Skip to content

softmax.cu spell error #371

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Aug 9, 2025
Merged

softmax.cu spell error #371

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f3bf632
softmax.cu spell error
napleon-liu Aug 8, 2025
4ed0285
format
napleon-liu Aug 8, 2025
3db60e7
format
napleon-liu Aug 8, 2025
File filter

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,7 +44,7 @@
<div align='center'>
<img src='https://github.com/user-attachments/assets/a5ec4320-d2f9-4254-888a-170b2d9e3784' height=170px>
</div>
-->
-->

- [2025-01-08]: **[🤖ffpa-attn](https://github.com/xlite-dev/ffpa-attn.git)** is released! Yet another Faster Flash Prefill Attention with O(1)🎉SRAM complexity for large headdim, **1.8x~3x↑**🎉 vs SDPA EA: [📈L20 ~1.9x↑🎉](https://github.com/xlite-dev/ffpa-attn?tab=readme-ov-file#L1-bench-l20), [📈A30 ~1.8x↑🎉](https://github.com/xlite-dev/ffpa-attn?tab=readme-ov-file#L1-bench-a30),[📈4090 ~2.1x↑🎉](https://github.com/xlite-dev/ffpa-attn?tab=readme-ov-file#L1-bench-4090).

Expand All @@ -54,7 +54,7 @@
<img src='https://github.com/user-attachments/assets/447e2937-f7c8-47c8-8550-8c0c71b910e6' height="170px" width="229px">
<img src='https://github.com/user-attachments/assets/65a8d564-8fa7-4d66-86b9-e238feb86143' height="170px" width="229px">
</div>
-->
-->
<div align='center'>
<img height="320px" alt="image" src="https://github.com/user-attachments/assets/ed30185b-2e11-4293-832f-43e9003d6ad9" />
</div>
Expand Down
12 changes: 6 additions & 6 deletions kernels/softmax/softmax.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -333,14 +333,14 @@ __global__ void online_safe_softmax_f32_per_token_kernel(const float *x,
// for softmax)
int local_tid = threadIdx.x;
int global_tid = blockIdx.x * NUM_THREADS + threadIdx.x;
const int WAPR_NUM = NUM_THREADS / WARP_SIZE;
const int WARP_NUM = NUM_THREADS / WARP_SIZE;
int warp_id = local_tid / WARP_SIZE;
int lane_id = local_tid % WARP_SIZE;
MD val;
val.m = global_tid < N ? x[global_tid] : -FLT_MAX;
val.d = global_tid < N ? 1.0f : 0.0f;

__shared__ MD shared[WAPR_NUM];
__shared__ MD shared[WARP_NUM];
MD res = warp_reduce_md_op<WARP_SIZE>(val);

if (lane_id == 0)
Expand All @@ -349,7 +349,7 @@ __global__ void online_safe_softmax_f32_per_token_kernel(const float *x,

if (local_tid < WARP_SIZE) {
MD block_res = shared[local_tid];
block_res = warp_reduce_md_op<WAPR_NUM>(block_res);
block_res = warp_reduce_md_op<WARP_NUM>(block_res);
if (local_tid == 0) {
shared[0] = block_res;
}
Expand All @@ -371,7 +371,7 @@ online_safe_softmax_f32x4_pack_per_token_kernel(float *x, float *y, int N) {
int local_tid = threadIdx.x;
int global_tid = (blockIdx.x * NUM_THREADS + local_tid) * 4;

const int WAPR_NUM = NUM_THREADS / WARP_SIZE;
const int WARP_NUM = NUM_THREADS / WARP_SIZE;
int warp_id = local_tid / WARP_SIZE;
int lane_id = local_tid % WARP_SIZE;
// compare local max value
Expand All @@ -382,15 +382,15 @@ online_safe_softmax_f32x4_pack_per_token_kernel(float *x, float *y, int N) {

MD local_md = {local_m, local_d};
MD res = warp_reduce_md_op<WARP_SIZE>(local_md);
__shared__ MD shared[WAPR_NUM];
__shared__ MD shared[WARP_NUM];

if (lane_id == 0)
shared[warp_id] = res;
__syncthreads();
// do block reduce
if (local_tid < WARP_SIZE) {
MD block_res = shared[local_tid];
block_res = warp_reduce_md_op<WAPR_NUM>(block_res);
block_res = warp_reduce_md_op<WARP_NUM>(block_res);
if (local_tid == 0)
shared[0] = block_res;
}
Expand Down