[Core] Ensure memory is pinned prior to async h2d copy (#45424)

Signed-off-by: Nick Hill <nickhill123@gmail.com>
This commit is contained in:
Nick Hill
2026-06-20 20:02:24 -07:00
committed by GitHub
parent 8dd1b702f2
commit 7df3d7dada
49 changed files with 254 additions and 264 deletions
+2 -3
View File
@@ -4,8 +4,7 @@
import torch
from vllm.model_executor.layers.utils import apply_penalties
from vllm.utils.platform_utils import is_pin_memory_available
from vllm.utils.torch_utils import make_tensor_with_pad
from vllm.utils.torch_utils import PIN_MEMORY, make_tensor_with_pad
def apply_all_penalties(
@@ -52,6 +51,6 @@ def _convert_to_tensors(
pad=vocab_size,
device="cpu",
dtype=torch.int64,
pin_memory=is_pin_memory_available(),
pin_memory=PIN_MEMORY,
)
return output_tokens_tensor.to(device, non_blocking=True)