Monday, June 2nd 2025
Microsoft's Agility SDK Update Brings Shader Execution Reordering, Opacity Micromaps, and Tiled Resource Tier 4 Support
Microsoft has quietly rolled out two updates to its DirectX Agility SDK, and developers are already finding reasons to celebrate. The 1.717-preview release brings Shader Execution Reordering (SER) and Cooperative Vectors to the table, while the 1.616-retail update introduces Opacity Micromaps (OMM) and D3D12 Tiled Resource Tier 4 support. Together, they promise smoother and more efficient ray tracing, as well as improved resource handling. In 1.717-preview, the arrival of Cooperative Vectors means that vector and matrix operations, essential for AI-driven effects, now enjoy dedicated hardware acceleration. Instead of performing complex calculations on the CPU, developers can offload these tasks to the GPU, potentially unlocking richer neural rendering techniques in real time. Additionally, SER provides ray tracing pipelines with a helpful feature: it reorganizes the order in which shader threads execute, reducing wasted cycles caused by thread divergence.
Early internal tests show up to a 2× performance boost in path-traced scenes when SER is enabled. The preview also updates Direct3D's video encoder with a HEVC Reference List extension for longer-term reference frames, a two-pass low-resolution first encode, and built-in PSNR metrics so creators can see exactly how crisp their output really is. The 1.616-retail SDK centers on Opacity Micromaps, a feature designed to handle alpha-tested geometry, such as foliage or chain-link fences, without firing off expensive AnyHit shader calls. With OMM, supported GPUs intelligently skip unnecessary work, and Microsoft's numbers suggest a speedup of as much as 2.3x in path-traced scenes. For example, one NVIDIA demo increased from approximately 55 FPS to around 90 FPS simply by enabling OMM. Today, only NVIDIA's RTX cards have driver-level support for these micromaps, but AMD and Intel are preparing their own updates for the coming months. That same 1.616-retail release also unlocks D3D12 Tiled Resource Tier 4, finally lifting the restriction on packed mipmaps in texture arrays. This means finer-grained texture streaming, less wasted memory, and more creative freedom in writing shader code. NVIDIA has confirmed its support for this tier, Intel's drivers are already shipping, and AMD is aiming to release compatible drivers by early June 2025.
Sources:
Microsoft, via Wccftech
Early internal tests show up to a 2× performance boost in path-traced scenes when SER is enabled. The preview also updates Direct3D's video encoder with a HEVC Reference List extension for longer-term reference frames, a two-pass low-resolution first encode, and built-in PSNR metrics so creators can see exactly how crisp their output really is. The 1.616-retail SDK centers on Opacity Micromaps, a feature designed to handle alpha-tested geometry, such as foliage or chain-link fences, without firing off expensive AnyHit shader calls. With OMM, supported GPUs intelligently skip unnecessary work, and Microsoft's numbers suggest a speedup of as much as 2.3x in path-traced scenes. For example, one NVIDIA demo increased from approximately 55 FPS to around 90 FPS simply by enabling OMM. Today, only NVIDIA's RTX cards have driver-level support for these micromaps, but AMD and Intel are preparing their own updates for the coming months. That same 1.616-retail release also unlocks D3D12 Tiled Resource Tier 4, finally lifting the restriction on packed mipmaps in texture arrays. This means finer-grained texture streaming, less wasted memory, and more creative freedom in writing shader code. NVIDIA has confirmed its support for this tier, Intel's drivers are already shipping, and AMD is aiming to release compatible drivers by early June 2025.
24 Comments on Microsoft's Agility SDK Update Brings Shader Execution Reordering, Opacity Micromaps, and Tiled Resource Tier 4 Support
Cyberpunk will probably be the first game to benefit, if I understand the video right, programmers will have to do the implementation from their side, it's not a driver level thing.
I wouldn’t be surprised if that’s the case, MS might have shared the preliminary SDK with all GPU vendors and NV could have jumped the gun. Pair that with the drivers being made to support a new architecture (and Blackwell IS new, all memes of “repackaged Ada” aside) and you can conceivably get the shitshow that was the 57x branch.
I would be very surprised if a game like cyberpunk 2077, in path tracing get any performance improvements at all from this, for Nvidia of course, it's already documented it was implemented in cyberpunk.
So this is a major setback for Nvidia as everyone will have it now
I'm sorry to disappoint but current titles are already optimized on Nvidia, there is no benefit for them.
I'm not well versed here, but it seems like it's simply going back to the way DirectX used to be in the past.
A library that can be used on multiple versions of Windows.
devblogs.microsoft.com/directx/gettingstarted-dx12agility/:
The only title known to use it is Alan Wake. AFAIA.
I'm done getting hyped over this kind of stuff. We've seen this cycle too many times with new feature drops, esp. when it takes months (or years) before games actually implement it properly, if they even do. And when things break or underperform, the blame ping pongs between the GPU drivers, Windows version, game devs or whatever engine they’re using. Even when you do have the right hardware, OS and drivers, you’re still at the mercy of game implementation.
Another round of wait and see (keeping expectations sensible)
But it has been available via NVAPI, and it’s not like AMD cards have been capable of path tracing.
Edit. Oh, it’s not just software. You have to have the hardware to run the software.
As i said, there is no benefit for Nvidia, they did the work already, it just wasn't in DX12 specs, now it is and unfortunately AMD and Intel can benefit from it.
They're gaining ground now with RDNA4 producing excellent results. I'm sure they knew about SER at high and low levels, but they just couldn't make it function with what they had. Now with this SDK update they should be able to take advantage of their hardware capabilities a lot better.
AMD can do it. They can. The problem is their entire concious is about their CPU's (Which is fair enough considering all the nonsense Intel kicked up thinking they will always be unbeatable), so they probably won't do it until they fall behind as usual, or, maybe somehow they shift up a gear and not miss an opportunity... As usual
Scum aside, at least nvidia is pushing the boundaries of what games and hardware can do. No bad products, just bad execution and bad practices.
videocardz.com/newz/geforce-rtx-5060-ti-appears-on-steam-hardware-survey-list-before-radeon-rx-9070-series
NVIDIA's mindshare dominance with the RTX 5060 Ti sales dominance despite the RX 9700 series. If AMD pulls discrete Radeon GPU products from the marketplace, NVIDIA's mindshare wouldn't even miss it.
PS: I'm guilty of supporting NVIDIA mindshare, and I plan to repeat it with RTX 5080 Super.SER support is a required feature in Shader Model 6.9. DXR Tier 1.2 needs Opacity Micro-maps (OMM) and Shader Execution Reordering (SER).
Intel: Support for SER is now available for Intel® Arc™ B-Series Graphics and Intel® Core™ Ultra Processors (Series 2) with the Intel® Arc™ Graphics developer preview driver available www.intel.com/content/www/us/en/download/737144/855958/intel-graphics-preview-features-windows.html
Intel Lunar Lake's iGPU has SER.
To be fair, AMD is too busy with the PlayStation PS5 Pro APU.