Update vllm/v1/structured_output/__init__.py

Co-authored-by: Russell Bryant <rbryant@redhat.com>

Author: benchislett

vllm/v1/structured_output/__init__.py

@@ -44,7 +44,10 @@ def __init__(self, vllm_config: VllmConfig):
  self.fill_bitmask_parallel_threshold = 128
  if self.fill_bitmask_parallel_threshold < max_batch_size:
  self.fill_bitmask_parallel_batch_size = 16
- max_workers = max(1, min(multiprocessing.cpu_count(), 8))
+ # Use:
+ # - at least 1 CPU
+ # - at most half the number of CPUs or 8, whichever is less
+ max_workers = max(1, min(multiprocessing.cpu_count() // 2, 8))
  self.executor_for_fillmask = ThreadPoolExecutor(
  max_workers=max_workers)