Revert "[PD Disaggregation] replace transfer with batch transfer for better performance (#7236)" (#7968)
This commit is contained in:
@@ -270,19 +270,17 @@ class MooncakeKVManager(BaseKVManager):
|
|||||||
|
|
||||||
# Worker function for processing a single layer
|
# Worker function for processing a single layer
|
||||||
def process_layer(src_ptr: int, dst_ptr: int, item_len: int) -> int:
|
def process_layer(src_ptr: int, dst_ptr: int, item_len: int) -> int:
|
||||||
src_addr_list = []
|
|
||||||
dst_addr_list = []
|
|
||||||
length_list = []
|
|
||||||
for prefill_index, decode_index in zip(prefill_kv_blocks, dst_kv_blocks):
|
for prefill_index, decode_index in zip(prefill_kv_blocks, dst_kv_blocks):
|
||||||
src_addr = src_ptr + int(prefill_index[0]) * item_len
|
src_addr = src_ptr + int(prefill_index[0]) * item_len
|
||||||
dst_addr = dst_ptr + int(decode_index[0]) * item_len
|
dst_addr = dst_ptr + int(decode_index[0]) * item_len
|
||||||
length = item_len * len(prefill_index)
|
length = item_len * len(prefill_index)
|
||||||
src_addr_list.append(src_addr)
|
|
||||||
dst_addr_list.append(dst_addr)
|
status = self.engine.transfer_sync(
|
||||||
length_list.append(length)
|
mooncake_session_id, src_addr, dst_addr, length
|
||||||
return self.engine.batch_transfer_sync(
|
)
|
||||||
mooncake_session_id, src_addr_list, dst_addr_list, length_list
|
if status != 0:
|
||||||
)
|
return status
|
||||||
|
return 0
|
||||||
|
|
||||||
futures = [
|
futures = [
|
||||||
executor.submit(
|
executor.submit(
|
||||||
|
|||||||
Reference in New Issue
Block a user