xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-24 13:06:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bug]: Limit num_reqs in dummy_run when max_num_seqs is small (#26144)

Browse Source 
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Benjamin Chislett 2025-10-03 00:00:20 -04:00 committed by yewentao256
parent a06bb9bf36
commit 56d0073f2a
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/v1/worker/gpu_model_runner.py
View File

@ -3060,7 +3060,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
assert not uniform_decode
# Create mixed batch:
# first half decode tokens, second half one prefill
num_decode_tokens = num_tokens // 2
num_decode_tokens = min(max_num_reqs - 1, num_tokens // 2)
num_prefill_tokens = num_tokens - num_decode_tokens
num_reqs = num_decode_tokens + 1
@ -3072,7 +3072,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
max_query_len = num_prefill_tokens
elif uniform_decode:
assert not create_mixed_batch
num_reqs = cdiv(num_tokens, max_query_len)
num_reqs = min(max_num_reqs, cdiv(num_tokens, max_query_len))
num_scheduled_tokens_list = [max_query_len] * num_reqs
if num_tokens % max_query_len != 0:
num_scheduled_tokens_list[-1] = num_tokens % max_query_len