apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 16:48:03 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[BugFix] Fix Configs (#2849)

Browse Source 
* fix config

* fix config
This commit is contained in:
YuanRisheng
2025-07-16 10:50:36 +08:00
committed by GitHub
parent 0fad10b35a
commit 101ad33332
10 changed files with 30 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
fastdeploy/worker/gpu_worker.py
View File

@@ -61,7 +61,8 @@ class GpuWorker(WorkerBase):
gc.collect()
paddle.device.cuda.empty_cache()
if self.parallel_config.enable_custom_all_reduce:
from fastdeploy.distributed.communication_op import use_custom_allreduce
from fastdeploy.distributed.communication_op import \
use_custom_allreduce
use_custom_allreduce()
else:
raise RuntimeError(
@@ -137,7 +138,7 @@ class GpuWorker(WorkerBase):
available_kv_cache_memory = after_run_meminfo.total * \
self.parallel_config.gpu_memory_utilization - after_run_meminfo.used - paddle_peak_increase
available_kv_cache_memory += model_block_memory_used * self.parallel_config.max_block_num
available_kv_cache_memory += model_block_memory_used * self.parallel_config.total_block_num
end_time = time.perf_counter()
logger.info((