[CPU] Add CPU FP8 W8A8 linear/MoE support (#49942)

Signed-off-by: yuwenzho <yuwen.zhou@intel.com>
Co-authored-by: Li, Jiang <jiang1.li@intel.com>
This commit is contained in:
Yuwen Zhou
2026-09-18 17:59:42 +08:00
committed by GitHub
co-authored by Li, Jiang <jiang1.li@intel.com>
parent 6df2b1a8c6
commit b346479065
20 changed files with 2924 additions and 46 deletions
@@ -1055,7 +1055,7 @@ class MLAAttention(nn.Module, AttentionLayerBase):
mqa_q_nope,
self.impl._w_uk_packed, # type: ignore[attr-defined]
True,
None,
self.impl._w_scale, # type: ignore[attr-defined]
)
mqa_ql_nope = mqa_ql_nope.transpose(0, 1)
else:
@@ -1401,7 +1401,7 @@ class MLAAttention(nn.Module, AttentionLayerBase):
x,
self.impl._w_uv_packed, # type: ignore[attr-defined]
True,
None,
self.impl._w_scale, # type: ignore[attr-defined]
)
else:
# Multiply + Transpose (N, B, L) x (N, L, V)->(N, B, V)->(B, N, V)