This was an ambitions attempt to implement INT4 and WMMA support in vulkan on the Strix Halo platform.

With some quick and dirty methods, I saw a 20% increase on performance on INT4 primatives. And an additional 10% improvement with WMMA using 16x16 matrices.

It was an extremely dirty hack and not production ready in any sense, but it shows that there's at least 30% possible performance gains on the strix halo.

I don't have the bandwidth to pursue this further.

Mesa - The 3D Graphics Library

Source

This repository lives at https://gitlab.freedesktop.org/mesa/mesa. Other repositories are likely forks, and code found there is not supported.

Build & install

You can find more information in our documentation (docs/install.rst), but the recommended way is to use Meson (docs/meson.rst):

$ meson setup build
$ ninja -C build/
$ sudo ninja -C build/ install

Support

Many Mesa devs hang on IRC; if you're not sure which channel is appropriate, you should ask your question on OFTC's #dri-devel, someone will redirect you if necessary. Remember that not everyone is in the same timezone as you, so it might take a while before someone qualified sees your question. To figure out who you're talking to, or which nick to ping for your question, check out Who's Who on IRC.

The next best option is to ask your question in an email to the mailing lists: mesa-dev@lists.freedesktop.org

Bug reports

If you think something isn't working properly, please file a bug report (docs/bugs.rst).

Contributing

Contributions are welcome, and step-by-step instructions can be found in our documentation (docs/submittingpatches.rst).

Note that Mesa uses gitlab for patches submission, review and discussions.