[AMDGPU] Avoid inserting noops during scheduling
authorAustin Kerbow <Austin.Kerbow@amd.com>
Mon, 19 Oct 2020 23:54:24 +0000 (16:54 -0700)
committerAustin Kerbow <Austin.Kerbow@amd.com>
Wed, 21 Oct 2020 00:11:36 +0000 (17:11 -0700)
commitebdcef20ce2921d1b6e8463ecf4031396cef79dc
tree2cb5049f6ee5d4431e6d8ecc1a2190343d91a5a2
parent37d907899f498c6944fc234ed2742b9923b88952
[AMDGPU] Avoid inserting noops during scheduling

Passes that are run after the post-RA scheduler may insert instructions like
waitcnt which eliminate the need for certain noops. After this patch the
scheduler is still aware of possible latency from hazards but noops will
not be inserted until the dedicated hazard recognizer pass is run.

Depends on D89753.

Reviewed By: foad

Differential Revision: https://reviews.llvm.org/D89754
12 files changed:
llvm/lib/Target/AMDGPU/GCNHazardRecognizer.cpp
llvm/lib/Target/AMDGPU/SIShrinkInstructions.cpp
llvm/test/CodeGen/AMDGPU/GlobalISel/extractelement.i128.ll
llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.fmas.ll
llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.update.dpp.ll
llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
llvm/test/CodeGen/AMDGPU/atomic_optimizations_pixelshader.ll
llvm/test/CodeGen/AMDGPU/global-saddr-atomics.ll
llvm/test/CodeGen/AMDGPU/llvm.amdgcn.update.dpp.ll
llvm/test/CodeGen/AMDGPU/memory_clause.ll
llvm/test/CodeGen/AMDGPU/nop-fold.mir [deleted file]
llvm/test/CodeGen/AMDGPU/power-sched-no-instr-sunit.mir