xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 02:15:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Kernel] update comment for KV shape in unified triton attn (#18099)

Browse Source 
Signed-off-by: haochengxia <xhc_1007@163.com>
This commit is contained in:
Percy 2025-05-20 13:19:34 -05:00 committed by GitHub
parent e1f5a71ed7
commit 980a172474
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
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/attention/ops/triton_unified_attention.py
View File

@ -31,8 +31,8 @@ def apply_softcap(S, x):
def kernel_unified_attention_2d(
output_ptr, # [num_tokens, num_query_heads, head_size]
query_ptr, # [num_tokens, num_query_heads, head_size]
key_cache_ptr, # [num_blks, num_kv_heads, head_size // x, blk_size, x]
value_cache_ptr, # [num_blks, num_kv_heads, head_size, blk_size]
key_cache_ptr, # [num_blks, blk_size, num_kv_heads, head_size]
value_cache_ptr, # [num_blks, blk_size, num_kv_heads, head_size]
block_tables_ptr, # [num_seqs, max_num_blocks_per_seq]
seq_lens_ptr, # [num_seqs]
alibi_slopes_ptr, # [num_query_heads]