Skip to content

Add Panthor compute-shader test

Daniel Stone igt-dev 1 version Applied

Status from this message: Applied all 1...4, thanks!

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest4 patches 10 repliesApplied

Patches in v1 Applied

#Subject and filesSizeRepliesApplied
1/4panthor: Use all available GPU coreslib/igt_panthor.c lib/igt_panthor.h+7 -234 replies
2/4panthor: Add igt_panthor_get_arch() helperlib/igt_panthor.c lib/igt_panthor.h+18 -02 replies
3/4panthor: Add more CSF instructionslib/igt_panthor.h+126 -02 replies
4/4panthor: Add simple compute-shader testlib/igt_panthor.h tests/panthor/panthor_group.c+444 -02 replies

Reviews trailers per patch

All 4 patches have a Reviewed-by or Acked-by.

PatchKamil KoniecznyLiviu Dudau
1/4A-b R-b
2/4A-b R-b
3/4A-b R-b
4/4A-b R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/4] Add Panthor compute-shader test

Hi,
This series adds a test to the Panthor suite to exercise a simple compute shader. Whilst the existing group_submit test does validate that we're able to execute CSF commands, it doesn't validate that we can use this to actually run any shaders. This series should help as minimal bringup and validation, inspired by a platform which was able to pass group_submit just fine, but would immediately wedge the firmware when trying to run a compute-shader test.

I don't plan to extend this much further - at least not with more complex GPU command types or shader ISA, since that would require a lot more divergent per-arch code to deal with changes in descriptor tables and commands - but it does ensure that we can actually schedule and execute jobs on a GPU.

Tested on v10 (RK3588), v13 (MT8196), and v14 (Lumex CSS FVP).

Cheers,
Daniel