[PATCH] D40158: AMDGPU: Use gfx9 carry-less add/sub instructions

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: lib/Target/AMDGPU/SIInstrInfo.cpp:3897
+
+    unsigned NewOpc = Opc == AMDGPU::S_ADD_I32 ?
+      AMDGPU::V_ADD_U32_e64 : AMDGPU::V_SUB_U32_e64;
----------------
rampitec wrote:
> Now you can send here S_ADD_U32 and you do not want to convert it into V_SUB_U32_e64.
S_ADD_U32 is never selected, so it never reaches here. I added it to the other switch so that in case it is selected, it will hit the assert here.


https://reviews.llvm.org/D40158