Microsoft DirectX Advances Gaming with AI, Faster Loads & DXR 2.0
▼ Summary
– Microsoft announced ML-powered DirectX updates, including DirectX Linear Algebra for matrix-based ML workflows and a DirectX Compute Graph Compiler for executing full ML models with native GPU performance.
– The company also introduced Advanced Shader Delivery for Windows, a system allowing gamers to download pre-compiled shaders to eliminate stutters and reduce load times, which hardware vendors like NVIDIA, AMD, and Intel are supporting.
– Major GPU vendors (AMD, Intel, NVIDIA, and Qualcomm) expressed support for the new DirectX ML features, highlighting benefits like developer control, portable performance, and seamless integration of ML into graphics pipelines.
– Microsoft is preparing DXR 2.0, the next version of DirectX Raytracing, which will require support for Shader Model 6.10 and Opacity Micromaps, with a preview expected in late summer 2026.
– The ML features, including DirectX Linear Algebra and the Compute Graph Compiler, are set for public and private previews starting in April and summer 2026, respectively, while Advanced Shader Delivery is expanding to PC platforms like NVIDIA GeForce RTX.
The future of PC gaming is set for a significant leap forward with Microsoft’s latest DirectX announcements, focusing on artificial intelligence integration, dramatically reduced load times, and next-generation ray tracing. Revealed at GDC 2026, these updates promise to reshape how developers create games and how players experience them, tackling long-standing performance hurdles head-on.
A core theme is the deep integration of machine learning directly into the graphics pipeline. This begins with Cooperative Vectors in Shader Model 6.9, already part of the Agility SDK, which enables advanced neural rendering techniques like Neural Texture Compression. To handle the complex matrix operations these ML workloads demand, Microsoft is introducing DirectX Linear Algebra. This new component provides developers with a unified programming model for both vector and matrix-based ML, granting precise control over math, data flow, and execution within shaders.
Complementing this is the DirectX Compute Graph Compiler, a new compiler API designed to execute full machine learning model graphs with native GPU performance. Its advantages are substantial, offering unified tooling with PIX for debugging, the ability to drop complete models into game engines without rewriting shaders, and automatic optimization for portable performance across different hardware vendors. Major industry players have voiced strong support. AMD highlights the new ways developers can integrate ML while retaining expected control and performance. Intel is excited to support Linear Algebra from day one, and NVIDIA sees flexible paths for integrating both shader-level and model-level ML seamlessly. Qualcomm notes the compiler is a meaningful step toward making full-model ML feel native inside real-time engines.
Beyond AI, Microsoft is addressing a perennial PC gaming annoyance: stutter and long load times caused by shader compilation. The solution is Advanced Shader Delivery for Windows, which allows game developers to provide fully compiled shaders for specific PC hardware in advance. This system, already proven on Xbox handhelds, relies on two new APIs in the Agility SDK. The App Identity API lets applications declare a standardized identity to graphics drivers, while the Stats API gives developers visibility into how well a precompiled shader database performs for specific hardware configurations. To enable this feature, developers must integrate and submit the necessary data through the Xbox Partner Center.
The industry response has been enthusiastic. AMD calls the results on handheld devices “truly remarkable,” while Intel states it is a critical step toward improving the overall gaming experience. NVIDIA confirms it is working to launch support for GeForce RTX users to eliminate shader-related stutters, and Qualcomm plans to debut the feature on its upcoming Adreno X2 GPUs.
Looking further ahead, Microsoft is already preparing DXR 2.0, the next major version of its DirectX Raytracing API. Full compliance will require support for Opacity Micromaps and Shader Model 6.10. The update introduces a tiered system; Tier 1.2 offers partial support for hardware lacking Opacity Micromap capabilities but retaining Shader Execution Reordering (SER). Shader Model 6.10 itself brings new features, including required support for TriangleObjectPositions. This structured approach allows hardware to support new capabilities without necessarily meeting every requirement of the top-tier DXR 2.0 specification, ensuring broader compatibility. The initial preview for these ray tracing advancements is currently slated for late summer 2026.
Together, these initiatives represent a coordinated push to solve key technical challenges. By weaving machine learning directly into the fabric of DirectX, eliminating shader compilation bottlenecks, and evolving ray tracing with clear hardware tiers, Microsoft is laying a foundation for more immersive, responsive, and visually stunning games on the Windows platform for years to come.
(Source: wccftech)
