Skip to content

[Bugfix] Fix DeepSeek V3.2 OOM during CG memory profiling #36691

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
MatthewBonanni merged 2 commits into from
Mar 11, 2026
+5 −7
Merged

[Bugfix] Fix DeepSeek V3.2 OOM during CG memory profiling #36691

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8161015
Fix
MatthewBonanni Mar 10, 2026
0af8be3
Add comment
MatthewBonanni Mar 10, 2026
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
12 changes: 5 additions & 7 deletions vllm/v1/worker/gpu_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5550,16 +5550,14 @@ def _init_minimal_kv_cache_for_profiling(self) -> None:
kv_cache_spec = self.get_kv_cache_spec()
kv_cache_groups = get_kv_cache_groups(self.vllm_config, kv_cache_spec)
min_blocks = self.compilation_config.max_cudagraph_capture_size or 1
if kv_cache_groups:
page_size = kv_cache_groups[0].kv_cache_spec.page_size_bytes
group_size = max(len(g.layer_names) for g in kv_cache_groups)
available_memory = min_blocks * page_size * group_size
else:
available_memory = 1 # Attention-free model

# Temporarily change num_gpu_blocks_override to allocate a minimal KV cache
saved_override = self.cache_config.num_gpu_blocks_override
self.cache_config.num_gpu_blocks_override = min_blocks
minimal_config = get_kv_cache_config_from_groups(
self.vllm_config, kv_cache_groups, available_memory=available_memory
self.vllm_config, kv_cache_groups, available_memory=0
)
self.cache_config.num_gpu_blocks_override = saved_override

self.initialize_kv_cache(minimal_config)
self.cache_config.num_gpu_blocks = minimal_config.num_blocks
Expand Down