Skip to content

stride is wrong when computing local var in the smem-layernorm.cu #1

New issue
New issue
Open
Open
stride is wrong when computing local var in the smem-layernorm.cu#1
@cailun01

Description

@cailun01
cailun01
opened on May 19, 2025

Hi, Arya!
I think it should be int stride = blockDim.x/2.

cuda/layernorm/kernels/smem-layernorm.cu

Line 52 in 18ee1d1

for(int stride = blockDim.x; stride > 0; stride /= 2){

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions