Skip to content

GitLab

"...git@developer.sourcefind.cn:2222/OpenDAS/vllm_cscc.git" did not exist on "fa4b70555bc1163c2ee8fa4b193be9957f05dda6"
Commit f39feef5 authored by zhuwenwen's avatar zhuwenwen
Browse files

update prefix_prefill params

parent 9807bb0f
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
vllm/attention/ops/prefix_prefill.py
View file @ f39feef5
...@@ -684,7 +684,7 @@ if triton.__version__ >= "2.1.0": ...@@ -684,7 +684,7 @@ if triton.__version__ >= "2.1.0":
sliding_window=None): sliding_window=None):
cap = torch.cuda.get_device_capability() cap = torch.cuda.get_device_capability()
BLOCK = 128 if cap[0] >= 8 else 32 BLOCK = 32 if cap[0] >= 8 else 32
# shape constraints # shape constraints
Lq, Lk, Lv = q.shape[-1], k.shape[-1], v.shape[-1] Lq, Lk, Lv = q.shape[-1], k.shape[-1], v.shape[-1]
assert Lq == Lk and Lk == Lv assert Lq == Lk and Lk == Lv
......
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