Skip to content

[Minor] Reduce num blocks of qknorm in small batch size #2264

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
yzh119 merged 1 commit into from
Dec 24, 2025
Merged

[Minor] Reduce num blocks of qknorm in small batch size #2264

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion include/flashinfer/norm.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -369,7 +369,8 @@ cudaError_t QKRMSNorm(T* input, T* weight, T* output, uint32_t batch_size, uint3
FLASHINFER_CUDA_CALL(cudaGetDevice(&dev_id));
FLASHINFER_CUDA_CALL(cudaDeviceGetAttribute(&num_sms, cudaDevAttrMultiProcessorCount, dev_id));

dim3 nblks(num_blocks_per_sm * num_sms);
const int needed_blocks = ceil_div(batch_size * num_heads, num_warps);
dim3 nblks(std::min(num_blocks_per_sm * num_sms, needed_blocks));
dim3 nthrs(32, num_warps);
config.gridDim = nblks;
config.blockDim = nthrs;
Expand Down