But is the cat a crack shot too?

Yes.

You’ve convinced me. I still think that secret optimizations are a possibility, but I must concede that there might very well be this sort of lock-in bullshit and nothing else.

Yes? I’m not saying Mesa as a whole is bad, but Mesa+Nouveau for Nvidia cards is terrible.

(It doesn’t help that Nvidia isn’t exactly cooperative when it comes to supporting open-source developers, but my point that driver development is non-trivial stands.)

Using an older kernel isn’t a long-term solution. And according to the kernel devs, either using and older kernel in that way or modifying the kernel to remove these protections still violates the license even if it bypasses the technical protections.

(I’m guessing Nvidia will keep shimming and rely on either not being sued or winning the lawsuit.)

Reaching 100% utilization is simple and entirely under the control of the user. Optimized drivers are for giving that user more computation at 100% utilization.

Of course software can’t exceed the physical limits of the hardware but reaching the physical limits of the hardware is non-trivial, especially for hardware as complex as a modern GPU.

Had this clause been found enforceable in any court case? (I’m not saying it hasn’t been - I don’t know.)

They weren’t forced to do it. They did it as part of a settlement. The outcome if they had gone to trial and lost could well have been different.

(Also how do you even violate the license for gcc while making a router?)

The Linux market is too small for nVidia to care.

The Linux gaming market is too small for Nvidia to care, but the GPU computing market isn’t.

I wouldn’t say AMD is doing pretty well - it isn’t a serious competitor to Nvidia in the GPU computing market.

Of course I can’t know for sure because the driver is closed-source, but I’d bet that a lot of what makes Nvidia hardware work fast is actually in the driver rather than the hardware itself. Plus, a proprietary driver lets them lock people in to buying their hardware. The company where I work doesn’t use Nvidia software because it buys Nvidia GPUs. It buys Nvidia GPUs because it uses Nvidia software.

I did say that I get why the Linux folks are doing this. The problem is that Nvidia drivers that obey these restrictions and as a result have significantly worse performance than Nvidia drivers on other operating systems aren’t the solution either. Anyone who does serious GPU computing will still have to switch away from Linux.

(IMO Nvidia would be insane to open-source their drivers. Like sue-corporate-officers-for-breach-of-duty level insane. So they can’t do more than what they’re already doing: coming up with workarounds.)

From my closed-source corporate perspective, Nvidia is trying to improve performance and the Linux kernel maintainers are trying to stop them. I don’t see why I would be annoyed at Nvidia in these circumstances.

Would having their own distro even help? It seems like working around this would require forking from Linux at a lower level, and even that would only circumvent technical (rather than copyright) barriers.

I get why the Linux folks are doing this, but I don’t expect that it will make them popular with anyone who actually uses Nvidia drivers on Linux (which is a lot of people). I’m sure that my employer will choose up-to-date Nvidia drivers over up-to-date versions of the kernel, at least in the short term. In the long term it probably won’t be an issue since Nvidia will figure something out, but if it did become an issue then ultimately Nvidia driver support is non-negotiable for the company where I work.

(No one cares what a small tech company does, but the big guys need Nvidia too so it should be possible to piggyback on whatever they do.)