Skip to content

[Minor] Use larger batch sizes for A100/B100/B200/MI300x #17073

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Apr 24, 2025
Merged

[Minor] Use larger batch sizes for A100/B100/B200/MI300x #17073

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
195f5f5
[Minor] Use large max_num_seqs for A100/B100/B200
WoosukKwon Apr 23, 2025
3fad03a
Merge branch 'main' into max-num-seqs
WoosukKwon Apr 23, 2025
95fd298
Use memory
WoosukKwon Apr 23, 2025
e6eb964
Merge branch 'main' into max-num-seqs
WoosukKwon Apr 23, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions vllm/engine/arg_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@
from vllm.test_utils import MODEL_WEIGHTS_S3_BUCKET, MODELS_ON_S3
from vllm.transformers_utils.utils import check_gguf_file
from vllm.usage.usage_lib import UsageContext
from vllm.utils import FlexibleArgumentParser, is_in_ray_actor
from vllm.utils import FlexibleArgumentParser, GiB_bytes, is_in_ray_actor

# yapf: enable

Expand Down Expand Up @@ -1625,13 +1625,13 @@ def _set_default_args_v1(self, usage_context: UsageContext) -> None:
# values for non-H100/H200 GPUs.
try:
from vllm.platforms import current_platform
device_name = current_platform.get_device_name().lower()
device_memory = current_platform.get_device_total_memory()
except Exception:
# This is only used to set default_max_num_batched_tokens
device_name = "no-device"
device_memory = 0

if "h100" in device_name or "h200" in device_name:
# For H100 and H200, we use larger default values.
if device_memory >= 70 * GiB_bytes:
# For GPUs like H100 and MI300x, use larger default values.
default_max_num_batched_tokens = {
UsageContext.LLM_CLASS: 16384,
UsageContext.OPENAI_API_SERVER: 8192,
Expand Down