[AMDGPU] Precommit test for D79481. NFC

Test shows unnecessary s_waitcnt between shader exports.
llvm · May 7, 2020 · 0d4d86c · 0d4d86c
1 parent f649aca
commit 0d4d86c
Showing 1 changed file with 18 additions and 0 deletions.
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.exp.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.exp.ll
@@ -540,5 +540,23 @@ end:
   ret void
 }
 
+; GCN-LABEL: {{^}}test_export_clustering:
+; GCN-DAG: v_mov_b32_e32 [[W0:v[0-9]+]], 0
+; GCN-DAG: v_mov_b32_e32 [[X:v[0-9]+]], s0
+; GCN-DAG: v_mov_b32_e32 [[Y:v[0-9]+]], s1
+; GCN-DAG: v_add_f32_e32 [[Z0:v[0-9]+]]
+; GCN-DAG: exp param0 [[X]], [[Y]], [[Z0]], [[W0]]{{$}}
+; GCN-DAG: v_sub_f32_e32 [[Z1:v[0-9]+]]
+; GCN: s_waitcnt expcnt(0)
+; GCN: v_mov_b32_e32 [[W1:v[0-9]+]], 1.0
+; GCN: exp param1 [[X]], [[Y]], [[Z1]], [[W1]] done{{$}}
+define amdgpu_kernel void @test_export_clustering(float %x, float %y) #0 {
+  %z0 = fadd float %x, %y
+  call void @llvm.amdgcn.exp.f32(i32 32, i32 15, float %x, float %y, float %z0, float 0.0, i1 false, i1 false)
+  %z1 = fsub float %y, %x
+  call void @llvm.amdgcn.exp.f32(i32 33, i32 15, float %x, float %y, float %z1, float 1.0, i1 true, i1 false)
+  ret void
+}
+
 attributes #0 = { nounwind }
 attributes #1 = { nounwind inaccessiblememonly }