Skip to content

Reduce blockDim in fake_quantize_kernel for improved SM occupancy #8115

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

Open
flutist wants to merge 1 commit into
base: master
Choose a base branch
from
+6 −1
Open

Reduce blockDim in fake_quantize_kernel for improved SM occupancy #8115

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
7 changes: 6 additions & 1 deletion csrc/quantization/fake_quantizer.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,8 +173,13 @@ void launch_fake_quantize_kernel(T* vals,
int num_bits,
cudaStream_t stream)
{
// Reduced from 1024 to 256: improves per-SM warp occupancy (32→48 warps/SM)
// for large group_num (ZeRO gradient compression on production models where
// group_num >> num_SMs * blocks_per_SM). Measured 28-77% speedup at group_num
// = 87K-1.4M. For small group_num (< num_SMs) this is slower, but that is not
// the production use case.
dim3 grid_dim(group_num);
dim3 block_dim(1024);
dim3 block_dim(256);

fake_quantize_kernel<<<grid_dim, block_dim, 0, stream>>>(
vals, total_count / group_num, num_bits);
Expand Down
Loading