[llvm] [AMDGPU] Legalize 64bit elements for BUILD_VECTOR on gfx942 (PR #145052)

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Tue Jun 24 10:03:44 PDT 2025


================
@@ -15486,6 +15502,72 @@ SDValue SITargetLowering::performClampCombine(SDNode *N,
   return SDValue(CSrc, 0);
 }
 
+SDValue
+SITargetLowering::performBuildVectorCombine(SDNode *N,
+                                            DAGCombinerInfo &DCI) const {
+  const GCNSubtarget *ST = getSubtarget();
+  if (DCI.Level < AfterLegalizeDAG || !ST->hasMovB64())
+    return SDValue();
----------------
arsenm wrote:

This is still worthwhile without v_mov_b64 if we are going to use s_mov_b64 for the final use

https://github.com/llvm/llvm-project/pull/145052


More information about the llvm-commits mailing list