| Issue |
243749
|
| Summary |
[X86] missed CompressEVEX opportunity
|
| Labels |
backend:X86,
missed-optimization
|
| Assignees |
|
| Reporter |
Andarwinux
|
found from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128770
https://godbolt.org/z/cn93aTYEE
```c
float
sel (float a, float b, float x, float y)
{
return a == b ? x : y;
}
```
gcc -O3 -march=znver5 -ffast-math
```asm
"sel":
vcmpneq_osss xmm0, xmm0, xmm1
vblendvps xmm0, xmm2, xmm3, xmm0
ret
```
clang -O3 -march=znver5 -ffast-math
```asm
sel:
vcmpeqss k1, xmm0, xmm1
vmovss xmm3 {k1}, xmm3, xmm2
vmovaps xmm0, xmm3
ret
```
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs