I will also be trying bluesky for a bit, there is a nice little graphics programmer community building up there, you can find me at

RT @davidspringate: Didn't think I'd have to say this again so soon.. I'm sorry to all the staff at Unity and Crytek who have been affecte…

This year the "new" DirectX 12 graphics API, becomes 10 years old (Vulkan becomes 9). Feeling old yet?

Good read with a great collection of screen space effects

Game engines and shader stuttering: Unreal Engine's solution to the problem

"Programming Parallel Computers" Aalto University course lecture notes

"Raymarching The Gunk" discussing the techniques used to render the slimy substance in the game, great read:

Nice video discussing, at a high level, how (NVidia) GPUs work with great visualisations:

AltDevBlogADay was a very interesting site that hosted posts from industry professionals, covering a wide range of topics. Unfortunately it went down about 10y ago and is accessible only via web archive Definitely worth browsing, many still relevant posts

While the long branch unnecessarily increased VGPRs, the compiler took advantage of it to batch many tex reads up front and cache the results in VGPRs. Without the inactive branch it serialised access issuing a tex load, wait, use result and then issue another, to save VGPRs. 2/2

The Nvidia Blackwell GPU architecture paper is now available online

Adding the precise modifier to a VS output will stop the compiler from doing any code refactoring for it, eg convert "a*b+c" to a mad: This can be good for precision sensitive output like motion vectors or to ensure z-pre and g-buffer pass depths match.

RT @davidspringate: Really sorry to hear about the loss of jobs at Ubisoft Leamington. :( That really sucks. If you're affected, please, r…

Mapping of logical shader stages to hardware shader stages for AMD's GCN and RDNA architectures:

Fantastic work! "After replacing acos() with an approximation (and seeing a nice performance boost) I realized that the shader could still be vastly improved and ended up removing acos() altogether." (from

@clabaultt Good catch, I'll rework the example, thanks!

The CUDA programming guide is worth studying, even if you're not programming in CUDA. A lot of this knowledge is transferable to DX12/Vulkan compute shaders, especially the GPU architecture/best practices/performance guidelines parts.

This is the amount of shader ISA code the compiler produces for a single atan2(). Avoid using it for time critical code.