Skip to content

fix bug: dp+tp warmup #3991

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 2 commits into from
Sep 29, 2025
Merged

fix bug: dp+tp warmup #3991

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bbb0774
fix bug: dp+tp warmup
Tsundoku958 Sep 19, 2025
54eabb3
assert dp size
Tsundoku958 Sep 19, 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
7 changes: 6 additions & 1 deletion lmdeploy/pytorch/engine/model_agent.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -395,12 +395,15 @@ def warmup(self):
with self.all_context():
max_batches = self.cache_config.max_batches
num_tokens = max_batches

dist_ctx = get_dist_manager().current_context()
dp = dist_ctx.dp
# warmup prefill
inputs = self.inputs_strategy.make_dummy(max_batches,
is_decoding=False,
device='cuda',
vocab_size=self.model_config.vocab_size)
if dp > 1:
inputs.build_dp_meta()
self._forward_impl(inputs)

# warmup decoding(with cuda graph)
Expand All @@ -411,6 +414,8 @@ def warmup(self):
is_decoding=True,
device='cuda',
vocab_size=self.model_config.vocab_size)
if dp > 1:
inputs.build_dp_meta()
self._forward_impl(inputs)

def _slice_outs(self, inputs: torch.Tensor, seq_length: torch.LongTensor):
Expand Down