projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 14818a8) | patch
AMDGPU: Correct const_index_stride for wave 32 for PAL ABI
authorDavid Stuttard <david.stuttard@amd.com>
Mon, 24 Feb 2020 21:19:15 +0000 (21:19 +0000)
committerDavid Stuttard <david.stuttard@amd.com>
Fri, 7 May 2021 12:42:57 +0000 (13:42 +0100)
commit606d4e806192013ff7da33351f671d08b4524438
tree1fe2507f637d7ff1d655d3936ed0d5979c9eb75atree | snapshot
parent14818a86d044909d8eeb1f39f689e2785a09823bcommit | diff
AMDGPU: Correct const_index_stride for wave 32 for PAL ABI

Retrying after revert and fix (removed implicit def flag from operand). Now
passes with expensive_checks enabled.

Since there is a single scratch resource descriptor for all shaders, if there is
a wave32 and a wave64 shader (for instance for VsFs pairs)
then the const_index_stride will be incorrect for wave32 shaders.

Differential Revision: https://reviews.llvm.org/D101830

Change-Id: Ie3b8b2921237968caca91527dd0c97b1b0cc0360
llvm/lib/Target/AMDGPU/SIFrameLowering.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/pal-simple-indirect-call.ll diff | blob | history
Domain: System / Toolchain;