xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-23 16:04:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

minor

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon 2025-09-18 12:44:21 -07:00
parent 9050087250
commit 92f337faeb
1 changed files with 4 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/v1/worker/gpu/model_runner.py
View File

@ -219,9 +219,10 @@ class GPUModelRunner:
idx_mapping_list = [ idx_mapping_list = [
self.req_states.req_id_to_index[req_id] for req_id in req_ids self.req_states.req_id_to_index[req_id] for req_id in req_ids
] ]
self.input_buffers.idx_mapping.np[:num_reqs] = idx_mapping_list idx_mapping = self.input_buffers.idx_mapping
idx_mapping_np = self.input_buffers.idx_mapping.np[:num_reqs] idx_mapping.np[:num_reqs] = idx_mapping_list
idx_mapping = self.input_buffers.idx_mapping.copy_to_gpu(num_reqs) idx_mapping_np = idx_mapping.np[:num_reqs]
idx_mapping = idx_mapping.copy_to_gpu(num_reqs)
# Block tables: num_kv_cache_groups x [num_reqs, max_num_blocks] # Block tables: num_kv_cache_groups x [num_reqs, max_num_blocks]
block_tables = self.block_tables.gather_block_tables(idx_mapping) block_tables = self.block_tables.gather_block_tables(idx_mapping)