Hacker News new | past | comments | ask | show | jobs | submit login

There is no chance of Linux winning this over Nvidia. The thing is there is no alternative to Nvidia, while there are many to Linux. The demand is seemingly infinite. Every big corp is rushing to secure as many as they can. For people talking about it being hard to keep their own forked or patched kernel, Nvidia is so big at this point that it can buy RedHat(from IBM), Suse, and almost every Linux vendor on the market and the cost would be mostly not relevant to them. There is zero chance of anyone avoiding Nvidia now due to Linux issues. Sure, 5 years ago this might have worked, but that ship has sailed.



Eh that's a pretty wild statement. Some big AI and compute custers don't outweigh the vast vast majority of servers and cloud instances running Linux that will never see or need a GPU. What are these alternatives to Linux?


For instances that don't need a GPU, this new change will simply not matter. My point is that this new change is unlikely to make Nvidia do anything differently compared to what they are already doing.


> The thing is there is no alternative to Nvidia, while there are many to Linux.

Huh? I am a backebd dev and a gamer who would never give up Linux for a proprietary OS, but has no NVidia hardware.

That being said, I know you're talking about industry. My question is, so are we supposed to just take it? NVidia has to be pushed to do the right thing here, as AMD and Intel have done or are we just slaves to 'too big to fail' and somehow this is a triumph of capitalism?


I am talking about industry solely, and the time for pushing was 5 years ago and before. I am not saying what would be the ideal outcome, but that at this point anyone running AI workloads will abandon Linux over Nvidia if there is a need to do so.


I suspect we're going to see more and more specialized hardware in this area before that happens, but what benefit exactly is there to having them on Linux anyway?

Most of that industry is highly opposed to being open in any sensible definition of that word, so what do they bring exactly?


NVDAs foothold here is not about gaming, but machine learning. Few people play on Linux anyway, but machine learning is basically not happening without Nvidia. Not that this is favorable, but it is true. And yes, I am taking it, I simply do not care that much.


And I am guessing ML workloads use up a single computer to do just that (and nothing else), meaning that NVidia could start their own fork of Linux, and people need to just eat it (or switch to AMD, Intel, etc.!)




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: