| Issue |
78856
|
| Summary |
[AMDGPU] Should v_permlane64_b32 be handled in fixVcmpxPermlaneHazards?
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
DadSchoorse
|
The GFX11 ISA doc says:
```
7.2.8. PERMLANE Specific Rules
V_PERMLANE may not occur immediately after a V_CMPX. To prevent this, any other VALU opcode may be
inserted (e.g. V_NOP).
```
As there is no restriction to `v_permlane(x)16`, should `isPermlane` be updated to include `v_permlaneb64_b32`?
![]()
_______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
