AMDGPU: Use less surprising form of ConstantFP::get

[arsenm]

Avoid deducing the IR type from the APFloat type

[arsenm]

• AMDGPU: Use less surprising form of ConstantFP::get #139248 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[llvmbot]

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

Changes

Avoid deducing the IR type from the APFloat type

---

Full diff: https://github.com/llvm/llvm-project/pull/139248.diff

1 Files Affected:

• (modified) llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp (+2-2)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp b/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
index 8679a7cd0f88d..e76396f6ffbb0 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp
@@ -906,8 +906,8 @@ GCNTTIImpl::instCombineIntrinsic(InstCombiner &IC, IntrinsicInst &II) const {
         if (const ConstantFP *C2 = dyn_cast<ConstantFP>(Src2)) {
           APFloat Result = fmed3AMDGCN(C0->getValueAPF(), C1->getValueAPF(),
                                        C2->getValueAPF());
-          return IC.replaceInstUsesWith(
-              II, ConstantFP::get(IC.Builder.getContext(), Result));
+          return IC.replaceInstUsesWith(II,
+                                        ConstantFP::get(II.getType(), Result));
         }
       }
     }