[PATCH] D56845: [AMDGPU] Add some missing always-uniform values.

[Neil Henning via Phabricator via llvm-commits]

sheredom updated this revision to Diff 182296.
sheredom added a comment.

Also tag s.getpc as scalar, and add tests in the Analysis folder like suggested.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D56845/new/

https://reviews.llvm.org/D56845

Files:
  lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
  test/Analysis/DivergenceAnalysis/AMDGPU/always_uniform.ll


Index: test/Analysis/DivergenceAnalysis/AMDGPU/always_uniform.ll
===================================================================
--- test/Analysis/DivergenceAnalysis/AMDGPU/always_uniform.ll
+++ test/Analysis/DivergenceAnalysis/AMDGPU/always_uniform.ll
@@ -1,6 +1,7 @@
 ; RUN: opt  -mtriple amdgcn-unknown-amdhsa -analyze -divergence -use-gpu-divergence-analysis %s | FileCheck %s
 
-define amdgpu_kernel void @workitem_id_x() #1 {
+; CHECK: for function 'workitem_id_x':
+define amdgpu_kernel void @workitem_id_x() {
   %id.x = call i32 @llvm.amdgcn.workitem.id.x()
 ; CHECK: DIVERGENT:  %id.x = call i32 @llvm.amdgcn.workitem.id.x()
   %first.lane = call i32 @llvm.amdgcn.readfirstlane(i32 %id.x)
@@ -8,7 +9,42 @@
   ret void
 }
 
+; CHECK: for function 'icmp':
+define amdgpu_kernel void @icmp(i32 inreg %x) {
+; CHECK-NOT: DIVERGENT:  %icmp = call i64 @llvm.amdgcn.icmp.i32
+  %icmp = call i64 @llvm.amdgcn.icmp.i32(i32 %x, i32 0, i32 33)
+  ret void
+}
+
+; CHECK: for function 'fcmp':
+define amdgpu_kernel void @fcmp(float inreg %x, float inreg %y) {
+; CHECK-NOT: DIVERGENT:  %fcmp = call i64 @llvm.amdgcn.fcmp.i32
+  %fcmp = call i64 @llvm.amdgcn.fcmp.i32(float %x, float %y, i32 33)
+  ret void
+}
+
+; CHECK: for function 's_buffer_load':
+define amdgpu_kernel void @s_buffer_load(<4 x i32> inreg %desc) {
+; CHECK-NOT: DIVERGENT:  %load = call <2 x i32> @llvm.amdgcn.s.buffer.load.v2i32
+  %load = call <2 x i32> @llvm.amdgcn.s.buffer.load.v2i32(<4 x i32> %desc, i32 0, i32 0)
+  ret void
+}
+
+; CHECK: for function 's_getpc':
+define amdgpu_kernel void @s_getpc(<4 x i32> inreg %desc) {
+; CHECK-NOT: DIVERGENT:  %load = call i64 @llvm.amdgcn.s.getpc
+  %load = call i64 @llvm.amdgcn.s.getpc()
+  ret void
+}
+
 declare i32 @llvm.amdgcn.workitem.id.x() #0
 declare i32 @llvm.amdgcn.readfirstlane(i32) #0
+declare i64 @llvm.amdgcn.icmp.i32(i32, i32, i32) #1
+declare i64 @llvm.amdgcn.fcmp.i32(float, float, i32) #1
+declare <2 x i32> @llvm.amdgcn.s.buffer.load.v2i32(<4 x i32>, i32, i32) #2
+declare i64 @llvm.amdgcn.s.getpc() #3
 
 attributes #0 = { nounwind readnone }
+attributes #1 = { nounwind readnone convergent }
+attributes #2 = { nounwind readonly convergent }
+attributes #3 = { nounwind readnone speculatable }
Index: lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -579,6 +579,10 @@
       return false;
     case Intrinsic::amdgcn_readfirstlane:
     case Intrinsic::amdgcn_readlane:
+    case Intrinsic::amdgcn_icmp:
+    case Intrinsic::amdgcn_fcmp:
+    case Intrinsic::amdgcn_s_buffer_load:
+    case Intrinsic::amdgcn_s_getpc:
       return true;
     }
   }


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D56845.182296.patch
Type: text/x-patch
Size: 2770 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190117/143fb3b6/attachment.bin>