youngkingdom/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
Haicheng Wu
2025-01-25 00:17:07 -05:00
committed by GitHub
parent fca21140a5
commit b353e36fe8
1 changed files with 1 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
README.md
View File

@ -99,12 +99,7 @@ CUTLASS team is working on a fix.
# Performance
CUTLASS primitives are very efficient. When used to construct device-wide GEMM kernels,
they exhibit nearly optimal utilization of peak theoretical throughput. The figure below
shows CUTLASS 3.8's performance as a % of theoretical peak utilization
on various input and output data types when run on NVIDIA Blackwell SM100 architecture GPU.
<p align="center"><img src=media/images/cutlass-3.8-blackwell-gemm-peak-performance.svg></p>
they exhibit nearly optimal utilization of peak theoretical throughput.
The two figures below show the continual CUTLASS performance improvements
on an [NVIDIA H100](https://www.nvidia.com/en-us/data-center/h100/) (NVIDIA Hopper architecture) since
CUTLASS 3.1.