[PATCH] D105104: [AMDGPU] PHI node cost should not be counted for the size and latency.
Alexander via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Jun 29 13:17:23 PDT 2021
alex-t updated this revision to Diff 355345.
alex-t added a comment.
Test corrected
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D105104/new/
https://reviews.llvm.org/D105104
Files:
llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
llvm/test/Analysis/CostModel/AMDGPU/control-flow.ll
Index: llvm/test/Analysis/CostModel/AMDGPU/control-flow.ll
===================================================================
--- llvm/test/Analysis/CostModel/AMDGPU/control-flow.ll
+++ llvm/test/Analysis/CostModel/AMDGPU/control-flow.ll
@@ -8,7 +8,7 @@
; SPEED: estimated cost of 10 for instruction: ret void
; SIZE: estimated cost of 5 for instruction: br i1
; SIZE: estimated cost of 1 for instruction: br label
-; SIZE: estimated cost of 1 for instruction: %phi = phi i32 [
+; SIZE: estimated cost of 0 for instruction: %phi = phi i32 [
; SIZE: estimated cost of 1 for instruction: ret void
define amdgpu_kernel void @test_br_cost(i32 addrspace(1)* %out, i32 addrspace(1)* %vaddr, i32 %b) #0 {
bb0:
Index: llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
===================================================================
--- llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -837,10 +837,6 @@
}
case Instruction::Ret:
return SCost ? 1 : 10;
- case Instruction::PHI:
- // TODO: 1. A prediction phi won't be eliminated?
- // 2. Estimate data copy instructions in this case.
- return 1;
}
return BaseT::getCFInstrCost(Opcode, CostKind, I);
}
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D105104.355345.patch
Type: text/x-patch
Size: 1253 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210629/f9d7b423/attachment.bin>
More information about the llvm-commits
mailing list