Record · corpus.blog/posts/01a07d5d-72e8-7140-82cd-cd9237967faa

Creating custom kernels for the AMD MI300

Hugging Face · published 9 July 2025

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at huggingface.co Wayback capture

0blogs citing

First cited: —
Most recent: —
Rank this month: not ranked
Links from this post: 31
Words captured: 6,534

Cited by

No blog we hold has cited this post yet.

Links from this post

Link

Host

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/title.png

huggingface.co

https://huggingface.co/meta-llama/Llama-3.1-405B

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/results_figure.png

huggingface.co

https://github.com/huggingface/hf-rocm-kernels

github.com

https://hub.docker.com/layers/rocm/vllm/rocm6.3.1_mi300_ubuntu22.04_py3.12_vllm_0.6.6/images/sha256-9a12ef62bbbeb5a4c30a01f702c8e025061f575aa129f291a49fbd02d6b4d6c9

hub.docker.com

https://discuss.pytorch.org/t/distributed-w-torchtitan-introducing-async-tensor-parallelism-in-pytorch/209487

discuss.pytorch.org

https://github.com/remi-or/vllm/tree/patch_hfrk

github.com

https://github.com/remi-or/vllm/blob/patch_hfrk/HFRK_readme.md

github.com

https://tinkerd.net/blog/machine-learning/cuda-basics/

tinkerd.net

How to Optimize a CUDA Matmul Kernel for cuBLAS-like Performance: a Worklog Read at the source ↗

siboehm.com

https://www.youtube.com/watch?v=OUIkkAPaw4M

youtube.com

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/thread_global.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/warp_global.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/cu_global.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/xcd.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/mi300x.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/disk_figure.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/bs_latency_figure.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/coalesced.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/rms_latency_figure.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/classic_gemm.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/skinny_gemm.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/split_k.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/MFMA_full.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/sample_sparse.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/sparsity_trick.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/gemm_AI.png

huggingface.co

https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/mi300kernels/async_work.png

huggingface.co

Outperforming cuBLAS on H100: a Worklog Read at the source ↗

cudaforfun.substack.com

https://github.com/huggingface/kernel-builder/

github.com

https://github.com/huggingface/kernels

github.com