[llvm] [AMDGPU] Fix isCanonicalized to require canonical input for NaN-propa… (PR #208733)

Wooseok Lee via llvm-commits llvm-commits at lists.llvm.org
Thu Jul 23 11:52:47 PDT 2026


================
@@ -15932,19 +15934,22 @@ bool SITargetLowering::isCanonicalized(SelectionDAG &DAG, SDValue Op,
     switch (IntrinsicID) {
     case Intrinsic::amdgcn_cvt_pkrtz:
     case Intrinsic::amdgcn_cubeid:
-    case Intrinsic::amdgcn_frexp_mant:
     case Intrinsic::amdgcn_fdot2:
+    case Intrinsic::amdgcn_trig_preop:
+    case Intrinsic::amdgcn_tanh:
+      return true;
+    case Intrinsic::amdgcn_frexp_mant:
     case Intrinsic::amdgcn_rcp:
     case Intrinsic::amdgcn_rsq:
     case Intrinsic::amdgcn_rsq_clamp:
     case Intrinsic::amdgcn_rcp_legacy:
     case Intrinsic::amdgcn_rsq_legacy:
----------------
wooseoklee wrote:

These instructions will be excluded in this fix. Sorry for the confusion.

https://github.com/llvm/llvm-project/pull/208733


More information about the llvm-commits mailing list