apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-09-26 20:41:53 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

[Feature] Support mixed deployment with adapter (#3517)

Browse Source
This commit is contained in:
chenjian
2025-08-21 18:19:01 +08:00
committed by GitHub
parent d18a637a17
commit 606d9e9c2c
2 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
fastdeploy/scheduler/local_scheduler.py
View File

@@ -208,6 +208,9 @@ class LocalScheduler:
"""
return (token_num + block_size - 1) // block_size
def get_unhandled_request_num(self):
return len(self.requests)
def get_requests(
self,
available_blocks,

4
fastdeploy/splitwise/internal_adapter_utils.py
View File

@@ -56,9 +56,9 @@ class InternalAdapter:
"splitwise_role": self.cfg.splitwise_role,
"block_size": int(self.cfg.cache_config.block_size),
"block_num": int(available_block_num),
"max_block_num": self.cfg.cache_config.total_block_num,
"max_block_num": int(self.cfg.cache_config.total_block_num),
"dec_token_num": int(self.cfg.cache_config.dec_token_num),
"available_resource": 1.0 * available_block_num / self.cfg.cache_config.total_block_num,
"available_resource": float(1.0 * available_block_num / self.cfg.cache_config.total_block_num),
"max_batch_size": int(available_batch_size),
"max_input_token_num": self.cfg.max_num_batched_tokens,
"unhandled_request_num": self.engine.scheduler.get_unhandled_request_num(),