Yeah I guess Nvidia have had to grow up pretty fast recently.
Linux
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
Nvidia have been big kernel contributers for a long time, even before the "fuck you nvidia" thing. They hold their graphics driver close to their chest, but have done a lot of other work for the kernel.
What's an example? I would have thought, back then especially, their driver (and maybe nvapi) was most of the software they shipped.
My memory is fuzzy, but they have had their tegra SOC since the 2000s, and somewhat more recently they have been a big player in data center networking.
And ever since CUDA became a thing they have been a big name in HPC and super computers, which is usually Linux based.
So they have done a lot of behind the scenes Linux work (and possibly BSD?).
Yeah, afaik the tegra was only used for embedded, closed source devices though, no? Did they submit any non-proprietary tegra support upstream?
And afaik CUDA has also always been proprietary bins. Maybe you mean they had to submit upstream fixes here and there to get their closed-source stuff working properly?
Tegra was used in android tablets, I had a couple. Not sure what the licence status was, but it was supported in cyanogen, so they must have had to make some changes to the kernel for that?
Certainly some of the stuff the upstreamed was to support their drivers, but they would have also been working on other more general things to support their super computers and other HPC stuff.
They also had a chipset for intel motherboards (which I can't find anything about), which may have had some work required?
I don't really know exactly the scope of all the work, but they have been in the top 20 companies for kernel development for a long time, and I assuming it can't just be supporting their own drivers.
Its hard to find the stats, but from here: https://bootlin.com/community/contributions/kernel-contributions/ you can click through and get breakdowns per kernel release: https://web.archive.org/web/20160803012713/remword.com/kps_result/3.8_whole.html
Cool, that's a good source to peruse, thanks.
I wish rust integrated more nicely with gnu and guix.
In what way?
You can't just run rustup or install the guix rust package because rust has its own standalone self-contained ecosystem which kinda clashes with the declarative nature of guix since it isn't built with integration with tools outside of the rust ecosystem in mind.
Yeah tldr is "rust good", "ai overrated", "i only care about the kernel and won't answer your questions"
How do you have 25 upvotes? Everything you wrote is wrong.
Linus said, that the rust infrastructure is not stable, is positive about AIs future and happy, that NVIDIA had to step up their open source game.
And even the interviewer mentioned, that the "I only care about the kernel" quote WILL be taken out of context.
And he answered even implied questions...
What? If you read the post you'll see that he was sad the rust adoption wasn't going faster.
Yes, but he also commented that the rust infrastructure isn’t super stable.
The point is that that Linus responses were not as overtly simplified and predictable as lung suggested.
Time is passing by so fast, really. I still remember linux 5 kernels to be modern, and 6 somehow still feels "the new thing"
And in reality they're all just in the 2.6 branch. I still remember the transition from 2.4.
If you think that if you cannot count a minor version number from from fingers and toes, and it is meaningless anyway, why not drop the current versioning system entirely? It would be fine if it was (major version).(patch)