apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 00:33:03 +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

8
fastdeploy/worker/dcu_worker.py
View File

@@ -13,18 +13,12 @@
# See the License for the specific language governing permissions and
# limitations under the License.
"""
import gc
import time
from typing import List, Optional
import paddle
import paddle.nn as nn
from fastdeploy.config import FDConfig
from fastdeploy.engine.request import Request
from fastdeploy.utils import get_logger
from fastdeploy.worker.gpu_model_runner import GPUModelRunner
from fastdeploy.worker.output import ModelRunnerOutput
from fastdeploy.worker.gpu_worker import GpuWorker
logger = get_logger("dcu_worker", "dcu_worker.log")
@@ -97,7 +91,7 @@ class DcuWorker(GpuWorker):
paddle_peak_increase = paddle_reserved_mem_after_run - paddle_allocated_mem_before_run
available_kv_cache_memory = total_gpu_memory * \
self.parallel_config.gpu_memory_utilization - after_used_gpu_memory - 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(