AMDGPU: Do not fold clamp instructions when sources are different

Patch by hakzsam (Samuel Pitoiset)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@314951 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/AMDGPU/SIFoldOperands.cpp b/lib/Target/AMDGPU/SIFoldOperands.cpp
index 9fc38aeefaa21b2d744dd0189c1ea65a7160deb7..a527afb25da032ce533a3970c54ab940af38a65d 100644 (file)
--- a/lib/Target/AMDGPU/SIFoldOperands.cpp
+++ b/lib/Target/AMDGPU/SIFoldOperands.cpp
@@ -737,6 +737,7 @@ const MachineOperand *SIFoldOperands::isClamp(const MachineInstr &MI) const {
     const MachineOperand *Src0 = TII->getNamedOperand(MI, AMDGPU::OpName::src0);
     const MachineOperand *Src1 = TII->getNamedOperand(MI, AMDGPU::OpName::src1);
     if (!Src0->isReg() || !Src1->isReg() ||
+        Src0->getReg() != Src1->getReg() ||
         Src0->getSubReg() != Src1->getSubReg() ||
         Src0->getSubReg() != AMDGPU::NoSubRegister)
       return nullptr;

diff --git a/test/CodeGen/AMDGPU/clamp.ll b/test/CodeGen/AMDGPU/clamp.ll
index 2d6b4f3c0c5741b80d61e32f88b29094e55ddc0a..216ecf76345661f0e93ad3864d2e92390816a0ac 100644 (file)
--- a/test/CodeGen/AMDGPU/clamp.ll
+++ b/test/CodeGen/AMDGPU/clamp.ll
@@ -663,6 +663,28 @@ define amdgpu_kernel void @v_clamp_v2f16_shuffle(<2 x half> addrspace(1)* %out,
   ret void
 }
 
+; GCN-LABEL: {{^}}v_clamp_diff_source_f32:
+; GCN: v_add_f32_e32 [[A:v[0-9]+]]
+; GCN: v_add_f32_e32 [[B:v[0-9]+]]
+; GCN: v_max_f32_e64 v{{[0-9]+}}, [[A]], [[B]] clamp{{$}}
+define amdgpu_kernel void @v_clamp_diff_source_f32(float addrspace(1)* %out, float addrspace(1)* %aptr) #0
+{
+  %gep0 = getelementptr float, float addrspace(1)* %aptr, i32 0
+  %gep1 = getelementptr float, float addrspace(1)* %aptr, i32 1
+  %gep2 = getelementptr float, float addrspace(1)* %aptr, i32 2
+  %l0 = load float, float addrspace(1)* %gep0
+  %l1 = load float, float addrspace(1)* %gep1
+  %l2 = load float, float addrspace(1)* %gep2
+  %a = fadd nsz float %l0, %l1
+  %b = fadd nsz float %l0, %l2
+  %res = call nsz float @llvm.maxnum.f32(float %a, float %b)
+  %max = call nsz float @llvm.maxnum.f32(float %res, float 0.0)
+  %min = call nsz float @llvm.minnum.f32(float %max, float 1.0)
+  %out.gep = getelementptr float, float addrspace(1)* %out, i32 3
+  store float %min, float addrspace(1)* %out.gep
+  ret void
+}
+
 declare i32 @llvm.amdgcn.workitem.id.x() #1
 declare float @llvm.fabs.f32(float) #1
 declare float @llvm.minnum.f32(float, float) #1