07:57 vedm: agd5f_: from my quick grep accross the drm and amdgpu, it doesn't seem to be hardcoded either, but dynamically detected
13:27 MrCooper: mareko: 7e959864b20bd961188cf78f6026a70baac3a824 ("radeonsi: enable NGG culling for non-monolithic TES and GS") broke lots of piglit & GLES CTS tests on Navi 14: ../src/gallium/drivers/radeonsi/si_state_shaders.cpp:1369: unsigned int si_get_input_prim(const si_shader_selector*, const si_shader_key*, bool): Assertion `gs->stage == MESA_SHADER_VERTEX' failed.
13:36 Venemo: weird, why didn't Mesa CI pick that up?
13:42 daniels: Venemo: there's no navi14 coverage
13:43 Venemo: sure, but it makes no sense why that change would even affect navi 14
13:44 daniels: there is navi21 coverage for gfx10 coverage, but that only runs vkcts
14:34 glehmann: Venemo: isn't navi14 special because ngg is disabled?
14:35 Venemo: yes, ngg is disabled, so it should behave like gfx6-9