3 ms·
The write up on the GitHub repo is much more informative than the blog. When running int8 matmul using onnx performance is ~0.6TF. https://github.com/usefulse
by fancyfredbot 2y ago
The write up on the GitHub repo is much more informative than the blog.
When running int8 matmul using onnx performance is ~0.6TF.
https://github.com/usefulsensors/qc_npu_benchmark https://github.com/usefulsensors/qc_npu_benchmark
- dang 2y agoThanks—we changed the URL to that from https://petewarden.com/2024/10/16/ai-pcs-arent-very-good-at-ai/ https://petewarden.com/2024/10/16/ai-pcs-arent-very-good-at-.... Readers may way want to look at both, of course!
- dhruvdh 2y agoOh, maybe also change the title? I flagged it because of the title/url not matching.