edcadet10/gpu-systems-lab

Compare the all-reduce benchmark with nccl-tests about 1 month ago

help wanted benchmark needs-design

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Add a native CUDA or CUTLASS comparison provider about 1 month ago

enhancement help wanted kernel needs-design

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Publish v0.2 RMSNorm measurements on two accelerator generations about 1 month ago

help wanted benchmark needs-review

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Implement and verify a residual-RMSNorm backward kernel about 1 month ago

enhancement help wanted kernel needs-design

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Compare the all-reduce benchmark with nccl-tests about 1 month ago

help wanted benchmark needs-design

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Publish v0.2 RMSNorm measurements on two accelerator generations about 1 month ago

help wanted benchmark needs-review

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Add a benchmark report validation command about 1 month ago

enhancement help wanted good first issue

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Add a native CUDA or CUTLASS comparison provider about 1 month ago

enhancement help wanted kernel needs-design

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Implement and verify a residual-RMSNorm backward kernel about 1 month ago

enhancement help wanted kernel needs-design

edcadet10/gpu-systems-lab

0

Reproducible GPU kernels, profiling, performance models, and collective communication experiments.

Python

#cuda#distributed-systems#gpu#kernel-fusion#nccl#nsight#open-source#performance-engineering#pytorch#rmsnorm#roofline-model#triton

Open Issues Need Help