Revert " Fp8 block scale quantization for fmha fwd (#3330)" (#3633)

This reverts commit dd0b4294af.
This commit is contained in:
Po Yen Chen
2026-01-23 13:21:19 +08:00
committed by GitHub
parent f30d04654e
commit de5a1d730d
14 changed files with 84 additions and 667 deletions

View File

@@ -95,11 +95,10 @@ run_fp8bf16_tests() {
for perm in 0 1 ; do
for b in 1 2 ; do
for hdim in 64 128 256 ; do
for scale in 1 2; do
$EXE -prec=fp8bf16 -init=3 -b=$b -h=1 -d=$hdim -s=128 -iperm=$perm -operm=$perm -vlayout=r -qscale=$scale -kname=$KNAME $COMMON_ARGS
$EXE -prec=fp8bf16 -init=3 -b=$b -h=1 -d=$hdim -s=128 -iperm=$perm -operm=$perm -vlayout=r -qscale=1 -kname=$KNAME $COMMON_ARGS
done ; done ; done ; done
done ; done ; done
}
run_fp8fp32_tests() {