[Bugfix][ROCm] Skip FP8 MLA prefill PS-metadata build for chunked-context batches (#52356)
Signed-off-by: Shantipriya Parida <shantipriya.parida@amd.com>
This commit is contained in:
@@ -777,7 +777,11 @@ class AiterMLAMetadataBuilder(MLACommonMetadataBuilder[AiterMLAMetadata]):
|
||||
attn_metadata.reduce_indptr = self._mla_reduce_indptr
|
||||
attn_metadata.reduce_final_map = self._mla_reduce_final_map
|
||||
attn_metadata.reduce_partial_map = self._mla_reduce_partial_map
|
||||
if self._fp8_prefill_enabled and attn_metadata.prefill is not None:
|
||||
if (
|
||||
self._fp8_prefill_enabled
|
||||
and attn_metadata.prefill is not None
|
||||
and attn_metadata.prefill.chunked_context is None
|
||||
):
|
||||
self._build_fp8_prefill_ps_metadata(attn_metadata, common_attn_metadata)
|
||||
return attn_metadata
|
||||
|
||||
|
||||
Reference in New Issue
Block a user