Skip to content

GitLab

Unverified Commit e80cfe57 authored by Woosuk Kwon's avatar Woosuk Kwon Committed by GitHub
Browse files

[MRV2] Avoid recompilation of _gather_block_tables_kernel (#37645) 


Signed-off-by: default avatarWoosuk Kwon <woosuk@inferact.ai>
parent d0532bf3
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
vllm/v1/worker/gpu/block_table.py
View file @ e80cfe57
......@@ -169,7 +169,7 @@ class BlockTables:
return self.slot_mappings[:, :num_tokens]
@triton.jit
@triton.jit(do_not_specialize=["num_reqs"])
def _gather_block_tables_kernel(
batch_idx_to_req_idx, # [batch_size]
src_block_table_ptrs, # [num_kv_cache_groups]
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment