Skip to content

GitLab

Unverified Commit a884bc62 authored by Xin Yang's avatar Xin Yang Committed by GitHub
Browse files

[LoRA] Update LoRA expand kernel heuristic (#32425) 


Signed-off-by: default avatarXin Yang <xyangx@amazon.com>
parent 7a103043
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
vllm/lora/ops/triton_ops/utils.py
View file @ a884bc62
...@@ -251,7 +251,7 @@ def get_lora_op_configs( ...@@ -251,7 +251,7 @@ def get_lora_op_configs(
else: else:
default = { default = {
"block_m": 64, "block_m": 64,
"block_n": 128, "block_n": max(64, next_power_of_2(128 // num_slices)),
"block_k": 16, "block_k": 16,
"num_warps": 4, "num_warps": 4,
"num_ctas": 1, "num_ctas": 1,
......
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