[PATCH] D133723: [AMDGPU][GFX11] Use VGPR_32_Lo128 for VOP1,2,C

Jay Foad via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Sep 20 02:22:52 PDT 2022


foad accepted this revision.
foad added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIModeRegister.cpp:177
     case AMDGPU::FPTRUNC_UPWARD_PSEUDO: {
-      // Replacing the pseudo by a real instruction
-      MI.setDesc(TII->get(AMDGPU::V_CVT_F16_F32_e32));
+      // Replacing the pseudo by a real instruction in place
+      if (TII->getSubtarget().hasTrue16BitInsts()) {
----------------
This is fine for now. As a follow up I'll try changing FPTRUNC_UPWARD_PSEUDO into an e64 instruction, so we can go back to using a simple setDesc call here.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D133723/new/

https://reviews.llvm.org/D133723



More information about the llvm-commits mailing list