[Fix][DSV4.1] SWA page size with bounded replay (#42128)
This commit is contained in:
@@ -2420,7 +2420,7 @@ class DeepseekV4AttnBackend(
|
||||
req_to_token=self.req_to_token,
|
||||
full_to_swa=self.token_to_kv_pool.full_to_swa_index_mapping,
|
||||
swa_window_size=SWA_WINDOW,
|
||||
swa_page_size=self.token_to_kv_pool.swa_kv_pool.page_size,
|
||||
swa_page_size=self.token_to_kv_pool.get_swa_key_page_size(),
|
||||
num_qo_tokens=num_qo_tokens,
|
||||
max_seq_len=max(seq_lens_cpu_list),
|
||||
total_swa=total_swa,
|
||||
@@ -3316,7 +3316,7 @@ class DeepseekV4AttnBackend(
|
||||
compress_ratio
|
||||
)
|
||||
|
||||
swa_kv_page_size = token_to_kv_pool.swa_kv_pool.page_size
|
||||
swa_kv_page_size = token_to_kv_pool.get_swa_key_page_size()
|
||||
assert swa_k_cache.ndim == 2
|
||||
# The kernel detects each cache's format from the last dim of this
|
||||
# view: 584 (V4), 528 (V4.1 fp8) or 288 (V4.1 fp4, extra cache only).
|
||||
|
||||
@@ -1860,6 +1860,11 @@ class DeepSeekV4TokenToKVPool(BaseSWAKVPool):
|
||||
# swa_kv_pool is None under the request window and unified_kv.
|
||||
return self.kv_layout
|
||||
|
||||
def get_swa_key_page_size(self) -> int:
|
||||
if self.request_window is not None:
|
||||
return self.request_window.page_size
|
||||
return self.swa_kv_pool.page_size
|
||||
|
||||
def get_swa_key_bytes_per_token(self) -> int:
|
||||
"""Last dim of the ``(pages, page_size, 1, bytes)`` view the attention
|
||||
kernel detects the SWA cache's format from."""
|
||||
|
||||
Reference in New Issue
Block a user