MCPcopy Create free account
hub / github.com/PaddlePaddle/FastDeploy / sleep

Method sleep

fastdeploy/worker/gpu_worker.py:199–201  ·  view source on GitHub ↗

Offload memory from GPU

(self, **kwargs)

Source from the content-addressed store, hash-verified

197 return self.model_runner.update_weights(version, verify_checksum)
198
199 def sleep(self, **kwargs) -> None:
200 """Offload memory from GPU"""
201 return self.model_runner.sleep(**kwargs)
202
203 def wakeup(self, **kwargs) -> None:
204 """Reload memory into GPU"""

Callers 15

initialize_kv_cacheMethod · 0.45
_keep_aliveMethod · 0.45
runMethod · 0.45
loop_scheduleMethod · 0.45
routine_reportMethod · 0.45
loop_expire_reqsMethod · 0.45
_monitor_queue_loadMethod · 0.45
_handle_connect_taskMethod · 0.45

Calls

no outgoing calls

Tested by 15

__init__Method · 0.36
slow_workMethod · 0.36
no_result_workMethod · 0.36
test_exception_scenarioFunction · 0.36
test_concurrentFunction · 0.36
test_cleanup_logicMethod · 0.36
worker_threadMethod · 0.36
test_timing_functionsMethod · 0.36
process_requestMethod · 0.36
worker_slicesMethod · 0.36