[Core] Ensure memory is pinned prior to async h2d copy (#45424)
Signed-off-by: Nick Hill <nickhill123@gmail.com>
This commit is contained in:
@@ -145,7 +145,6 @@ def _generate_fake_sampling_metadata(
|
||||
vllm_config.scheduler_config.max_num_seqs,
|
||||
num_spec,
|
||||
device,
|
||||
PIN_MEMORY_AVAILABLE,
|
||||
)
|
||||
fake_sampling_metadata = SamplingMetadata(
|
||||
temperature=torch.full((batch_size,), 0.0),
|
||||
@@ -880,7 +879,6 @@ def test_maybe_create_thinking_budget_holder_without_reasoning():
|
||||
cfg.scheduler_config.max_num_seqs,
|
||||
0,
|
||||
torch.device("cpu"),
|
||||
False,
|
||||
)
|
||||
is None
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user