Microsoft Rolls Out Shader Model 6.9, DXR 1.2 & DX12 Enhancements In Agility SDK 1.619, NVIDIA With Widest Support

Microsoft has just rolled out its Agility SDK 1.619 with major DX12 improvements such as Shader Model 6.9 & DXR 1.2.

Microsoft Agility SDK Adds Shader Model 6.9, DXR 1.2, & Various DX12 Improvements: NVIDIA RTX, AMD Radeon & Intel Arc GPUs Ready

In the latest Agility SDK 1.619 release, Microsoft is making three updates: Shader Model 6.9, DXR 1.2, and DX12-related improvements. The features are listed below:

• Shader Model 6.9 (via DXC 1.9.2602.16):
  • Long Vector
  • 16-bit float Specials
  • 16-bit and 64-bit shader ops and wave ops are now required features
  • HLSL exposure of the following in DXR 1.2:
• DXR 1.2:
  • Opacity Micromaps (Released earlier, with a small SM 6.9 portion lit up now)
  • Shader Execution Reordering
• D3D Customer-Requested Features:
  • Revised Resource View Creation APIs
  • Periodic Trim Notifications
  • Increased Dispatch Grid Limit
  • CPU Timeline Query Resolves

The biggest part of this Agility SDK release is the introduction of Shader Model 6.9, which adds four new features: Long Vector, 16-bit float Specials, 16-bit / 64-bit shader op & wave ops, and HLSL exposure for DXR 1.2. With Long Vectors, HLSL can load, store, and perform HLSL vectors longer than 4 elements & up to 1024 elements. 16-bit and 64-bit shader ops and wave ops are also a required feature now with Shader Model 6.9.

Moving over to DXR (DirectX Ray Tracing) updates, the latest features include Opacity Micromaps, which were released earlier, and SER (Shader Execution Reordering).

For D3D12 Opacity Micromaps, or OMMs in short, Microsoft states that the new feature will enable the hardware to handle alpha-tested geometry much more efficiently than relying on costly AnyHit shared invocations.

In previous examples, Microsoft has touted up to 2.3x performance gains in path-traced titles for games that utilize OMM. One of the demos highlighted by NVIDIA shows over 60% gains. On the left, you can see the reference scene at 55 FPS, and the frame on the right has Opacity Micro-maps enabled, resulting in 90 FPS. Currently, only NVIDIA offers driver support for OMM for its RTX GPUs, while support from other vendors is coming in the future.

Alan Wake, from Remedy, is another game that relies heavily on alpha-tested geometry. A single scene can have anywhere from 4.5M + 9.3M triangles in total, with 5.2M skinned vertices and 2.2K skinned instances. The game is also one of the most intensive visual titles that utilizes path tracing, and with ray tracing set on high, the game can generate 10 rays per pixel or 36.9M per frame. An RTX 4090 without any optimizations takes 16.8ms on average to render a scene, but with SER and OMM, the frame can be rendered in just 10.2ms.

Opacity Micromaps

Opacity Micromaps (OMMs) enable hardware to handle alpha tested geometry more efficiently than relying only on costly AnyHit shader invocations. The overall feature shipped previously and just the small HLSL portion is coming out of preview. 

Shader Execution Reordering

Shader Execution Reordering (SER) enables application shader code to inform hardware on how to find coherency across rays so they can be sorted for better parallel execution. 

This feature is coming out of preview now.  The addition since preview is that apps can query if a device actually reorders.

via Microsoft

Lastly, Microsoft's Agility SDK 1.619 is adding some nice D3D enhancements for its DirectX 12 Ultimate API. The updates address the previous limitations in buffer views, which needed to evolve with the advances in GPU architectures. With the new update, buffer views can now be measured using byte offsets & sizes.
Revised Resource View Creation APIs

As GPU architectures have evolved, the original D3D12 view‑creation model has shown limitations, especially around buffer access patterns, descriptor management, and alignment rules. Revised View Creation modernizes this area of the API in response to multiple customer requests.

Previously, buffer views were limited to being measured in elements. With this change, they can now be measured using byte offsets and sizes too. In addition, variants of view creation have been added that return HRESULT rather than void to allow for programmatic error handling, as opposed to relying on the debug layer validation and dealing with a device removal.

Periodic Trim Notifications

Kernel-level trim notifications are now available through D3D12 runtime interfaces, enabling applications to receive notifications when the system should trim residency. No new driver support is required for this feature.

Increased 1D Dispatch Limit

Increases the maximum 1-Dimensional Dispatch/DispatchMesh size (Currently 65535) to a device specific value which is much larger for most recent hardware.

• D3D12_FEATURE_DATA_D3D12_OPTIONS22.Max1DDispatchSize
• D3D12_FEATURE_DATA_D3D12_OPTIONS22.Max1DDispatchMeshSize

CPU Timeline Query Resolves

A new kind of Query Heap which can be resolved on the CPU timeline, avoiding unnecessary GPU work and overhead. Introduces ID3D12Device15::CreateQueryHeap1 along with ID3D12Device15::ResolveQueryData.

via Microsoft

GPU Vendor Hardware / Software Support for Microsoft Agility SDK 1.619

As for support, all three GPU vendors have support ready for Microsoft's Agility SDK 1.619 and its latest features. The level of support varies from architecture to architecture.

Hardware Support

IHV	Driver Link(s)
AMD	AMD Software: Adrenalin Edition 26.2.1 AMD Software: AgilitySDK Developer Preview Edition 25.30.21.01
Intel	Intel Arc Graphics – Windows
NVIDIA	Download The Official NVIDIA Drivers | NVIDIA or the NVIDIA App, which handles automatic driver updates. Driver version 595 and newer.

For NVIDIA, the GPU maker lists that all of its RTX GPUs support the latest features. For DXR1.2, OMM are hardware accelerated on RTX 40 GPUs and above, while SER is supported on RTX 40 GPUs and above. The OMM support is limited to software emulation on older hardware.

As for AMD, they list support for RX 7000, RX 900, and Ryzen AI 300/400 GPUs, essentially indicating support for RDNA 3, RDNA 3.5, and RDNA 4 architectures. The RDNA 3 and RDNA 3.5 architectures carry D3D enhancements, while DXR 1.2 and Shader Model 6.9 features are limited to RDNA 4 (Radeon RX 9000) GPUs.

Lastly, Intel lists support for Arc B-Series (Battlemage) for all features, while certain D3D features, such as VPblit 3DLUT is applicable to Lunar Lake CPUs too.

Interestingly, OMMs are only supported by RTX 40 GPUs and above, while SER support shows reordering on NVIDIA RTX 40+ GPUs, Arc B-Series GPUs, but doesn't feature reordering on RX 9000 GPUs.

	AMD	Intel	NVIDIA
Long Vector	AMD Radeon RX 9000 series	Intel Arc B-Series Graphics	All RTX hardware
16 bit float Specials	AMD Radeon RX 9000 series	Intel Arc B-Series Graphics	All RTX hardware
Opacity Micromaps (OMM)	N/A	N/A	All RTX hardware. Hardware-accelerated on RTX 4xxx+ GPUs, software-emulated on older.
Shader Execution Reordering (SER)	AMD Radeon RX 9000 series supports API but doesn’t reorder.	RTX 4xxx+ GPUs support the API and do reordering.	Intel Arc B-Series Graphics. Existing 64k limit, to increase for future drivers.
Revised Resource View Creation APIs	AMD Radeon RX 7000 and 9000 series	Intel Arc B-Series Graphics	All RTX hardware
Periodic Trim Notifications	N/A	Intel® Arc B-Series Graphics	All RTX hardware
Increased Dispatch Grid Limit	AMD Radeon RX 7000 and 9000 series. UINT_MAX compute, 64k mesh.	Intel Arc B-Series Graphics. Existing 64k limit, to increase in future drivers.	All RTX hardware. Existing 64k limit, to increase in future drivers.
CPU Timeline Query Resolves	AMD Radeon RX 7000 and 9000 series	Intel Arc B-Series Graphics	All RTX hardware
Fence Barriers (preview)	AMD Radeon RX 7000 and 9000 series	Intel Arc B-Series Graphics	Contact your developer relations representative for in-development driver access.
VPblit 3DLUT (preview)	AMD Radeon RX 7000 series graphics cards and Ryzen AI 300/400 series processors with integrated graphics	Intel Core Ultra processor family Lunar Lake and Panther Lake platforms	Contact your developer relations representative for specifics.

Follow Wccftech on Google to get more of our news coverage in your feeds.