[PATCH] D135447: [AMDGPU] Add llvm.is.fpclass intrinsic to existing SelectionDAG fp class support for AMDGPU

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

Can do it as a follow up commit, but the existing combines we have for AMDGPU::FP_CLASS should be ported to use the generic intrinsic. Also, llvm.amdgcn.class should get bitcode upgraded to the generic



================
Comment at: llvm/test/CodeGen/AMDGPU/llvm.is.fpclass.ll:3
+; RUN:  llc -global-isel=0 -march=amdgcn -mcpu=gfx908 -verify-machineinstrs < %s | FileCheck --check-prefix=SELDAG %s
+
+
----------------
Should also test/handle globalisel


================
Comment at: llvm/test/CodeGen/AMDGPU/llvm.is.fpclass.ll:182
+declare i1 @llvm.is.fpclass.f64(double, i32)
+
----------------
Should add some vector cases too


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135447/new/

https://reviews.llvm.org/D135447