Elektrine lite

← Feed

@gfxstrand@social.treehouse.systems

Post #2263753

2026-05-07 20:32 UTC

RE: https://floss.social/@collabora/116533723198138565 In case you missed, it, my blog post about memory optimization in NIR was published today.

Replies (3)

  • @aras@mastodon.gamedev.place 2026-05-08 03:50

    @gfxstrand@social.treehouse.systems nice post! Audience is probably dozens of us who are both into compilers and shaders, but this audience likes it very much :)

    Open ##2290659

  • @dotstdy@mastodon.social 2026-05-08 05:31

    @gfxstrand@social.treehouse.systems "In general, the approach is to always assume two derefs may alias and try to prove otherwise. If the proof fails, we have to assume the worst." :'( it was somewhat surprising to me how conservative nir / shader frontends are on this front, given that spir-v gives carte blanche to assume memory declarations don't alias unless they're explicitly marked with Aliased (as long as you're not in OpenCL mode anyway).

    Open ##2290664

  • @atmind@mstdn.social 2026-05-08 05:50

    @gfxstrand@social.treehouse.systems thanks for writing this up. I sporadic read the NIR codebase for investigating driver bugs. Now I know better how the code is structured.

    Open ##2290674