Windows 11 lets users split memory for gaming and AI as RTX Spark launch nears

Windows could soon let you control how your PC splits memory for gaming and AI

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. Connecting the dots: PCs without a dedicated GPU already have to share a single memory pool between graphics, AI, and system tasks, and Windows currently makes that split on its own. That arrangement is becoming a much bigger deal as Apple, Qualcomm, Nvidia, and others build AI PCs around unified memory pools that could potentially stretch into the hundreds of gigabytes. A recent experimental Windows 11 build hides evidence that Microsoft is preparing to let users control the split between system and video memory in unified pools. The feature could give users more flexibility when juggling gaming and AI workloads, especially on upcoming unified-memory devices like Nvidia's RTX Spark platform. Windows 11 build 29648.1000, released August 17, carries a new feature called "IntelligentCarveout" that isn't mentioned anywhere in Microsoft's patch notes. Windows watcher @phantomofearth confirmed to Windows Latest that the build's code strings reference reserving memory for accelerators, graphics, and AI. That would let users manually set aside unified memory for games and AI-heavy applications, with that reserved portion off-limits to everything else. Similar references previously turned up in Nvidia documentation. If the feature ships publicly, it could fundamentally change how users configure performance on unified-memory chips. Unified memory has traditionally shown up in Macs, some laptops, handheld gaming PCs, and other systems built around integrated rather than discrete graphics. Depending on the workload, performance on these systems can vary significantly compared to discrete GPUs with dedicated VRAM, and while GPU-heavy tasks can pull from system memory when they run short, doing so usually comes with a steep performance penalty. Since running local AI models demands large amounts of fast memory, high-end GPUs once reserved mainly for gaming have become popular tools for AI work. Meanwhile, PCs built from the ground up for local models have featured progressively larger unified memory pools. Macs configured with 24GB of unified RAM or more have faced shortages amid the local AI boom, and Microsoft and Nvidia have answered that shift with RTX Spark devices offering up to 128GB. Nvidia's dominance in GPUs owes a lot to software, not just silicon - CUDA programming, DLSS upscaling, Reflex latency reduction, and the broader RTX suite have become real selling points on top of raw hardware. Carrying that same software stack into the integrated-graphics space gives RTX Spark a built-in edge over AMD, Qualcomm, and Intel, none of which can match Nvidia's software ecosystem on AI or gaming performance yet. When it unveiled the RTX Spark chip, Nvidia already touted it as carrying more addressable memory than any of its GPUs. Because IntelligentCarveout would let users route a large share of that 128GB pool toward graphics and AI tasks, RTX Spark's GPU could end up with access to far more memory than even the RTX 5090's 32GB of dedicated VRAM - though that shared, LPDDR5x-based pool is considerably slower than the RTX 5090's GDDR7, so it's a capacity advantage rather than a speed one.

Windows 11 Tests User-Controlled Unified Memory Reservations for GPUs and AI

Microsoft is experimenting with a Windows 11 feature that could allow users to determine how much shared system memory is explicitly reserved for graphics and AI acceleration. The feature is internally named IntelligentCarveout and has appeared in experimental Windows 11 build 29648.1000, released August 17. It is currently hidden and unavailable through the normal Settings interface. The concept is particularly relevant to systems where the CPU, GPU and NPU share one unified memory pool. Instead of letting Windows dynamically allocate the entire pool, users could potentially reserve a specific quantity for accelerator workloads. Memory assigned to this carve-out would no longer be available to ordinary applications. NVIDIA's new RTX Spark PC platform is an obvious use case. RTX Spark combines Arm CPU cores and Blackwell graphics with as much as 128 GB of unified memory. Microsoft previously modified Windows so substantially more system RAM could be exposed directly to the GPU for large AI workloads. This becomes increasingly important for local AI. A 70B or larger model can require tens of gigabytes of addressable memory. On systems with unified memory, a manually configurable GPU allocation could provide more predictable model-loading behavior than relying entirely on automatic Windows memory management. The same concept could eventually benefit AMD Strix Halo-class APUs and Intel's large integrated GPU platforms, although Microsoft has not identified support for those products. There are major caveats. Microsoft's Experimental Future Platforms channel is specifically intended for unfinished operating-system functionality. Features can be substantially changed or removed altogether before reaching retail Windows releases. So this is not yet a promised Windows 11 feature. Its existence is nevertheless notable because Windows historically hides most UMA framebuffer allocation behind firmware and driver logic. Giving the end user an OS-level memory-carveout control would represent a significant change as PCs increasingly blur the distinction between "system RAM" and "VRAM."

Windows 11 Is Getting A New Feature Where Users Can Allocate Memory For The GPU To Tackle AI Workloads, Likely To Prepare For RTX Spark's Imminent Launch

The idea of bringing unified memory controls to Windows 11 will bridge the gap when running those denser AI models, and according to the latest report, users can decide how much RAM to allocate. Apparently, Microsoft is working to bring a feature to its OS, enabling you to decide how much memory the GPU needs. With laptops powered by the RTX Spark just around the corner, this change will be necessary. The new Windows 11 feature is called "IntelligentCarveout" and is hidden in an experimental build With the RTX Spark offering support of up to 128GB of LPDDR5X unified memory, there are limited use cases for individuals to utilize nearly all of that just for system RAM. Instead, Windows Latest and VideoCardz report that a new feature called IntelligentCarveout has been spotted in a new experimental Windows 11 build on August 17, with an addition labeled "SettingsHandlers_UnifiedMemory.dll" Even though the feature is unreleased, the experimental Windows 11 build mentions that "Let Windows reserve additional unified memory for graphics and AI intensive games and applications. Reserved memory is not available for other applications." In short, the operating system lets users control how much unified memory the RTX Spark's Blackwell GPU uses for games and AI applications. To be fair, it isn't surprising that Microsoft is bringing this feature, especially seeing as how NVIDIA's new SoC is right around the corner. With the Windows 11 developer driver available, it's a sure sign that we'll see various notebooks featuring the RTX Spark before the end of the year. Then again, it's not the first time that Windows 11 is witnessing the arrival of a system with unified memory, but Microsoft is bringing a feature that will give users control. Well, that's because it didn't need to develop such a feature in the first place. AMD's Ryzen AI Max+ 395, with up to 128GB of memory, offered a similar feature via its graphics drivers, reserving up to 96GB for the GPU. There's no telling what the limit will be for allocating GPU memory on the RTX Spark, but during NVIDIA's announcement, the company mentioned that running 120B AI models was possible, hinting that a minimum of 8GB RAM will be reserved as system memory. Let us keep our fingers crossed for the RTX Spark launch, and we'll update our readers on the latest in the coming weeks. Follow Wccftech on Google to get more of our news coverage in your feeds.