VP9 Encoder & Other Media Functionality Of Tesla's FSD Chip To Be Upstreamed In Linux

[phoronix]

Phoronix: VP9 Encoder & Other Media Functionality Of Tesla's FSD Chip To Be Upstreamed In Linux

With the Linux 5.18 kernel reaching stable in the next week or two there is basic support for Tesla's FSD chip. That Samsung-based SoC for powering Tesla's full-self driving technology has the basic support pieces in place for this kernel while Samsung engineers are working on ironing out other portions of the SoC support for future kernel releases...

https://www.phoronix.com/scan.php?page=news_item&px=Tesla-FSD-Chip-MFC-Patches

[Kjell]

Linux mailing list.. Feels just like the 90s.

[RahulSundaram]

Linux mailing list.. Feels just like the 90s.

It's actually pretty impressive tech behind the scenes. Refer to https://lwn.net/Articles/748184/ and https://lwn.net/Articles/878205/ for some details and https://blog.ffwll.ch/2017/08/github...he-kernel.html for an additional useful reference.

[wagaf]

Impressive but obsolete...

Sometimes it's time to let go.

[tildearrow]

Impressive but obsolete...

Sometimes it's time to let go.

1MB mail client memory usage or 100MB browser tab memory usage?
Not every kernel developer has a graphical interface available.
The mail client doesn't require anything other than an input/output stream (terminal client, serial port, SSH, etc.) for its interface.
On the other hand a browser requires a graphical display for it, which may not be readily available (e.g. when debugging a display driver).

[wagaf]

1MB mail client memory usage or 100MB browser tab memory usage?
Not every kernel developer has a graphical interface available.
The mail client doesn't require anything other than an input/output stream (terminal client, serial port, SSH, etc.) for its interface.
On the other hand a browser requires a graphical display for it, which may not be readily available (e.g. when debugging a display driver).

You don't need a web UI to push git commits to remotes for review. It's actually much better than sending emails.

And any tool like Gitlab, Gerrit, GitHub etc have webhooks allowing to do reviews from the command line if you wish so.

[RahulSundaram]

You don't need a web UI to push git commits to remotes for review. It's actually much better than sending emails.

And any tool like Gitlab, Gerrit, GitHub etc have webhooks allowing to do reviews from the command line if you wish so.

I guess you didn't read to the blog post I linked earlier yet. Refer to https://blog.ffwll.ch/2017/08/github...he-kernel.html

[wagaf]

I guess you didn't read to the blog post I linked earlier yet. Refer to https://blog.ffwll.ch/2017/08/github...he-kernel.html

I don't advocate for Linux to use GitHub

The development process for the Linux Kernel is now mature and well established, and because of the unique size of the project, it has specific needs and requirements. A major change is not immediately needed, and would be risky and costly, for sure.

However, the process was mostly designed before git existed. Using the git protocol to exchange commits and branches is much more powerful than sending emails. And IMHO mailing lists are completely obsolete as a communication tool in general. The more they wait to replace the current system, the more "organisational debt" will accumulate and progressively hinder the project.

Kernel developers would probably bring a lot to the ecosystem by contributing the required features to an existing tool, or by building their own modern tool, like Linus did with git, or Google with Gerrit.

[Quackdoc]

it's a shame VP9 didn't take off the same way HEVC did, It is interesting to see VP9 being used here though.

[cl333r]

1MB mail client memory usage or 100MB browser tab memory usage?
Not every kernel developer has a graphical interface available.

I think even Nigerian people have computers with displays, what do you mean?