say if its display will be any good, or if it'll have a decent processor, a high amount of RAM, or a reasonable amount of storage. And while it may not seem as important to the average gamer, a well-designed chassis is vital. You don't want your laptop to feel like a flimsy piece of junk, and we often find that the cheapest laptops typically suffer from this exact problem. For context, back during Nvidia's CES 2023 keynote, the company claimed the RTX 4050, 4060, and 4070 laptops would be starting at $999, whereas the RTX 4080 and 4090 laptops would start at $1,999. In reality, the first RTX 4070 laptop we reviewed was the Asus ROG Strix G16 (G614J), and the model we tested cost $1,999. Another one of the earliest RTX 4070 laptops we reviewed is the Origin EON-16S, which cost $2,386 at the time of testing. While Nvidia is likely being honest about the $1,299 starting point for the RTX 5070 series, it's only a starting point for a reason. You're most likely still going to spend up to $2,000 or more if you want to ensure the laptop has a solid processor, plenty of RAM and storage, and a decent display. But that beckons the question: Regardless of whether or not the estimated price point is a generous one, is there any validity to Nvidia's claim about the performance potential of RTX 5070? Let's break down the RTX 5070 mobile version's specs: It's built with 4,608 CUDA cores, which determines how many parallel processing tasks it can perform at once. Essentially, the higher the number, the better. It also features 798 AI TOPS, which measures how many computing operations it can handle in its AI processing. Finally, it sports 8GB of GDDR7 vRAM, which is vital for smooth gameplay experiences (vRAM determines how much graphical data a GPU can store). Comparing it directly to the RTX 4090 mobile version makes Nvidia's claims a bit questionable, as it's built with 9,728 CUDA cores and 16GB of GGDR6 vRAM. It's a bit worse at AI, with only 686 AI TOPS, but there's no avoiding a very obvious conclusion here: In terms of pure technical power, the RTX 4090 is significantly stronger than the RTX 5070. In reality, what Nvidia is talking about all boils down to DLSS, otherwise known as Deep Learning Super Sampling. It's AI-powered upscaling and frame generation that can vastly improve visible framerates. Nvidia recently unveiled DLSS 4, which apparently can generate three additional frames for every one frame originally existing. Nvidia is calling this "DLSS Multi Frame Generation," which is exclusive to the GeForce RTX 50-series. The Verge spoke to Lars Weinand, Senior Technical Product Manager at Nvidia, in order to get some clarification of the company's claims. "Using neural rendering and DLSS 4 we can reach performance levels [on an RTX 5070] that were only possible with an RTX 4090," Weinand says. So, this isn't about the RTX 5070 being able to match the RTX 4090 in power on every level but instead refers to what the company has seen while using DLSS 4. Keep in mind that not all games support DLSS 4, but it seems like the ones that do will shine on RTX 50-series GPUs. While it's a bit less exciting to know that the RTX 5070 isn't really as powerful as an RTX 4090, it's also not remotely surprising. However, the performance differences absolutely still make it worth waiting until the 50-series launches, and if you're doubtful, wait until we get our hands on the hardware and put Nvidia's claims to the test.

Nvidia announces the RTX 50-series, led by the $1,999 RTX 5090 with 'twice the performance of the 4090'

With more AI acceleration than ever even the RTX 5070 is claimed to offer RTX 4090-level performance for $549. Nvidia has announced the RTX 50-series, codename Blackwell, at CES 2025. The new cards come with massive improvements to AI acceleration, with CEO Jensen Huang claiming "AI is coming home to GeForce". CEO Huang announced the entire lineup at CES 2025, from the RTX 5090 to the RTX 5070. Let's get right into prices: The Blackwell family will be led by the RTX 5090, which is a mighty beast. No surprises there. Nvidia cites up to 125 shader TFLOPS, 380 RT TFLOPS, and 4,000 AI TOPS for the Blackwell family, presumably the top GPU configuration with 92 billion transistors, though the RTX 5090 is noted with 3,400 AI TOPS -- so a little below the maximum. For comparison, the RTX 4090 has just over 82 FP32 shader TFLOPS from just over 76 billion transistors. The cards also feature GDDR7 memory, with a maximum of 1.8 TB/s of memory bandwidth. As for performance comparisons, Nvidia didn't give us much to go on except the promise that the RTX 5090 will offer 'twice the performance of the 4090' and the RTX 5070, a $549 card, will compete with the RTX 4090. Though it's likely DLSS 4 and new AI features are playing a big role here. "4090 performance at $549," proclaimed Huang during the show. "Impossible without artificial intelligence. Impossible without 4 teraops of AI Tensor cores. Impossible without the G7 memories." That does suggest that DLSS 4 has a big part to play in the RTX 5070's comparative performance. At the beginning of the stream, Nvidia showed off a scene rendered (inferred?) in real-time on Blackwell, which included new AI features to massively accelerate performance and save memory. The RTX 50-series features an AI Management Processor, and Huang alluded to in-shader AI acceleration to help spread the workload of the new AI systems. As Huang notes, for every four frames in the scene, only one was rendered. The rest were generated by AI. Availability for the RTX 50-series will be sometime this month.

The Desktop GeForce RTX 50 Series, Unveiled: Nvidia's New Graphics Cards Promise Massive AI Gains

For as long as I can remember, I've had love of all things tech, spurred on, in part, by a love of gaming. I began working on computers owned by immediate family members and relatives when I was around 10 years old. I've always sought to learn as much as possible about anything PC, leading to a well-rounded grasp on all things tech today. In my role at PCMag, I greatly enjoy the opportunity to share what I know. During its CES 2025 keynote, Nvidia gave us our first honest look at its upcoming GeForce RTX 50-series graphics cards. Though these new GPUs, based on Nvidia's cutting-edge "Blackwell" architecture, were highly anticipated, the extent to which these new GPUs surpass their predecessors, at least on paper, is still impressive. For example: The GeForce RTX 5070, at $549, will reportedly be able to match the performance of the company's older GeForce RTX 4090, which launched at $1,599, marking an astonishing leap forward in performance if it pans out in testing. Nvidia's GeForce RTX Blackwell Launch Lineup For starters, Nvidia told us that at least four new graphics chips are due in the next few months: the Nvidia GeForce RTX 5090, the GeForce RTX 5080, the RTX 5070 Ti, and the RTX 5070. At launch, Nvidia has not indicated that it will release an Nvidia GeForce RTX 5060 or Nvidia GeForce RTX 5060 Ti, but those levels of card are typical, too, so we'll very likely see one or both of those graphics cards later. At the CES keynote, we didn't get info on fixed shader counts, or any traditional performance numbers (like frame rates), for any of these graphics cards. Instead, Nvidia focused on AI performance and provided some speed info for that metric. The GeForce RTX 5090, which is the fastest model and will feature GDDR7 memory, can reportedly process 3,352 AI TOPS (trillion operations per second) and will contain 92 billion transistors, making it possibly the largest single chip in terms of transistor count ever produced. We don't know the transistor count for the Nvidia GeForce RTX 5080, but its AI processing power is rated at 1,801 TOPS. This suggests that the RTX 5080 likely has roughly 50% of the resources of the RTX 5090, marking a steep step down. The price of these two GPUs also supports this, as the RTX 5090 is priced at $1,999, while the RTX 5080 is priced at $999. The GeForce RTX 5070 Ti, which costs $749, reportedly can process 1,406 AI TOPs. The Nvidia GeForce RTX 5070, at $549, has the power to handle 988 AI TOPS. Blackwell Conjures Up Nvidia DLSS 4 & AI Processing The prices Nvidia set on its new GPUs weren't all that surprising, as they fall roughly in line with those set by Nvidia on its current GeForce RTX 40-series GPUs. Far more startling was Nvidia's claim that its new GeForce RTX 5070 would be able to match the RTX 4090 in gaming performance. This is shocking not only because the RTX 4090 is currently the world's fastest single graphics chip, but also due to the significant price difference between the two, with the RTX 4090 having an MSRP of $1,599 and frequently selling for more than that. A close look at the rest of Nvidia's keynote gives us some insights into how this is possible, and it's all down to AI. The AI hardware on these new GPUs is cutting-edge and pushed to provide far greater performance than previous generations could handle. "Blackwell, the engine of AI, has arrived for PC gamers, developers, and creatives," said Jensen Huang, founder and CEO of Nvidia. "Fusing AI-driven neural rendering and ray tracing, Blackwell is the most significant computer graphics innovation since we introduced programmable shading 25 years ago." During the keynote, Huang stated that for one of its demonstrations, the GPU only had to render 2 million pixels in a scene that contained 33 million pixels, with all of the remaining pixels generated by AI. Two key technologies make this possible. One is Nvidia's new DLSS 4 technology, which takes frame generation to a new level by generating more frames based on fewer pixels. It also utilizes ray reconstruction and super-resolution models to do this with greater image fidelity than previous generations of DLSS. Nvidia also introduced a technology called RTX Neural Shaders, which similarly helps generate higher-quality images such as 3D character models. By taking advantage of these technologies and the powerful AI hardware, this is likely how the RTX 5070 can match the RTX 4090 in games. However, at the same time, it's likely that the Nvidia GeForce RTX 5070 won't be able to match the RTX 4090 in games that don't support DLSS, which makes support for this feature critically important. The RTX 50 Series: Shipping Soon If you're eager to get your hands on one of Nvidia's new RTX 50-series GPUs, you won't have to wait much longer. The high-end RTX 5090 and RTX 5080 will be available on January 30, with the GeForce RTX 5070 Ti and RTX 5070 to follow in February. Nvidia will also have mobile versions of these graphics chips coming just slightly later in March. The mobile versions of these chips are listed as having less overall AI performance, likely because their power consumption and clocks were reduced to work better inside a mobile platform. Laptops with an Nvidia GeForce RTX 5090 will start at $2,899, while systems with an RTX 5080 will start at $2,199. Notebooks equipped with the RTX 5070 Ti are priced at $1,599, and PCs with the RTX 5070 are comparatively affordable, starting at $1,299. But, of course, if you want the utmost power, that will reside in Nvidia's two-grand desktop GPU behemoth, the GeForce RTX 5090. Stay tuned for our full review of the new lineup in the coming weeks.

CSE 2025: Is NVIDIA's $2,000 RTX 5090 the Future of GPUs or Overpriced Hype?

The RTX 5090 recently unveiled by NVIDIA has received a lot of attention, not only because of the technology behind it but also because of the $2,000 price that makes it the most expensive desktop graphics card in the market. During a question and answer segment at CES 2025, executive director Jensen Huang called for such a push by stating, "When someone would like to have the best, they just go for the best." Huang insists the RTX 5090 is not just a new addition to its graphics card line; it is a step forward in evolution. The new flagship is built on NVIDIA's ground-breaking fourth-generation DLSS technology, which includes multi-frame generation and maxes out the performance of the previous models. He said that these features are impossible to implement without Artificial Intelligence and, thus, introduced the RTX 5090 as a future-oriented tool for gamers and creators. Although the price of $ 2,000 crossed everyone's mind, this is not the first time NVIDIA has left a rather reasonable price for other RTX 50-series cards. The RTX 5080 costs $1000 and is similarly priced to the RTX 4080 Super; the RTX 5070 Ti costs $750 which is $50 cheaper than the previous model, the RTX 5070 costs $550. This strategic pricing ensures that we continue to appeal to both the hobby category and the more general populace. Key to the RTX 5090's appeal is its exclusive DLSS 4 technology, which includes groundbreaking 4X frame generation. With this feature, only one out of every four frames is rendered by the GPU, dramatically improving efficiency and performance. NVIDIA claims the RTX 50-series delivers twice the performance of the previous generation, promising a transformative gaming and content creation experience. Although Huang hinted at future expansions, including a potential RTX 5060, the RTX 5090's launch on January 30 will be the ultimate test of whether it lives up to its lofty promises.

RTX 5090 will empty your wallet but provide twice the performance of the RTX 4090

Blackwell based graphics cards are here, with massive increases in performance and pricing. While most had feared a price increase, the new RTX 5090 will be released at $2,000, or more than €2,200, and will be available for purchase on January 30th. This is made even sadder by the fact that Nvidia's Founder Edition cards are usually the cheapest option, but also extremely limited, so prices will most likely be close to, and above €2,500 for third party cards. The card is, as everything lately, crammed with AI and neural based technology, with an explanation on this provided by Jensen Huang, founder and CEO of Nvidia, as the following: "Blackwell, the engine of AI, has arrived for PC gamers, developers and creatives. Fusing AI-driven neural rendering and ray tracing, Blackwell is the most significant computer graphics innovation since we introduced programmable shading 25 years ago." RTX 5090 features 92 billion transistors, providing over 3,352 trillion AI operations per second (TOPS). Nvidia has, as usual, provided a collection of numbers for us to compare, but as with anything in tech, we will have do verify those numbers independently when possible. So, what do you get? A lot actually. Nvidia is promising twice the performance from 32GB of VRAM, and you finally get to use your PCIe 5.0 slot as intended. It also uses DLSS 4 with multi-frame generation, instead of making one frame each time, and it can now generate three out of thin air, vastly improving FPS. With transformer-based DLSS Ray Reconstruction (it's called "transformer model", and no, no Autobots were harmed in the making) and improved Super Resolution, ghosting has been reduced, anti-aliasing enhanced, and stability improved according to Nvidia. While a massive 75 games are supported at release, DLSS 4 does not seem to be compatible with RTX 40 Series cards or older. The full specs can be seen in the image below: It's also noteworthy that Nvidia seems to be returning to the two-slot format, and while the RTX 5090 has had a massive uplift in pricing, the cheaper cards are hit less by price increases. In addition, AI is now integrated into the programmable shaders that use the new RTX Neural Shaders system, further giving us film-quality graphics according to Nvidia. A separate system for faces has also been implemented.

Nvidia's new RTX 5070 will deliver 'RTX 4090 performance at $549' when it launches in February

The lowest spec new RTX Blackwell GPU is capable of frame rates akin to the $1,599 RTX 4090 of the previous generation, though likely only through new DLSS 4 AI smarts. During today's Nvidia CES 2025 keynote, Jen-Hsun Huang announced the new RTX Blackwell GeForce GPUs, with the RTX 5070 propping up the new range, offering "RTX 4090 performance at $549." That's a pretty spectacular statement, with a previous gen GPU many tiers above the RTX xx70 level being superseded by a card costing a third less. The RTX 4090 was a $1,599 GPU at best, and generally costed far more throughout its lifetime, but is now being matched by a $549 card. But, it is worth pointing out that's likely down to the new DLSS 4 magic that Nvidia is baking into its new RTX 50-series chips. With the new Multi Frame Generation feature of the RTX Blackwell family you are going to be able to compute one frame and then have up to three further frames generated by Nvidia's AI chops. That's likely how Nvidia gets to its claims of RTX 4090 performance; it's going to be more about the experience you get in one of the 75 apps and games that will support DLSS 4 at launch than the sort of all-round RTX 4090 performance you will get as standard through the silicon alone. Still, that's an impressive offering, and speaks to the power of the new 'transformer' model that Nvidia has switched its entire DLSS stack over to. The full stack features four separate cards, with the $1,999 RTX 5090 topping it off, the RTX 5080 appearing at the same $999 as the RTX 4080 Super, and then the RTX 5070 Ti and RTX 5070 rounding it off for $749 and $549 respectively. Which means all those rumours of massive price increases for the RTX 50-series were rather overblown. Unsurprisingly. Okay, the RTX 5090 is considerably more expensive than the RTX 4090 was -- and is reportedly going to offer twice the performance of the top Ada card -- but that last-gen card was the best-value graphics card of the entire Ada stack. It's no surprise that its successor its a $2,000 GPU now. The fact the RTX 5080 is sticking to the same $999 price point is pretty powerful stuff... depending on the GPU inside of it. Nvidia is claiming the card is twice as powerful as the RTX 4080, but again that's when using DLSS 4 and Multi Frame Generation. But they're all coming soon. The RTX 5090 and RTX 5080 will be launching on January 30, with the RTX 5070 Ti and RTX 5070 will be following in February.

At CES 2025, Nvidia dropped the mic while Radeon dropped the ball

CES 2025 was packed with all kinds of techie announcements, but for me the highlight was all the new graphics cards. Nvidia and AMD both played their hands with next-generation products, with some caveats. Nvidia certainly stole the show, with the GeForce RTX 50 Series announcement serving as the key opening message in CEO Jensen Huang's keynote speech. AMD was much softer in its talk, with sparse details on its new RDNA 4 graphics cards, resulting in similarly sparse enthusiasm. Let's go over what we learned from both companies about their GPUs and where your attention should be in 2025. Right off the bat, here's the good news for us: We're definitely getting some new GPUs, starting with Nvidia's GeForce RTX 50 Series. The top-dog in that group will be the RTX 5090, featuring a staggering 32GB of fast GDDR7 VRAM. According to Nvidia, it may be as much as twice as fast as the outgoing flagship RTX 4090. While that's certainly impressive, Nvidia is likely basing its claims on its DLSS 4 numbers, which feature several new technologies that help it achieve all this. Chief among them is Multi Frame Generation, using the power of AI to reach higher peaks in performance with little to no apparent drawbacks. Even more impressive is the claim by Nvidia that the new RTX 5070 will have similar performance to the RTX 4090. That's a bold statement considering that the RTX 5070 will have an MSRP of $549 while the RTX 4090 sold with an MSRP of $1,599. (More on this later, as once again AI is a big factor in these results.) The high end GPUs -- such as the RTX 5070 Ti and RTX 5080 -- also look promising, with both having 16GB of GDDR7 VRAM. The RTX 5080 appears to be half of what the RTX 5090 is, both in hardware specs along with its $999 MSRP. That's where the good news is. Pricing. The GeForce RTX 5080 has stuck with the revamped $999 price point of its RTX 4080 Super predecessor, a marked improvement over the original RTX 4080's unpopular $1,199. The RTX 5070 fares even better at $549, a $50 reduction compared to the current RTX 4070 Super's $599 price tag. Couple with that a decent performance increase and it's clearly a good value for the money. The RTX 5070 Ti also sees a price drop to $749, which is good news if you're shopping in the high end. But then we have an outlier in the group: the mighty RTX 5090, which climbed 25 percent of the RTX 4090's price, up to $1,999. Yet while that might seem steep, early rumors pinned the price of the RTX 5090 at well over this amount. We won't know for sure until we review this GPU, but the price increase could be justified on paper. Not only does it feature a first-ever 32GB of VRAM in a gaming GPU, but the rest of the spec sheet also reads like a superhero comic book. It will be a stellar GPU for not just gaming, but productivity and other workloads, too. The new technologies embedded within these GPUs will produce some stunning games in the future. Nvidia's neural rendering looks very promising, utilizing the the 5th generation tensor cores. Upgraded RT cores will also have a significant impact on ray tracing, something that enthusiast gamers have started to find almost necessary for the most visually pleasing eye candy. These demanding technologies pair like fine wine with DLSS 4 and Multi Frame Generation, erasing any FPS performance penalties while not impacting visuals to a noticeable degree. Note: Early performance numbers should be taken with a grain of salt, of course, because Nvidia is showcasing the likes of DLSS 4 and Multi Frame Generation for its new GPUs. We don't yet have the full context of what performance will be like across most games, especially those that may not support these new technologies. While Nvidia was busy dropping the mic, Radeon RDNA 4 GPUs were barely touched on by AMD. Details were amiss in AMD's presentation, and specs for the new products were hard to find. The good news is that we at least have confirmation that GPUs like the Radeon RX 9070 XT will have 16GB of VRAM and will be coming soon. Even though the company devoted zero air time to the RTX 9070 in its CES keynote, AMD has a lot of positives with AI and RDNA 4 that should help its GPUs perform well. FSR 4 is also a notable technology that will allow AMD to compete with Nvidia's DLSS 4, although Nvidia does have a head start as far as AI tech in the gaming space. AMD has stated in the past that it isn't chasing the crown for fastest GPU this generation, instead focusing on the mid-to-entry-high-range GPUs that'll result in more gamers choosing its products. In turn, this will mean there are more developers willing to optimize for their GPUs and their larger base of users. (AMD has previously done well with RDNA 3 GPUs, such as the Radeon RX 7800 XT, bringing some great price-performance to the table in the battle against Nvidia.) While the strategy is fine, AMD did seem a bit too hush-hush on its new products at CES 2025. Perhaps they wanted Nvidia to fully play their hand, like announcing their intent to launch the $549 GeForce RTX 5070, a product that competes directly in the price range and market that AMD is targeting. The RTX 5090, not so much -- unless AMD pulls out a surprise like they did with the RDNA 3-based Radeon RX 7900 XTX a few years ago. There's no doubt that AMD will have some excellent GPUs with RDNA 4, but the optics of a high-profile event like CES 2025 are also important. Even Nvidia, with its massive revenue stemming from data centers and AI, still gave proper attention to its gaming products during their keynote. Something like this can cause an early mindshare shift with gamers, since we don't have details aside from a few pictures and sparse information. This fails to build enthusiasm and only gives the nod to Nvidia to bulldoze this generation of graphics cards for enthusiasts. To be sure, apart from Nvidia-curated graphs, we don't know the full details of the GeForce RTX 50 Series and its capabilities. But at least we have some rough ideas to get excited about, along with the all-important pricing that dictates what gamers will ultimately buy. We would've liked to see more info from AMD, even a slight mention in its presentation, to give us a sense of what to expect. Sadly, AMD really the dropped the ball. Only time will tell how much it affects Radeon's prospects.

Nvidia RTX 50 Series: Everything We Know So Far -- PCIe 5.0, GDDR7 RAM Confirmed

For as long as I can remember, I've had love of all things tech, spurred on, in part, by a love of gaming. I began working on computers owned by immediate family members and relatives when I was around 10 years old. I've always sought to learn as much as possible about anything PC, leading to a well-rounded grasp on all things tech today. In my role at PCMag, I greatly enjoy the opportunity to share what I know. Nvidia's CES keynote announcement of its upcoming GeForce RTX 50-series graphics cards was exciting, but with few details revealed except names and prices, I was eager to learn more. Sooner than expected, that wait is over. I tracked down core counts and clock speeds, giving us a better picture of these GPUs' capabilities and the first solid basis for comparison with the outgoing RTX 40-series. Naturally, the biggest focus of the RTX 50-series has been on its new flagship, the Nvidia GeForce RTX 5090. This card has a monster of a graphics chip with 21,760 CUDA cores clocked at 2.01GHz with a boost clock of 2.41GHz. It also has 32GB of GDDR7 memory and a 512-bit wide memory interface. Though incredibly powerful, this isn't the graphics card that gamers and gaming enthusiasts should consider, as its $1,999 price tag and advanced AI processing technology all but ensure it will be snatched up by businesses and be relatively rare in gaming PCs. Anyone who does buy it should ensure they have a potent power supply, as Nvidia lists this card as requiring a 1,000-watt power supply and says the card can consume up to 575W by itself. RTX 5080 Far more obtainable for gamers will be the Nvidia GeForce RTX 5080, which is essentially half of an Nvidia GeForce RTX 5090. The RTX 5080 will feature 10,752 CUDA cores clocked at 2.3GHz with a boost clock of 2.62GHz. That's only a modest increase from the current Nvidia GeForce RTX 4080, which has 9,728 CUDA cores, and the Nvidia GeForce RTX 4080 Super, which has 10,240, but everything from Nvidia so far suggests the AI hardware has improved substantially. According to an Asus product page, the RTX 5080 GPU will have 30Gbps GDDR7 RAM connected to the GPU die over a 256-bit memory interface. The math gives us an effective bandwidth of 960GB/s between the GPU die and the memory. The last-gen RTX 4080 and RTX 4080 Super also featured a 256-bit wide memory interface, but those used slower GDDR6X memory, and their bandwidth topped out at 716GB/s and 736GB/s, respectively. This means the new RTX 5080 has a more significant 30% bandwidth advantage over the RTX 4080 Super compared with the more meager 5% increase in CUDA cores. Nvidia has not released details on the rest of its shading resources. Still, often the proportion of texture mapping units (TMUs) and raster operation processors (ROPs) to CUDA cores doesn't change from one generation to the next. Nvidia packs most of its shading resources into a logic block called a streaming multi-processor (SM) for ease of scalability. If this remains unchanged from the RTX 40 series, the RTX 5080 will have 84 SMs with 336 TMUs. The ROPs aren't packaged inside the SMs, which makes them harder to estimate, but they are instead closely associated with the memory interface and are likely to remain unchanged or relatively close to the previous generation in overall count. That suggests the RTX 5080 will have just 112 ROPs. Other than those specs, I found just two other noteworthy pieces of information on the product page. First, the RTX 5080 supports a PCIe 5.0 x16 connection, an upgrade from the PCIe 4.0 x16 connection used on the last-gen RTX 40-series cards. In fact, all of these new graphics cards use PCIe 5.0 connections up and down the stack. We'll see whether this holds up with eventual 5060 and 5050 cards. Second, the recommended PSU Nvidia listed on its spec page is 850W, an increase from the 750W rating Nvidia set on the RTX 4080 and RTX 4080 Super. This suggests the new RTX 5080 will be more power-hungry with a max power consumption of 360W. RTX 5070 Ti For now, those are all the details I could find on the new RTX 5080, but similar info for both the Nvidia GeForce RTX 5070 and the GeForce RTX 5070 Ti is available. These GPUs will have 6,144 and 8,960 CUDA cores, respectively. One of the more interesting details we didn't know before is that both cards will also ship with GDDR7 RAM. When new high-end memory technologies first come out, it's common that they are only used on the highest-end products in a line due to cost and supply limitations, so this wasn't at all a guaranteed thing before now. The RTX 5070 Ti is clocked at 2.3GHz with a max turbo of 2.45GHz. It is paired with 16GB of GDDR7 memory that Gigabyte suggests are 28Gbps memory chips, which would give it a bandwidth of 896GB/s. Using similar math as I used for the RTX 5080 above, I would estimate that this card will have 70 SMs, 280 TMUs, and 112 ROPs, though this is speculative. Nvidia recommends a 750W PSU for the RTX 5070 Ti and lists its max power draw at 300W. RTX 5070 As for the Nvidia GeForce RTX 5070 (no-Ti), its 6,144 CUDA cores are clocked at 2.16GHz with a higher turbo of 2.51GHz than the RTX 5070 Ti to help make up for the lower core count. I would estimate it to have 48 SMs based on its CUDA core count with 192 TMUs and around 80 ROPs. This model also has a 192-bit wide memory interface and will connect to 12GB of 28Gbps GDDR7, according to Gigabyte, giving it an effective bandwidth of 672GB/s. Power consumption is lower here, with a max power draw of 250W, and the recommended PSU power rating is 650W. It's also worth remembering that Nvidia said its GeForce RTX 5070 would be able to match the RTX 4090 in games, which, if true, is all the more impressive based on its hardware resources and power draw. Both are far closer to the last-gen Nvidia GeForce RTX 4070, with the core count even lower than the Nvidia GeForce RTX 4070 Super. The RTX 4070 Super also has similar clocks to the RTX 5070, which suggests it could be faster than the RTX 5070 in some situations. The RTX 5070 is likely only able to match the RTX 4090 when using its AI hardware to gain an advantage, so it'll be exciting to see whether the RTX 5070 can beat the RTX 4070 Super when DLSS 4 isn't used. For now, that's all the information I've gleaned about the new RTX 50 series, but stay tuned for more updates as we approach the launch of these graphics cards this month.

Nvidia Seeks To Turbocharge AI PC Development With GeForce RTX 50 GPUs

While Nvidia positioned the Blackwell-based GeForce RTX 50 GPUs as a significant upgrade for PC gamers at CES 2025, the AI computing giant is also hoping to lure AI developers and content creators with a bevy of new hardware and software capabilities. Nvidia has revealed its much-anticipated GeForce RTX 50 GPUs for desktops and laptops, relying on the same Blackwell architecture at the center of its new AI data center chips to bring forth PC advancements in graphics, content creation and productivity. At its CES 2025 keynote Monday, Nvidia called the GeForce RTX 50 series the "most powerful" consumer GPUs "ever created," positioning them as a significant upgrade for gamers by claiming they will offer up two times faster graphics performance than the previous generation when using the new AI-powered DLSS 4 image upscaling feature. [Related: Opinion: Why Nvidia, MediaTek May Enter The PC CPU Market Soon] But the AI computing giant is also hoping to lure AI developers and content creators with a bevy of new hardware and software capabilities. For instance, the GPU family's new fifth-generation Tensor Cores pack a "massive amount of AI processing horsepower" to run AI models two times faster using less graphics memory when taking advantage of the newly supported 4-bit floating point (FP4) format, according to Nvidia. On the desktop side, the flagship GPU, the 32-GB GeForce RTX 4090, will cost $1,999 when it becomes available along with the $999, 16-GB GeForce RTX 4080 on Jan. 30. The $749, 16-GB GeForce RTX 5070 Ti and $549, 12-GB GeForce RTX 5070 will become available in February. The GPUs will be available from Nvidia and add-in board partners as well as in desktops from system builders such as Falcon Northwest and Maingear. Laptops equipped with RTX 4090, RTX 5080 and RTX 5070 Ti GPUs will debut in March while laptops with the RTX 5070 will launch in the following month from several OEMs, including Acer, Asus, Dell Technologies, HP Inc., Lenovo and MSI. To foster development of AI PC applications on the new RTX 50 series and other recent generations of GeForce GPUs, Nvidia said it plans to release a "pipeline" of Nvidia NIM microservices and Nvidia AI Blueprints that use first- and third-party models to enable use cases ranging from PDF extraction, computer vision and speech to image generation, large language models and embedded models for retrieval-augmented generation. These models include Nvidia's newly released Llama Nemotron family of models, which are versions of Meta's Llama models that have been optimized to aid with the development of agentic AI use cases ranging from instruction following and chat to coding and math. To demonstrate how NIMs can be used to build AI agents and assistants, Nvidia plans to release a vison-enabled PC avatar called Project R2X that can read and summarize documents, fetch information and "assist with desktop apps and video conference calls." It will also be able to connect with cloud AI services such as OpenAI's GPT4o. When it comes to content creation, Nvidia said the new RTX 50 GPUs come with new hardware features to boost video editing and 3-D rendering workloads on top of new software capabilities for image generation as well as voice and video communication. For video editing, Nvidia said the RTX 50 series comes with new video encoders and decoders that provide a "generational leap" in capabilities with support for the 4:2:2 pro-grade color format, the multi-view extension of HEVC (high-efficiency video coding) for 3-D and virtual reality video as well as the new AV1 Ultra High Quality mode. For 3-D rendering, Nvidia said the GPUs come with fourth-generation RT Cores, which enable applications to run 40 percent faster. One way the chips speed up 3-D rendering is through its fourth-generation AI-based image upscaling technology, DLSS 4, which introduces Multi Frame Generation to increase frame rates. For image generation, Nvidia highlighted how the new FP4 support in RTX 50 GPUs will enable models used for such purposes to take up significantly less VRAM compared to the default 16-bit floating point (FP16) format. As an example, the company said the FLUX.1 [dev] model from Black Forest Labs will only need less than 10 GB of memory with FP4, which means it can run on each of the four new RTX 50 GPUs since they range from 32-12 GB in VRAM. By contrast, the FLUX.1 [dev] model running with FP16 would require more than 23 GB of memory, which restricts it to the RTX 4090 and professional GPUs from the last generation. Nvidia said it plans to offer a NIM microservice based on FLUX.1 [dev], which will be made available in an Nvidia AI Blueprint for 3-D guided image generation next month. For voice and video communication, Nvidia said it plans to add two new features to the Nvidia Broadcast app for AI-enhanced video and voice effects. The first feature, Studio Voice, will make a user's microphone sound like a high-quality microphone while the second, Virtual Key Light, can relight a "subject's face to deliver even coverage." These will initially require an RTX 4080 or higher when they become available in February. The flagship GPU in the RTX 50 series, the RTX 5090, is made up of 92 billion transistors, up 21 percent from the 76 billion transistors of its predecessor, the RTX 4090, which debuted in 2022 using Nvidia's previous-generation Ada Lovelace architecture. Across the RTX 50 series, the GPUs come with new fourth-gen RT Cores and fifth-gen Tensor Cores as well as a streaming multiprocessor that "has been updated with more processing throughput and a tighter integration with the Tensor Cores in order to optimize the performance of neural shaders," according to Nvidia. The RTX 5090 comes with 32 GB of GDDR7 memory and 21,760 CUDA cores, up from the 24 GB of GDDR6X memory and 16,385 CUDA cores of the RTX 4090. The GPU's boost and base clock frequencies are 2.41GHz and 2.01GHz, respectively, which are lower than the 2.52GHz and 2.23GHz clock speeds of the RTX 4090. As for performance, the RTX 5090's Tensor Cores are capable of hitting 3,352 trillion operations per second (TOPS) in AI computing performance while the RT Cores can achieve 318 trillion floating-point operations per second (TFLOPS). These figures are 2.5 times and 60 percent faster, respectively, than the 1,321 AI TOPs achieved by the RTX 4090's Tensor Cores and the 191 TFLOPS of the GPU's RT Cores. Nvidia did not disclose Shader Core performance numbers for the RTX 50 series after providing such information for the previous generation. The total graphics power required by the RTX 5090 is 575 watts, up 27 percent from the 450 watts needed for the RTX 4090. The lowest-end GPU, the RTX 5070, requires 250 watts, which is 25 percent higher than the 200 watts needed for the RTX 4070.

GeForce RTX 50 Series announced, flagship RTX 5090 $1999, incredible RTX AI rendering

TL;DR: NVIDIA announced the GeForce RTX 50 Series at CES 2025, featuring the RTX 5090 with a 2X performance boost over the RTX 40 Series, priced at $1999. The series includes the RTX 5070, 5070 Ti, and 5080, with prices starting at $549. Powered by Blackwell, it introduces AI features like Neural Shaders and DLSS 4. Availability begins January 2025. NVIDIA has formally announced the GeForce RTX 50 Series, powered by Blackwell and incredible new AI rendering features like Neural Shaders and DLSS 4. The flagship GeForce RTX 5090, presented by NVIDIA CEO Jensen Huang at CES 2025 in sleek, cutting-edge Founders Edition form, delivers a 2X performance increase over the RTX 40 Series and will ship with an MSRP of $1999. NVIDIA announced four GeForce RTX 50 Series GPUs at CES 2025, and we were there in person (alongside many thousands). The GeForce RTX 5070 will be priced at $549 and is set to deliver GeForce RTX 4090 levels of performance. Incredible! The GeForce RTX 5070 Ti will be priced at $749 (cheaper than the RTX 4070 Ti SUPER's $799 price point). Finally, there's the GeForce RTX 5080 priced at $999 - half the cost of the GeForce RTX 5090. Here are images of all four (it looks like there won't be a GeForce RTX 5070 Ti Founder Edition coming). NVIDIA has confirmed that availability begins this month, January 2025. The GeForce RTX 5090 is an AI beast; it features 92 billion transistors and can deliver 3352 TOPS of AI performance. All of this AI power is being put to good use as DLSS 4's Multi Frame Generation now generates up to three frames per rendered frame. "Blackwell, the engine of AI, has arrived for PC gamers, developers, and creatives," said Jensen Huang, founder and CEO of NVIDIA. "Fusing AI-driven neural rendering and ray tracing, Blackwell is the most significant computer graphics innovation since we introduced programmable shading 25 years ago." With the RTX 50 Series, NVIDIA also introduces RTX Neural Shaders, which will unlock "film-quality materials, lighting, and more in real-time." This includes RTX Neural Faces and RTX Mega Geometry, enabling 100X more ray-traced triangles in a scene. NVIDIA showcased a stunning demonstration of this during its keynote, and as we're at CES, stay tuned for hands-on, in-depth coverage on all things GeForce RTX 50 Series, DLSS 4, RTX Neural Shaders, and more.

Here's how Nvidia's CEO defends the RTX 5090's $2,000 price tag

CES 2025 Read and watch our complete CES coverage here Updated less than 5 minutes ago "When someone would like to have the best, they just go for the best," said Nvidia's CEO Jensen Huang in a Q&A session with media at CES 2025. Huang was speaking on the newly-announced RTX 5090, and its new price tag of $2,000, making it the most expensive desktop graphics card Nvidia has ever released. It's a new high for Nvidia, but also a bold departure from the rest of the range. The next card down in Nvidia's stack, the RTX 5080, comes in at $1,000 -- half the price of the flagship. Huang suggested that customers don't want to deal in micro-segmentation minutia. "$2,000 is not small money, it's fairly high value," Huang said. "But a lot of customers, they just absolutely want the best." Recommended Videos In the session, the decorated CEO covered everything from tariffs to AI scaling to gaming. A lot of focus was placed on Nvidia's new RTX 50-series GPUs, however, which Huang suggests is much more than just a new graphics card generation. "Without AI, and without the Tensor cores and all of the innovation around DLSS 4, these capabilities simply wouldn't be possible." Huang also hinted at possible GPUs lower down the stack, saying, "Is there a 60s? I don't know. It is one of my favorite numbers, though," hinting at a possible RTX 5060 in the future. Get your weekly teardown of the tech behind PC gaming ReSpec Subscribe Check your inbox! Privacy Policy Although the $2,000 price of the RTX 5090 has grabbed headlines, Nvidia's pricing lower down the stack is actually lower than the previous generation. The RTX 5080 matches the RTX 4080 Super at $1,000, while the RTX 5070 Ti and RTX 5070 are $50 less expensive than their last-gen counterparts, clocking in at $750 and $550, respectively. Those are reasonable prices considering Nvidia says that its new range of graphics cards delivers two times the performance of the previous generation. That performance boost largely comes on the back of Nvidia's new DLSS 4 tech, however. DLSS 4 includes upscaling and frame generation like previous versions, but exclusive to RTX 50-series GPUs, DLSS 4 introduces multi-frame generation. With it, cards like the RTX 5090 have access to 4X frame generation, where only one out of every four frames is rendered by the GPU. It's hard to say now if the RTX 5090 will truly be worth $2,000, but it certainly looks like a monster GPU. It won't be long until we know for sure, with both the RTX 5090 and RTX 5080 set to release on January 30.

NVIDIA GeForce RTX 50 Series specs, pricing, and performance - plan that upgrade today

NVIDIA's GeForce RTX 50 Series is almost here, and the company revealed its latest generation of GeForce RTX graphics at CES 2025. The flagship GeForce RTX 5090 and GeForce RTX 5080 are set to launch on January 30, 2025, with stunning new compact Founders Edition models and dozens of revamped designs from NVIDIA's partners, such as MSI, ASUS, GIGABYTE, and many, many, more. The GeForce RTX 50 Series will introduce NVIDIA's latest groundbreaking AI tech for PC gaming, DLSS 4, and laptop GeForce RTX 50 Series chips, which are on the way. DLSS 4 is set to bring an incredible boost to game performance with Path Tracing, as it now generates multiple new frames instead of one. DLSS 4 is bringing a lot more to the table, and the good news is that GeForce RTX 40, 30, and even 20 Series owners will reap the benefits. With NVIDIA unveiling the GeForce RTX 50 Series, we finally have the official specs, pricing, and performance for the lineup. Each model will deliver an incredible 2X performance increase over its GeForce RTX 40 Series counterpart. However, this is with DLSS 4's new Multi Frame Generation enabled. First, though, here's a breakdown of the specs. And here's a look at the GeForce RTX 5090 vs. RTX 4090, GeForce RTX 5080 vs. RTX 4080, GeForce RTX 5070 Ti vs. RTX 4070 Ti, and GeForce RTX 5070 vs. RTX 4070 performance in a range of games, including Cyberpunk 2077, Alan Wake 2, and Black Myth: Wukong. It's hard to tell each model's actual raw performance uplift with these charts, so we'll have to wait for reviews for that information. Speculation and rumor are that these figures will be in the same 20-50% range as previous generations. Stay tuned for more on the GeForce RTX 50 Series at CES 2025, including hands-on previews of all the new features, capabilities, and tech.

Where to buy RTX 5090: retailers I'd pre-order Nvidia's latest graphics card at first

Nvidia's latest flagship GPU is has only just been announced, and I'm already thinking about where to buy the RTX 5090. The new Blackwell model will set you back $1,999 if you manage to grab a Founder's Edition card, but plenty of manufacturers will join the green team's bandwagon with their own custom options. First things first, you'll want to start thinking about RTX 5090 pre-orders. It's not just gamers looking to grab the Nvidia's next best graphics card contender on January 30, as it's AI abilities make it coveted by entire industries. Still, it's my mission to help you grab the lavish GPU as soon as it hits virtual shelves, in turn transforming your PC into the ultimate powerhouse. I'm here to take you through where to buy the RTX 5090, and that starts with checking some reliable retailers. I check these specific stores every time a new GPU emerges on the battlefield, and while stock issues have plagued PC players in recent years, my goal is to make picking up a pre-order far easier than before. The RTX 5090 will undoubtedly change PC gaming when it launches January 30, but its $2,000 price tag means it absolutely isn't for everyone. If you're planning on putting together an absurdly powerful 4K rig that can also venture into 8K territory using the power of AI, you'll ultimately end up eyeing up this card. However, if even the RTX 4090 feels like more than enough juice to run your Steam library, there are far cheaper options this generation. During Nvidia's CES 2025 keynote, CEO Jensen Huang also unveiled the RTX 5080, RTX 5070 Ti, and RTX 5080. Each class of card should offer up frame rates that scale with price, but even the latter 70-series card can apparently provide Nvidia GeForce RTX 4090 performance. Better still, that particular model comes in at $549, so current gen 4K gaming is about to get a hell of a lot cheaper in the generation to come. Of course, there is a slight catch, as such feats are achieved using new AI tricks made possible by Blackwell. DLSS 4 will effectively give your PC a boost by generating graphics on its behalf, with Nvidia claiming the tool can fill in 33 million pixels for every 2 million generated by your actual hardware. If you'd prefer to keep things native, the results might be slightly less impressive, but you're still going to be getting a huge performance boost across the entire Blackwell range. It's still early days for the RTX 5090, and I'll be looking to share my own benchmarks soon. That'll ultimate paint a real performance picture that'll help you decide on whether to buy the flagship, but its price tag already suggests it's for power users and extreme PC builders alike.

Nvidia's Best New GeForce RTX 50-Series Graphics Card Costs $2,000 And Is An Absolute Power Hog

Nvidia's new GeForce RTX 50-series graphics cards have been officially revealed and they are beastly. The RTX 5090 will cost $2,000 and require a recommended power supply of 1kW, or roughly the equivalent of running a belt sander non-stop. The latest line of GPUs comes as Nvidia rushes to fill demand in the current AI rat-race. The new GPUs promising to take PC gaming rigs to the next level were revealed onstage at CES 2025 on Monday night by CEO Jensen Huang. You're probably familiar with DLSS, Nvidia's upscaling technology that squeezes even better resolutions at higher framerates out of existing computing resources, but the company is now also touting RTX Neural Shaders which will use machine learning to compress textures and free up even more processing power for other stuff. The 5080 and 5090 are both available later this month, while the 5070 and 5070 Ti will ship in February. Here are the prices: The big thing to note here is that Nvidia is promising big results even from the low-end budget card. All cards are supposed to be twice as fast on average as their corresponding predecessors, and Huang even claimed that this year's $550 RTX 5070 would be as effective in practice as the previous $1,600 RTX 4090, seemingly due to machine learning and other optimizations. At the high end, Nvidia showed a graph promising double the performance of the RTX 4090 from the newer (and only $400 more expensive) RTX 5090 across games like Alan Wake 2 and Black Myth: Wukong. The company even showed Cyberpunk 2077 running at over double the framerate with full raytracing enabled, because one can never have too many frames when navigating Night City's treacherous underworld. But while the 5090 is a power hog, it's not quite the hulking big boy that it's predecessor was. The 4090 founder's edition was a three-slot card that weighed over five pounds. The 5090 will only be two slots, allowing it to fit into more compact PC builds at launch.

Nvidia DLSS 4 key to RTX 50-series' claimed 2X performance boost

AI, more than new silicon, makes the difference against RTX 40-series Nvidia's CES 2025 presentation showed off its new RTX 50-series graphics cards proudly, with the RTX 5090 through 5070 all offering around double the performance of their previous-gen counterparts in the RTX 40-series. Nvidia credited this in its keynote to the brand-new Blackwell architecture. However, a look at Nvidia's in-house benchmarks reveals that 2x improvements in games come more from the software suite of DLSS 4 than pure silicon vs. silicon matchups. Deep Learning Super Sampling, or DLSS, has been a tentpole feature in Nvidia's consumer graphics cards since the 20-series. DLSS 4's software suite includes a bevy of new and improved AI features that increase in-game performance and visual quality over last-gen's DLSS 3/3.5. Software and hardware improvements lead to upgrades in DLSS's Super Resolution upscaling, ray reconstruction, and frame generation, a feature which AI generates one extra frame per rendered frame. DLSS 4 steps this up to Multi Frame Generation, a process which can generate three frames ahead of one rendered frame. This can effectively step a 30 FPS experience up to 120 FPS by generating and inserting the extra frames. The DLSS 4 suite is, per Nvidia's testing, better in every way than DLSS 3. Nvidia's own relative performance benchmarks for the RTX 5090, 5080, 5070 Ti, and 5070, which pin the new cards against their previous-generation counterparts (RTX 4090, etc.), seek to prove this point exactly. Both cards more than double last-gen performance in Cyberpunk 2077, Black Myth: Wukong, and Alan Wake 2 all in 4K Max settings and a D5 Render productivity benchmark. However, the fine print reveals that though some DLSS 4 features will come to 40-series and older cards, DLSS 3 was used on the 40-series cards in testing. There is no problem with testing in this way; DLSS 4 has not yet rolled out to older cards and Multi Frame Generation will only ever be feasible on 50-series cards thanks to its specific hardware needs. But the lead shrinks to a 1.5x boost or lower in games where DLSS 4 isn't yet available. A Plague Tale: Requiem was tested with DLSS 3 on all cards, while Far Cry 6 was tested with DLSS off completely and turned in a 1.25x generational improvement. Let those gamers who purchase the RTX 50 series remember to crank all DLSS 4 settings and features up to eleven to see true performance boosts beyond the 40 series. On release of the 50-series, beginning this month, some DLSS 4 upgrades will also roll down to older cards, raising their boats with the 50 Series. DLSS 4's Multi Frame Generation is also made interesting thanks to how it plays with input latency. Because of how Multi Frame Generation works (one frame is rendered, and the next three are generated based on extrapolating the previous one), some amount of input latency is introduced as some frames are generated without input data. Reflex 2, a technology that seeks to squash input lag, seeks to make up for Multi Frame Generation's extra required headroom. The appropriately named sequel to 2020's Reflex, Reflex 2 reduces latency in games by up to 75%, beating Reflex 1's 50% reduction. A standard latency pipeline follows mouse/keyboard inputs from the CPU, to the render queue, to the GPU, which then outputs an image around 58 milliseconds later. Reflex is an SDK that seeks to better synchronize the CPU and GPU, submitting updated input data from CPU to GPU just before it is needed, bypassing the render queue, and taking latency as low as 28ms. 90% of gamers keep Reflex on in its over 100 supported games, per Nvidia. Reflex 2 performs this improved queueing process even faster than Reflex 1, and adds a bonus layer to its pipeline. "Frame Warp" can take super-quick mouse movements and, rather than render the new angle, simply warp the existing frame in the direction of the input, further reducing input lag by removing the need for a full render. A predictive algorithm paints in the gaps in the new warped frame's edges, and voila; latency is further cut down to 14ms. If you're confused by this witchcraft, the simplified explanation is that Reflex 2 further improves input lag and can help overcome the added frame latency of DLSS 4's Multi Frame Gen for those on RTX 50 Series. A much deeper explanation can be found in Nvidia's Reflex 2 video below. Nvidia's RTX 50-series on Blackwell has been long awaited by consumers, and its twice-as-fast performance numbers over the RTX 40-series seem incredibly impressive. But while Blackwell may pack in double the memory bandwidth and double the ray-tracing cores over Ada silicon, the gap in real-world performance between the two generations seems to really come from DLSS 4's software-side improvements over the last gen. And of course, this is all based on Nvidia's own in-house benchmarks; once RTX 5090 and 5080 arrive for testing sometime this month, we will be able to more quantitatively support these numbers. Check back in for our full testing of the RTX 50-series once it launches sometime this month, and be sure to keep an eye on Tom's Hardware's CES 2025 coverage as the conference continues this week.

NVIDIA GeForce RTX 5090, RTX 5080, RTX 5070 Ti and RTX 5070 announced

NVIDIA today unveiled its latest line of consumer GPUs, the GeForce RTX 50 Series, available for both desktop and laptop platforms. Built on the new NVIDIA Blackwell architecture, these GPUs incorporate fifth-generation Tensor Cores and fourth-generation RT Cores, promising significant advancements in AI-driven rendering, including neural shaders, digital human technologies, and enhanced geometry and lighting. The GeForce RTX 5090, the most powerful of the series, features 92 billion transistors and delivers over 3,352 trillion AI operations per second (TOPS). NVIDIA states that innovations within the Blackwell architecture and the introduction of DLSS 4 enable the RTX 5090 to outperform the previous generation RTX 4090 by up to a factor of two. The Blackwell architecture extends to laptops with the RTX 50 Series Laptop GPUs, offering the same features as their desktop counterparts. NVIDIA's latest generation of Max-Q technology promises up to a 40% improvement in battery life, maintaining sleek laptop designs without compromising on power or performance. The new DLSS 4 introduces Multi Frame Generation, which leverages AI to generate up to three additional frames for each rendered frame. In conjunction with other DLSS technologies, this results in a performance increase of up to eight times compared to traditional rendering, while maintaining responsiveness through NVIDIA Reflex technology. DLSS 4 also incorporates a transformer model architecture for its Ray Reconstruction and Super Resolution models. This model uses double the parameters and four times the compute power of previous models, reportedly resulting in increased stability, reduced ghosting, greater detail, and improved anti-aliasing. Over 75 games and applications are expected to support DLSS 4 at launch NVIDIA Reflex 2 introduces Frame Warp, a technique designed to lower latency by updating rendered frames based on the most recent mouse input immediately before display. This can reduce latency by up to 75%, providing a potential advantage in competitive gaming and improving responsiveness in single-player titles. NVIDIA introduces RTX Neural Shaders with the RTX 50 Series, integrating small AI networks into programmable shaders. This is claimed to unlock more realistic materials, lighting, and other effects in real-time games. Rendering realistic digital humans is a key focus. RTX Neural Faces utilizes generative AI to render high-quality, temporally stable digital faces in real-time, taking rasterized face and 3D pose data as input. This is complemented by new RTX technologies for ray-traced hair and skin, and RTX Mega Geometry, which supports up to 100 times more ray-traced triangles per scene. The RTX 50 Series GPUs provide the AI TOPS necessary to power autonomous game characters alongside game rendering. NVIDIA's new suite of NVIDIA ACE technologies enables game characters to perceive, plan, and act similarly to human players. These ACE-powered characters are being integrated into games like KRAFTON's PUBG: BATTLEGROUNDS and InZOI, as well as Wemade Next's MIR5. NVIDIA is releasing a pipeline of NIM microservices and AI Blueprints for RTX AI PCs, sourced from developers like Black Forest Labs, Meta, Mistral, and Stability AI. These microservices cover a range of AI applications, including LLMs, vision language models, image generation, speech, and more. Project R2X was also previewed, demonstrating a vision-enabled PC avatar powered by NIM microservices. The RTX 50 Series GPUs are designed to enhance creative workflows. Notably, they are the first consumer GPUs to support FP4 precision, reportedly doubling AI image generation performance for models like FLUX and allowing generative AI models to run locally with reduced memory usage. The NVIDIA Broadcast app gains two AI-powered beta features for streamers: Studio Voice, which enhances microphone audio, and Virtual Key Light, which relights faces. Streamlabs is also introducing an Intelligent Streaming Assistant, powered by NVIDIA ACE and Inworld AI. The GeForce RTX 5090 and RTX 5080 will be available on January 30th, priced at USD 1,999 (Rs. 1,71,285 approx.) / Rs. 2,14,000 in India and USD 999 (Rs. 85,600 approx.) / Rs. 1,07,000 in India, respectively. The GeForce RTX 5070 Ti and RTX 5070 will be available in February, priced at USD 749 (Rs. 64,180 approx.) / Rs. 80,000 in India and USD 549 (Rs. 47,040 approx.) / Rs. 59,000 in India, respectively. NVIDIA Founders Editions will be available directly from nvidia.com and select retailers. Custom models will be available from various add-in card providers and system builders. Laptops featuring RTX 5090, RTX 5080, and RTX 5070 Ti GPUs will be available starting in March priced at USD 2899 (Rs. 2,48,435 approx.), USD 2199 (Rs. 1,88,445 approx.) and USD 1599 (Rs. 1,37,010 approx.), respectively, while RTX 5070 Laptop GPUs will be available in April price starting at USD 1299 (Rs. 1,11,305 approx.). These laptops will be available from manufacturers such as Acer, ASUS, Dell, GIGABYTE, HP, Lenovo, MECHREVO, MSI, and Razer.

Nvidia's New GeForce RTX 50-Series Graphics Cards are a Gift to Powermongers. Here's Why

There's just something about a new graphics processor architecture that gets everyone's juices flowing, either to covet it or complain. Or oddly, sometimes both. So happy CES 2025 from Nvidia! The company on Monday announced four 50-series desktop graphics cards, plus their laptop equivalents, using its new Blackwell architecture. That doesn't preclude an even lower-end model arriving later -- it's likely we'll see RTX 5060 and RTX 5050 options, at the very least mobile versions for entry-level gaming or creative laptops. The market for the latter has changed a bit this year, though, with graphics integrated into the CPU getting beefed up in order to handle AI and basic gaming. While Nvidia has been putting AI-focused Tensor cores in its cards since the RTX 20 series (Turing), this year it's kicked it up a notch because, well, AI is now The Thing and it's for a lot more than just DLSS (Nvidia's ML-based upscaling). Plus, the new fifth generation of Tensor cores bring new capabilities to DLSS over the 40-series' fourth-gen, and in conjunction with the new fourth-gen RT cores, improvements to render quality. Specifically: There's also the generally-faster-than-the-previous-generation angle thanks to more of everything and optimizations that occur from year to year. We won't really know specifics until we get a chance to test the cards. But Nvidia makes the usual claims that "this year's model performs like the higher-level of last year's." The 50-series imposes the usual year-over-year growing pains. Namely, they can be bigger and require more power. For instance, Nvidia specs the 5090 and 5080 as dual slot, but I've seen several much fatter listings; for example, the Asus ROG Astral GeForce RTX 5080 OC Edition lists as requiring 3.8 slots, which really means four. And even the midrange RTX 5070 models may hit the triple-slot mark. Plus, make sure your power supply's up for it. The RTX 5080 requires the same level as the RTX 4090; the 1,000-watt recommended PSU for the RTX 5090 may number among any records Nvidia has broken. The good thing about the laptop chips is that you don't have to worry about making sure anything fits or will run. On the other hand, do check to see how much power the manufacturer allows the GPU to pull. Because if you need the power of an RTX 5090, you don't want a system that will throttle it back when you need it most; you might as well save money on a lower class of card. Also, think about your motherboard and CPU (as always). The new cards support PCIe 5, and it's possible that if you stick it into a last-gen motherboard paired with a year's-old processor then one or both may bottleneck the performance.

These Are NVIDIA's New RTX 5000-Series GPUs

The RTX 4000-series was unveiled more than two years ago, and while the lineup is still holding up pretty well, we're already due to successors. Now, NVIDIA has just unveiled its RTX 5000-series, and it's one of the company's biggest generational improvements ever. ✕ Remove Ads NVIDIA has announced its next-generation GeForce RTX 5000-series GPUs, powered by the new Blackwell architecture. There's a total of four graphics cards for this initial announcement: the RTX 5070, the RTX 5070 Ti, the RTX 5080, and the RTX 5090. Out of these, the RTX 5070 and the RTX 5090 are the ones that will make the most headlines. The RTX 5070, the lowest-end card, is apparently just as powerful as the RTX 4090, at just a fraction of the cost. If this is true, it will likely quickly become a top option for gamers. The biggest and most powerful card is, as you might expect, the RTX 5090. According to NVIDIA, this one is two times as powerful as its predecessor, which is an insane generational improvement. This card won't be for everyone -- more on that later -- but if this claim holds up in real life (as usual, it's good practice not to take what a company says or claims as gospel and to wait for independent testing) it will be an impressive development. ✕ Remove Ads We're not getting just raw performance improvements, either. The RTX 5000-series has a few other party tricks down their sleeve. For starters, this generation introduces RTX Neural Shaders, bringing AI networks into programmable shaders to enhance real-time rendering of materials and lighting. RTX Neural Faces, along with technologies for ray-traced hair, skin, and Mega Geometry, aim to significantly improve the realism of game characters and environments. In other words, these cards use neural networks and AI in general to enhance the way they actually render games. We'd need to see how much of a game-changer this is, if it actually is one, but NVIDIA seems pretty confident on what it's doing here. ✕ Remove Ads NVIDIA also showed off DLSS 4, which is baked into this new generation of graphics cards. DLSS 4 can enable up to an 8x performance boost over traditional rendering methods, thanks to Multi Frame Generation, which uses AI to generate additional frames. For this, DLSS 4 is actually moving on to a transformer model for the first time. In case you don't know what a transformer model is, it's what powers generative AI and large language models in general -- a transformer model can weigh the importance of different parts of the input data when processing each element. This allows DLSS 4 to generate not only more frames but also higher-quality frames with fewer visual artifacts than before. DLSS 3 was already generating new frames from scratch using AI, and the fact that we have a transformer model here means that DLSS 4 can do it much more efficiently. This transformer also enhances DLSS's Super Resolution and Ray Reconstruction features, resulting in sharper images and more realistic lighting. Over 75 games and applications will be supporting DLSS 4 at launch, and more will come down the road. ✕ Remove Ads The RTX 5090 and RTX 5080 will be available on January 30, and the only bad part about them is the price -- the cards are priced at $1,999 and $999, respectively, which is actually a bit outrageous. Luckily, the lower-end ones, which are still great, are a bit more sensibly-priced. The RTX 5070 Ti and RTX 5070, which will follow in February, are priced at $749 and $549, respectively. $549 is actually a really good price for a card that performs just as good as the RTX 4090, mind you. Laptop versions of these GPUs will be available starting in March and April. Founders Editions and models from various manufacturers will be available worldwide -- individual manufacturers should be making announcements soon on their own cards, while you'll be able to buy the Founders Edition cards from NVIDIA's website. Source: NVIDIA

Nvidia announces next-gen RTX 5090 and RTX 5080 GPUs

Nvidia is officially announcing its RTX 50-series GPUs today. After months of leaks and rumors, the next-generation RTX Blackwell GPUs are now official and there are are four of them on the way. Nvidia CEO Jensen Huang revealed the RTX 50-series GPUs during a CES keynote this evening, announcing a $1,999 RTX 5090, a $999 RTX 5080, a $749 RTX 5070 Ti, and a $549 RTX 5070. The RTX 50-series GPUs will be available starting in January, and include a new design for the Founders Edition, with just two double flow through fans, a 3D vapor chamber, and GDDR7 memory. Huang demonstrated Nvidia's RTX Blackwell GPUs with a real time rendering demo at the beginning of the company's CES keynote today. The demo included new RTX Neural Materials, RTX Neural Faces, text to animation, and even DLSS 4. "The new generation of DLSS can generate beyond frames, it can predict the future," says Huang. "We used GeForce to enable AI and now AI is revolutionizing GeForce." Nvidia's RTX 50-series announcement comes more than two years after the RTX 4090 and RTX 4080 were announced, based on Nvidia's Ada Lovelace architecture. Nvidia's RTX 40-series of GPUs focused on improving ray tracing with Deep Learning Super Sampling (DLSS) version 3, and the RTX 4090 delivered some truly impressive performance gains over the previous RTX 3090 GPU.

Nvidia announces GeForce RTX 50-series graphics cards, starting with the 5070 for $549

Cameron Faulkner (he/him) is Polygon's commerce editor. He began writing about tech and gaming in 2013, and migrated from The Verge in 2023. After months of rumors, Nvidia debuted its latest generation of desktop and laptop graphics cards at CES 2025. The RTX 50-series is the fastest lineup of GPUs Nvidia has developed. Starting with its RTX 5070, Nvidia claims it delivers the equivalent performance to the previous generation's top model, the RTX 4090, at just $549 instead of the 4090's $1,599. Moving on up, the RTX 5070 Ti will cost $749, $999 for the RTX 5080, and $1,999 for the RTX 5090. All of these cards will be available starting later this month. The RTX 50-series marks the jump to GDDR7 video memory (VRAM), which is both more powerful and power efficient than the GDDR6x utilized in most RTX 40-series cards. Chief among the new features in the RTX 50-series cards -- aside from their horsepower increase -- is DLSS 4. It builds on Nvidia's previous iterations, all of which aim to increase gaming performance while making a series of visual trade-offs (i.e. lowering the render resolution, using AI to generate new frames to make games run smoother) that are becoming more and more difficult to spot as the tech improves. Nvidia touts that DLSS 4 taps into AI even more -- generating 3 frames with trained AI for every 1 rendered on the GPU -- to allow games to look better and run faster. Leaning so heavily on Tensor AI cores on the GPU is part of what makes Nvidia's RTX 50-series more efficient, so says Nvidia. It was a big day for new gaming GPUs, as AMD also announced some new hardware at CES. The company's new RDNA 4 graphics architecture and FSR4 upscaling and frame generation tech will debut in graphics cards coming soon, including the AMD Radeon RX 9070 XT and the RX 9070. Pricing and availability haven't been announced for these cards, though they'll likely compete directly against the GPUs in Nvidia's RTX 50-series lineup.

The Wait Is Over: Nvidia's Next-Gen RTX 50-Series GPUs Are Here

Expertise Laptops | Desktops | All-in-one PCs | Streaming devices | Streaming platforms Blackwell has arrived. At CES on Monday night, Nvidia unveiled its long-awaited next generation of GeForce RTX graphics cards based on its latest Blackwell microarchitecture, and the race will soon begin by AI outfits training large language models and PC gamers alike to gobble up these new cards as soon as they are released. Nvidia CEO Jensen Huang took the stage in Las Vegas and announced four new GeForce RTX desktop GPUs: the flagship RTX 5090 along with the RTX 5080, 5070 Ti and 5070. The GeForce RTX 5090 is twice as fast as the previous RTX 4090, according to Huang, thanks to the new Blackwell architecture and DLSS 4. It features 92 billion transistors capable of more than 3,352 trillion AI operations per second (TOPS). DLSS 4 introduces Multi Frame Generation that improves frame rates by using AI to generate up to three frames per rendered frame. DLSS 4 will be supported in over 75 games and applications the day of launch. And launch day is coming soon. The RTX 5090 and 5080 will start shipping on Jan. 30, and the 5070 Ti and 5070 will follow sometime in February. Here's the pricing: In addition to Nvidia's Founders Edition cards, versions using the GPUs will be available from Nvidia's usual partners, including Asus, Gigabyte, MSI, PNY and Zotac. Laptops with mobile versions of the RTX 5090, 5080, 5070 Ti and 5070 are slated to start shipping in March. In this age of AI, Huang focused on the AI TOPS figures of the new GPUs rather than CUDA cores. Here are the AI TOPS counts of the four new desktop GPUs: And AI is in the new graphics cards. In addition to DLSS 4, Nvidia's new RTX Neural Shaders uses AI to compress textures for greater memory efficiency, and RTX Neural Faces uses AI to improve the quality of characters' faces in games.

Nvidia GeForce RTX 50-Series Graphics Cards: What You Need to Know

There's just something about a new graphics processor architecture that gets everyone's juices flowing, either to covet it or complain. Or oddly, sometimes both. So happy CES from Nvidia! The company on Monday announced four 50-series desktop graphics cards, plus their laptop equivalents, using its new Blackwell architecture. That doesn't preclude an even lower-end model arriving later -- it's likely we'll see RTX 5060 and RTX 5050 options, at the very least mobile versions for entry-level gaming or creative laptops. The market for the latter has changed a bit this year, though, with graphics integrated into the CPU getting beefed up in order to handle AI and basic gaming. While Nvidia has been putting AI-focused Tensor cores in its cards since the RTX 20 series (Turing), this year it's kicked it up a notch because, well, AI is now The Thing and it's for a lot more than just DLSS (Nvidia's ML-based upscaling). Plus, the new fifth generation of Tensor cores bring new capabilities to DLSS over the 40-series' fourth-gen, and in conjunction with the new fourth-gen RT cores, improvements to render quality. Specifically: There's also the generally-faster-than-the-previous-generation angle thanks to more of everything and optimizations that occur from year to year. We won't really know specifics until we get a chance to test the cards. But Nvidia makes the usual claims that "this year's model performs like the higher-level of last year's." The 50-series imposes the usual year-over-year growing pains. Namely, they can be bigger and require more power. For instance, Nvidia specs the 5090 and 5080 as dual slot, but I've seen several much fatter listings; for example, the Asus ROG Astral GeForce RTX 5080 OC Edition lists as requiring 3.8 slots, which really means four. And even the midrange RTX 5070 models may hit the triple-slot mark. Plus, make sure your power supply's up for it. The RTX 5080 requires the same level as the RTX 4090; the 1,000-watt recommended PSU for the RTX 5090 may number among any records Nvidia has broken. The good thing about the laptop chips is that you don't have to worry about making sure anything fits or will run. On the other hand, do check to see how much power the manufacturer allows the GPU to pull. Because if you need the power of an RTX 5090, you don't want a system that will throttle it back when you need it most; you might as well save money on a lower class of card. Also, think about your motherboard and CPU (as always). The new cards support PCIe 5, and it's possible that if you stick it into a last-gen motherboard paired with a year's-old processor then one or both may bottleneck the performance.

Nvidia introduces RTX 5090, RTX 5080, and RTX 5070 laptop GPUs

Nvidia has announced the RTX 50-series (Blackwell) GPUs for laptops in parallel with its desktop family, promising a substantial lift in performance while preserving efficiency. Blackwell on mobile extends to the RTX 5090 laptop, featuring a massive 24GB VRAM, 50% more than the last generation. Retail availability is expected by March and April, and we're already seeing a compelling array of laptops showcased by partners as CES unfolds. Kicking things off with the RTX 5090 laptop GPU, Nvidia has packed this beast with 10,496 CUDA cores for 84 Streaming Multiprocessors, almost mirroring the desktop RTX 5080. Employing 3GB GDDR7 modules likely from Samsung, the RTX 5090 laptop GPU is packaged with 24GB of VRAM despite the 256-bit interface. Like all Blackwell, the revamped Tensor cores now support the FP4 datatype, with the RTX 5090 laptop GPU boasting 1,824 AI TOPS. TGP (Total Graphics Power) is paramount for GPU performance, and manufacturers can configure the RTX 5090 between 95W and 150W for every form factor and chassis. The RTX 5080 laptop GPU drops to 7,680 CUDA cores (60 SMs) and retains 16GB of memory as its predecessor. Despite the reduction in core count, the RTX 5080 still packs a punch, capable of delivering 1,334 TOPS of AI performance with an 80-150W TGP. The RTX 5070 family is divided into the RTX 5070 Ti laptop GPU with 5,888 CUDA cores and 12GB of memory and the base RTX 5070 laptop GPU, limited to 4,608 cores with just 8GB of memory. Nvidia showcased a slide during its presentation that detailed the price structure of these Blackwell mobile GPUs. Going by the data, the RTX 5090, RTX 5080, RTX 5070, and RTX 5070 Ti (all mobile) are set to cost $2,899, $2,199, $1,599, and $1,299 respectively. It's reasonable to assume this number represents the cost Nvidia's partners have to pay, so expect a higher price tag for retail RTX 50 laptops. With Blackwell, Nvidia has introduced a suite of changes to its Max-Q technology, promising up to 40% longer battery life. Let's go over them breifly. Advanced Power Gating turns off unused parts of the GPU when not needed, while Low Latency Sleep allows the GPU to go to sleep frequently, which, according to Nvidia, can save power even when the GPU is being used. Other features include the ability to switch clocks within microseconds and the inclusion of voltage-optimized GDDR7 memory featuring "ultra" low-voltage states. Performance is an important aspect, and so is efficiency since we're talking about laptops. Nvidia claims the (50-100W) RTX 5070 laptop GPU can match the RTX 4090 at half the power. Jensen showcased a desktop RTX 4090, so we can reasonably assume the desktop counterpart is the GPU used in the comparison. That's all fine and dandy, but wasn't the 250W RTX 5070 desktop on par with the RTX 4090? We'll have to wait for independent reviews, as Nvidia's first-party performance charts likely don't show the whole picture. GeForce RTX 5090, RTX 5080, and RTX 5070 Ti laptops will be available in March, with the RTX 5070 laptop GPU scheduled a month later for April. We can expect many models from Nvidia's partners, including Acer, Asus, Dell, Gigabyte, HP, Lenovo, Mechrevo, MSI, and Razer.

Nvidia Mobile GeForce RTX 5090, RTX 5080, RTX 5070 Ti, and RTX 5070 also announced

Nvidia introduced its latest series of laptop graphics cards, the GeForce RTX 5090, 5080, 5070 Ti, and 5070, during the CES keynote led by CEO Jensen Huang. These models, built on Nvidia's new Blackwell architecture, mark the next generation in mobile GPU technology. The announcement highlighted an array of technical advancements, including improved memory configurations and AI computational capabilities. Each card in the series has been engineered to balance performance, power efficiency, and support for emerging graphical and AI computing standards, making them suitable for high-end gaming laptops and professional workstations. The flagship model, GeForce RTX 5090, features 24GB of GDDR7 memory and delivers 1824 trillion operations per second (Tops) in AI computing power. This performance is facilitated by three ninth-generation Nvenc encoders and two sixth-generation Nvdec decoders, which are designed to accelerate video encoding and decoding workflows. In contrast, the RTX 5080 is equipped with 16GB of GDDR7 memory and achieves 1334 Tops, utilizing dual encoders for increased video processing efficiency. The mid-range RTX 5070 Ti and the entry-level RTX 5070 configurations offer 12GB and 8GB of GDDR7 memory, delivering 992 Tops and 798 Tops respectively. These models integrate advanced tensor and ray tracing cores, as well as Nvidia's DLSS 4 upscaling technology, to ensure consistent imaging quality and performance optimization across diverse applications. All models in the RTX 50 series incorporate fifth-generation Tensor Cores and fourth-generation Ray Tracing Cores. The technical specifications differentiate based on encoder and decoder configurations, with lower models featuring a single Nvenc encoder and an Nvdec decoder, while the RTX 5080 includes dual encoders. The standardized integration of DLSS 4 across the series and support for additional AI-based improvements showcase Nvidia's commitment to technical advancement and compatibility with modern software environments. While detailed pricing and availability have not been disclosed, the timeline indicates desktop versions of the RTX 5090 and RTX 5080 will launch on January 30, with the 5070 Ti and 5070 following in February.

NVIDIA Launches GeForce RTX 50 Series with Blackwell Architecture

At CES 2025, NVIDIA unveiled the GeForce RTX 50 Series, the latest GPUs developed for gamers, creators, and developers. Using the Blackwell architecture, the RTX 50 Series delivers improved performance and efficiency through AI-driven innovations like neural shaders and DLSS 4. The RTX 50 series, revealed during a CES keynote, includes the RTX 5090, 5080, 5070 Ti, and 5070. The GeForce RTX 5090 will be available starting January 30 for $1,999, while the other models are priced between $999 and $549, with launches scheduled for February. The flagship GeForce RTX 5090 GPU offers a significant leap in power, featuring 92 billion transistors and over 3,352 trillion AI operations per second (TOPS). According to NVIDIA, the RTX 5090 GPU doubles the performance of its predecessor, the RTX 4090. "Blackwell, the engine of AI, has arrived for PC gamers, developers, and creatives," said Jensen Huang, founder and CEO of NVIDIA. "Fusing AI-driven neural rendering and ray tracing, Blackwell is the most significant computer graphics innovation since we introduced programmable shading 25 years ago." The GPUs are equipped with fifth-generation Tensor Cores and fourth-generation RT Cores, enabling advanced capabilities such as RTX Neural Shaders and DLSS 4. The latter features Multi Frame Generation, which uses AI to generate up to three frames per rendered frame, boosting frame rates by up to 8x over traditional rendering. NVIDIA announced that the RTX 50 Series is now available for laptops, providing desktop-level performance in a portable form. With NVIDIA Max-Q technology, battery life is extended by up to 40% without compromising on design or performance. Laptops featuring RTX 50 Series GPUs will roll out from March. To improve realism in game characters and environments, NVIDIA introduced RTX Neural Faces and other AI-driven features. RTX Neural Faces uses generative AI to create high-quality, stable digital faces in real time, while RTX Mega Geometry allows for up to 100x more ray-traced triangles in scenes. The GPUs also power autonomous game characters through NVIDIA ACE technologies. These characters can perceive, plan, and act, creating dynamic interactions in games such as PUBG: BATTLEGROUNDS and the upcoming InZOI. RTX 50 Series GPUs support FP4 precision, optimising AI image generation and enabling generative AI models to run efficiently on local hardware. New tools, such as NVIDIA Broadcast's Studio Voice and Virtual Key Light, further enhance creative workflows.

Hands-on with DLSS 4 on Nvidia's new GeForce RTX 5080

Last night, Nvidia finally unveiled the RTX 50-generation graphics cards - but over the holiday period, I had the opportunity to spend some time testing out the brand new DLSS 4 upscaling and frame generation technology on a PC kitted out with the new GeForce RTX 5080, running an updated build of Cyberpunk 2077 - and it's impressive. Nvidia has upgraded its super resolution upscaling and ray reconstruction technology with a new 'vision transformer' model, offering some dramatic quality upgrades, while frame generation is boosted from one interpolated frame to two - or even three. The end result is a better-looking Cyberpunk 2077 capable of running the full path-traced experience at frame-rates well, well north of 120fps. There's a lot to discuss here but to be clear, this is preview, first look coverage. The RTX 5080 I had access to is an engineering sample. The drivers are not final. I can offer a broad idea of how DLSS 4 works and ballpark frame-rate increases, but exact numbers will have to wait for review hardware and final drivers. Also, the limited time window I had with the hardware limits the extent of the testing I could carry out - but regardless, I saw enough and captured enough to have an initial response to the new technology. The fruits of my visit are helpfully embedded below, but speaking of capture, just like DLSS 3 frame-gen before it, showing off how DLSS 4 actually looks is somewhat difficult, owing to the frame-rate amplification factor of the new frame generation system. Multi-frame generation makes DLSS 4 a good fit for the latest wave of QD-OLED 4K 240Hz monitors - but there is no capture technology on the market capable of capturing ultra HD at 240 frames per second. In the video b-roll, I tried to show the quality of frame generation by capping to 120fps, then slowing down by 50 percent to fit within the limits of YouTube's top-end 4K 60fps container. However, running unlocked, frame-rates are much higher than the 120fps limit I imposed. So while the video gives you an idea of how the new frame-gen presents, the real-world experience is something quite different: frame persistence will be significantly lower and therefore frame generation artefacts are far less noticeable. In effect, I needed to hold back DLSS 4 to give any kind of representative media on a video platform. DLSS 4 with full multi-frame-gen enabled is designed for the latest generation of high refresh rate displays and I experienced it on an Alienware AW3225QF QD-OLED 4K 240Hz monitor running fully unlocked - and it's quite the thing. Thankfully, improvements to DLSS super resolution and ray reconstruction can be shown with 4K 60fps capture - and there are key improvements here. The existing DLSS uses a convolutional neural network - Nvidia says this "generates new pixels by analysing localised context and tracking changes in those regions over successive frames". The model has improved over time, but it can only go so far and the latest releases have shown only iterative improvements. The new DLSS model uses a vision transformer, similar to the base technology behind the likes of ChatGPT, Gemini and Flux. Nvidia says this "enables self-attention operations to evaluate the relative importance of each pixel across the entire frame and over multiple frames." There are twice as many parameters as the prior CNN model and as a result, we're promised greater stability, reduced ghosting, higher detail and enhanced anti-aliasing. The transformer model is highly scalable, so Nvidia is predicting further improvements to quality with better training. It's also backwards compatible, meaning it can be retrofitted to prior DLSS titles running on the 2.0 version or higher. It'll also work on all prior RTX cards, going back to 2018's Turing-based RTX 20-series but the increased complexity of the model may have some kind of performance impact (possibly offset by the increase to quality - this really needs thorough testing). Even though the captured material we had was limited, the general impression that image quality specialist Alex Battaglia and I have from the b-roll is that despite running at the same 1080p internal resolution as the CNN performance mode, the vision transformer equivalent doesn't look like it - it's better. And there are some clear signs that some of the long-standing issues we've had with DLSS are much improved. This is best shown in the video above, but smearing and ghosting are drastically reduced. Shimmer in motion and unwanted 'stylisation' of content are mostly gone, Meanwhile, ray reconstruction - essentially upscaling for ray tracing - shows similar, highly welcome improvements. To compare the vision transformer model to the existing CNN version of DLSS is essentially to compare two very different AI upscalers - well outside of the scope of the hours I had with the RTX 5080 and something we'll be looking at once we've got hardware in-house. However, the implications of the super resolution and ray reconstruction improvements are significant. DLSS has enjoyed only iterative improvements since the arrival of the 2.0 version in 2020 - and yet it remains the highest quality upscaler around, while DLSS ray reconstruction has yet to be delivered by any of the competition. Assuming dramatic improvements with no downside are delivered on both counts, this takes Nvidia a further step ahead of its rivals. DLSS frame generation also gets new features - but these are tied exclusively to the new RTX 50-series cards. The single generated frame-gen tech used in 40-series is augmented with two and three frame generation in 50-series. This is tricky stuff, particularly when it comes to pacing those intermediate frames in a smooth, consistent manner - so Nvidia says that the new Blackwell architecture contains a hardware component for ensuring even pacing. I used the black market in the Petrochem stadium for testing - an area we've previously used for identifying a graphics bottleneck in the console version of Cyberpunk 2077. Running fully unlocked at 4K output resolution with the RT Overdrive path-tracing setting enabled, the existing single frame generation tech combined with the vision transformer model for super resolution and ray reconstruction saw a 535 percent frame-rate multiplier compared to the game running at native 4K resolution. The gains continue as we made more generated frames. Adding two frames sees that rise to 725 percent, while three frames predictably delivers the biggest increase of all - 913 percent vs native resolution rendering. Comparing the RTX 5080 with full frame generation and the transformer model enabled to the outgoing 4080 Super with single frame-gen and the existing CNN super resolution/ray reconstruction tech, I logged an improvement of 91 percent to frame-rate. Returning to frame-pacing, you'll see some visualisations of frame-times comparing single frame generation to full, three frame-gen output, both running on RTX 5080 - and it is reassuring to see that the latter not only means faster frame-rates, but also smoother frame-rates too. I took metrics here using a new version of Nvidia FrameView, but verified them with FCAT - where our software measures frame persistence based on the output of the GPU. Both results tallied. My concern with the new frame generation technique was latency. When DLSS 3 launched, frame generation was achieved by buffering an extra frame then calculating the intermediate one. Both of these factors add latency, which was offset to a degree by the mandatory inclusion of Nvidia Reflex to claw back precious milliseconds. In the video, you'll see that I've included PC latency results - again measured by FrameView. The averages across a circa two and a half minute run through the black market work out like this: a 50.97ms average latency with a single generated frame, rising to 55.5ms with two generated frames, and 57.3ms with the full, three-frame option. Based on this sample then, we're looking at a 6.4ms average addition to latency with maxed out frame generation in exchange for a 71 percent increase in frame-rate - it's an equitable trade overall. Looking at these results, it seems to me that the majority of the extra latency still comes from buffering that extra frame, but adding further intermediate frames comes with a relatively minimal increase in latency. That means that the gameplay is still responsive in Cyberpunk and unless you're super-attuned to input lag, you're unlikely to tell the difference with the existing DLSS frame generation solution. My time spent with the RTX 5080 was promising then. It's still quite remarkable to see a path-traced Cyberpunk 2077 running well in excess of 120 frames per second with the new frame generation option, so it's likely that less demanding games will run a whole lot faster. And then there's the fact that frame generation has proven quite useful in overcoming CPU limitations in many games - it'll be interesting to see the role multi frame-gen plays there. Meanwhile, the super resolution and ray reconstruction quality improvements look to address some of the key concerns we've had with DLSS, despite its status as the current highest quality upscaling solution. We'll be testing all components of DLSS 4 in much, much more depth as soon as we can, and - of course - we'll be reviewing all of the RTX 50-series GPUs in due course.

NVIDIA GeForce RTX 5090 24 GB, RTX 5080 16 GB, 5070 Ti 12 GB & 5070 8 GB Laptop GPUs Unveiled, 50-150W Max-Q For Mobile Gamers

NVIDIA Blackwell Enters The Gaming Laptop Segment With 24 GB GeForce RTX 5090, RTX 5080, 5070 Ti & 5070 GPUs NVIDIA's Blackwell AI GPU architecture is coming to high-end gaming laptops with the RTX 50 series lineup. These GPUs will offer expanded capabilities through the updated architecture while utilizing new upgrades to expand their efficiency. According to NVIDIA, the RTX 50 laptop GPUs will come in designs as thin as 14.9mm, which are made possible using new Max-Q innovations. One advantage comes through the use of GDDR7, which provides ultra-low voltage states that allow these laptops to run longer and also faster. In terms of specifications, the NVIDIA GeForce RTX 5090 Laptop GPU is going to be the flagship chip with a humongous 24 GB GDDR7 VRAM running across a 256-bit bus interface. This chip will offer a total of 10,496 CUDA cores in a 95-150W TGP package and offer up to 1824 AI TOPS. In addition to the flagship RTX 5090, NVIDIA is also going to offer RTX 5080, RTX 5070 Ti, and RTX 5070 GPUs for laptops. Starting with the RTX 5080, the chip gets 7680 CUDA cores with 16 GB of GDDR7 memory and an 80-150W TGP, the RTX 5070 Ti gets 5888 CUDA cores with 12 GB of GDDR7 memory and a 60-115W TGP, while the RTX 5070 gets 4608 CUDA cores with 8 GB of GDDR7 VRAM and a 50-100W TGP. There's no word on the performance, but given what we have seen on the desktop graphics cards, we can expect the performance to be doubled over the existing RTX 40 "Ada" lineup along with the support for DLSS 4. NVIDIA anticipates the launch of the first GeForce RTX 50-powered laptops in March 2025, so within Q1 2025. We can expect designs from leading partners such as Acer, ASUS, Dell, GIGABYTE, HP, Lenovo, MECHREVO, MSI, and Razer. NVIDIA GeForce RTX 50 "Blackwell" Laptop GPU Lineup (Official):

Nvidia RTX 50-series takes powerful laptop and desktop GPUs to the next generation

Blackwell is here, and this is everything we know so far about the upcoming Nvidia series Nvidia officially launched the GeForce RTX 5000 series graphics cards for both desktops and laptops at CES on Monday in Las Vegas. Nvidia's "Blackwell" generation GPUs have been highly anticipated since Computex in Taipei back in June of 2024. And finally, they're here. So, without further ado, here's everything we know about the new Nvidia GPUs so far: "Blackwell" was the codename of the latest generation of Nvidia graphics cards. Named after mathematician David Harold Blackwell, these chips are capable of raytraced gaming but were engineered with generative AI in mind. These massively powerful GPUs have been highly anticipated by gamers, content creators, and AI enthusiasts because of their raw graphics computing potential which can be used for next-gen gaming, video editing, and for AI workloads that hit the GPU. As far as laptop GPUs go, these are the confirmed RTX 50-series graphics cards: While the RTX 50-series graphics cards, at least the 5090 to 5070, are now official, that doesn't mean you can take a laptop home with the new Nvidia GPUs just yet. Gaming and content creator laptops with the new Nvidia GPUs are expected to launch later in March for systems with the RTX 5090, 5080, and 5070 TI. Laptops with the RTX 5070 will launch in April. So you can have a new Nvidia RTX 5090 laptop by spring. For folks looking for more budget-friendly GPUs like the RTX 5060 or 5050, those cards have not been confirmed for desktops or laptops. But, if Nvidia further expands the RTX 50-series GPU lineup, those GPUs would likely launch later this year.