apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 00:17:25 +08:00
Code Issues Actions 19 Packages Projects Releases Wiki Activity

cpmmot (#7105)

Browse Source 
Co-authored-by: “liuruian” <liuruian@baidu.com>
This commit is contained in:
周周周
2026-03-31 16:13:44 +08:00
committed by GitHub
parent 5c5dc66aa7
commit fd44bb7cbf
1 changed files with 1 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files
custom_ops/gpu_ops/append_attn/multiquery_attention_c16_impl.cuh
+1 -19
View File
@@ -485,26 +485,8 @@ __global__ void multi_query_append_attention_warp1_4_kernel(
const int *block_table_now = block_table + batch_id * max_block_num_per_seq;
const uint32_t q_len = seq_lens[batch_id];
if (q_len <= 0) {
return;
}
const uint32_t kv_len = seq_lens_kv[batch_id] + q_len;
uint32_t kv_len = seq_lens_kv[batch_id];
if (ENABLE_PREFILL) {
kv_len += q_len;
if (kv_len <= 0) {
return;
}
} else {
if (kv_len <= 0) {
return;
}
kv_len += q_len;
}
const int seq_len_enc = seq_lens_encoder[batch_id];
if (seq_len_enc > 0) {
return;
}
const uint32_t num_chunks_this_seq = div_up(kv_len, chunk_size);
if (chunk_idx >= num_chunks_this_seq) {
return;