[llvm] [AMDGPU] Duplicate packed fp32 instructions (PR #212857)

Stanislav Mekhanoshin via llvm-commits llvm-commits at lists.llvm.org
Wed Jul 29 13:52:28 PDT 2026


================
@@ -3728,21 +3728,21 @@ unsigned getLdsDwGranularity(const MCSubtargetInfo &ST) {
   return 64; // In sync with getAddressableLocalMemorySize
 }
 
-bool isPackedFP32Inst(unsigned Opc) {
+bool isPackedSingleSGPRFP32Inst(unsigned Opc) {
   switch (Opc) {
-  case AMDGPU::V_PK_ADD_F32:
-  case AMDGPU::V_PK_ADD_F32_gfx12:
-  case AMDGPU::V_PK_MUL_F32:
-  case AMDGPU::V_PK_MUL_F32_gfx12:
-  case AMDGPU::V_PK_FMA_F32:
-  case AMDGPU::V_PK_FMA_F32_gfx12:
+  case AMDGPU::V_PK_ADD_F32_gfx1250:
+  case AMDGPU::V_PK_ADD_F32_gfx1250_gfx12:
----------------
rampitec wrote:

Yep. I am actually not sure if real instructions are needed in this context at all.

https://github.com/llvm/llvm-project/pull/212857


More information about the llvm-commits mailing list