[GLM5.3 Perf] Optimize glm 5.3 metadata op, 1.6~4.8x kernel level performance improvement (#58450)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2026-09-25 13:15:18 -04:00
committed by GitHub
parent 2617fe9383
commit c88f4824ff
4 changed files with 26 additions and 25 deletions
@@ -291,17 +291,7 @@ class SparseMLACommonMetadataBuilder(AttentionMetadataBuilder[T]):
self,
common_attn_metadata: "CommonAttentionMetadata",
) -> torch.Tensor:
num_tokens = common_attn_metadata.num_actual_tokens
starts = np.asarray(common_attn_metadata.query_start_loc_cpu, dtype=np.int32)
seg_lengths = np.diff(starts)
req_id_per_token = np.repeat(
np.arange(seg_lengths.shape[0], dtype=np.int32), seg_lengths
)
self.req_id_per_token_buffer.fill_(0)
self.req_id_per_token_buffer[: req_id_per_token.shape[0]].copy_(
np_to_pinned_tensor(req_id_per_token), non_blocking=True
)
return self.req_id_per_token_buffer[:num_tokens]
return common_attn_metadata.token_to_req_indices(self.req_id_per_token_buffer)
def _build_chunked_context_fields(
self,