Record · corpus.blog/posts/01a0e418-ecbb-7160-9bcc-3e2ee6f6f50b

Outperforming cuBLAS on NVFP4

cudaforfun.substack.com · published 15 August 2026

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 cudaforfun.substack.com Wayback capture

1blogs citing

First cited: 11 Sept 2026
Most recent: 11 Sept 2026
Rank this month: not ranked
Links from this post: 30
Words captured: 4,946

Cited by

Blog

In the post

Date

osmarks.net

Fixing Graphcore Read at the source ↗

11 Sept 2026

Links from this post

Link

Host

tcgen05 for dummies Read at the source ↗

gau-nernst.github.io

https://danielvegamyhre.github.io/2026/03/29/mxfp8-gemm.html

danielvegamyhre.github.io

https://www.paulwillchan.com/articles/outperforming-cublas-b200

paulwillchan.com

Modular: Matrix Multiplication on Blackwell: Part 4 - Breaking SOTA Read at the source ↗

modular.com

https://substackcdn.com/image/fetch/$s_!BZE_!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F51acc575-2368-4126-8ff2-b2fc20e9371c_1917x1115.png

substackcdn.com

https://github.com/pranjalssh/fast.cu

github.com

https://substackcdn.com/image/fetch/$s_!Twu8!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3842ffb6-9d76-419d-b6b9-1ed146cf5f2d_1560x1086.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!TAw0!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff5a92785-f118-4a2b-bf23-7bb2d120fc16_1512x994.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!UjFY!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7483ceb5-9e4e-4ba3-bfa2-bbb8d572a50b_2365x1395.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!6OsV!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffa018a75-e12a-4126-ade6-61187865acaa_1822x1118.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!pv_3!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F03b2ce05-f6b9-48c1-af4d-057e96185005_1931x1164.png

substackcdn.com

https://docs.nvidia.com/cuda/parallel-thread-execution/

docs.nvidia.com

https://substackcdn.com/image/fetch/$s_!Qy82!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fde363e62-c826-4ea3-90b5-9c975ed6b155_1822x748.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!qk3N!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8be04a27-7566-4fad-b9af-94f8803de906_2365x1395.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!yAf0!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9f9f829d-985b-48e1-a3f3-5f986cababa7_1931x1164.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!GAXU!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa24c43d3-b9b6-4f85-bc71-7a15f8658629_1931x1025.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!GY46!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F322739e6-3b3e-454c-8c67-f493ff6e1f65_1822x951.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!vbBF!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fecab08f6-4a3d-4025-98a7-7dbcef91b6a8_2365x1395.png

substackcdn.com

https://docs.nvidia.com/cuda/archive/13.0.0/cuda-c-programming-guide/index.html

docs.nvidia.com

https://docs.nvidia.com/cuda/parallel-thread-execution/index.html

docs.nvidia.com

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

cudaforfun.substack.com

https://substackcdn.com/image/fetch/$s_!9Jfh!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F25bd2248-4e92-4b39-9219-8db8078fcea3_1931x1010.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!G5Iu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa11c13df-0f73-4e61-8764-58eba858f484_1512x994.png

substackcdn.com

https://images.nvidia.com/aem-dam/en-zz/Solutions/data-center/nvidia-ampere-architecture-whitepaper.pdf

images.nvidia.com

https://www.nvidia.com/en-us/on-demand/session/gtcspring21-s33322/

nvidia.com

https://github.com/ademeure/QuickRunCUDA/blob/main/tests/side_aware.cu

github.com

https://substackcdn.com/image/fetch/$s_!ZZis!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6b08b25c-a4a0-4ae2-9c3f-bca87de49eed_1512x686.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!mbSz!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8ee426f4-d632-4110-bfde-5893639a21b0_1560x682.png

substackcdn.com

https://substackcdn.com/image/fetch/$s_!6J5n!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7d9b656f-b134-440e-8f6c-132c67759503_2596x1076.png

substackcdn.com

https://arxiv.org/pdf/2604.15379

arxiv.org