[Misc]Minor Changes about Worker (vllm-project#11555)

Signed-off-by: Chenguang Li <[email protected]>

Author: noemotiovon

vllm/v1/worker/gpu_worker.py

@@ -132,7 +132,6 @@ def determine_num_available_blocks(self) -> Tuple[int, int]:
         # Execute a forward pass with dummy inputs to profile the memory usage
         # of the model.
         self.model_runner.profile_run()
-        torch.cuda.synchronize()
 
         free_gpu_memory, _ = torch.cuda.mem_get_info()
         # NOTE(woosuk): Here we assume that the other processes using the same

vllm/worker/worker.py

@@ -200,7 +200,6 @@ def determine_num_available_blocks(self) -> Tuple[int, int]:
                               weights_memory_in_bytes=self.model_runner.
                               model_memory_usage) as result:
             self.model_runner.profile_run()
-            torch.cuda.synchronize()
 
         self._assert_memory_footprint_increased_during_profiling()