apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[Iluvatar] Support V1_KVCACHE_SCHEDULER and paddleocr-vl rope mode (#5555)

Browse Source
This commit is contained in:
yzwu
2025-12-18 18:14:25 +08:00
committed by GitHub
parent 48f3e9797e
commit ac013803f3
24 changed files with 1212 additions and 1090 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
custom_ops/gpu_ops/get_padding_offset.cu
View File

@@ -28,8 +28,13 @@ __global__ void PrefixSumKernel(int64_t *ids_remove_padding,
const int max_seq_len) {
const int bi = blockIdx.x;
const int tid = threadIdx.x;
#ifdef PADDLE_WITH_COREX
const int warp_id = threadIdx.x / 64;
const int lane_id = threadIdx.x % 64;
#else
const int warp_id = threadIdx.x / 32;
const int lane_id = threadIdx.x % 32;
#endif
int cum_seq_len = 0;