Nvidia bans using translation layers for CUDA software to run on other chips

#1

Post by justinclift » Mon, Mar 04, 2024, 4:58 PM UTC

Nvidia bans using translation layers for CUDA software to run on other chips
tomshardware.com

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#2

Post by justinclift » Mon, Mar 04, 2024, 5:00 PM UTC

On the face of it, this sounds like abusive behaviour by a monopolist.

Wonder if it'll be seen that way legally though?

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#3

Post by mindcrime » Mon, Mar 04, 2024, 5:11 PM UTC

Ironically, this is just going to increase interest in ROCm and other alternatives.

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#4

Post by jstanley » Mon, Mar 04, 2024, 5:22 PM UTC

People keep using proprietary software and keep getting burned by it. When will we learn?

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#5

Post by superkuh » Mon, Mar 04, 2024, 5:24 PM UTC

jstanley wrote: post #4

People keep using proprietary software and keep getting burned by it. When will we learn?

What's the alternative? AMD only supports their (consumer) GPUs for ~4 years via ROCm in some instances. If you buy the card at any time except release day you only get a couple years of compute support.

To answer my own question: opencl, and it's just as bad as it was in 2014. Or, slowly, people are starting to do compute with Vulkan. This might be the best way forwards even if it's an awkward choice.

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#6

Post by andersa » Mon, Mar 04, 2024, 5:25 PM UTC

> a new clause in CUDA 11.5 reads

Huh? That was released like 3 years ago.

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#7

Post by nottorp » Mon, Mar 04, 2024, 5:25 PM UTC

justinclift wrote: post #2

On the face of it, this sounds like abusive behaviour by a monopolist. Wonder if it'll be seen that way legally though?

For the US, didn't Oracle win the java api lawsuit?

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#8

Post by andersa » Mon, Mar 04, 2024, 5:25 PM UTC

jstanley wrote: post #4

People keep using proprietary software and keep getting burned by it. When will we learn?

There are no viable alternatives if you require high performance or even just the software to actually work properly.

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#9

Post by SomeHacker44 » Mon, Mar 04, 2024, 5:26 PM UTC

IANAL, but I don't see how this can be enforced (in the US) given the legal history of emulation. This is effectively just emulation.

c.f., https://emulation.gametechwiki.com/index.php/Licensing

Re: Nvidia bans using translation layers for CUDA software to run on other chips

#10

Post by justinclift » Mon, Mar 04, 2024, 5:27 PM UTC

nottorp wrote: post #7

justinclift wrote: post #2

On the face of it, this sounds like abusive behaviour by a monopolist. Wonder if it'll be seen that way legally though?

For the US, didn't Oracle win the java api lawsuit?

Doesn't seem like it, though it had to work it's way to the Supreme Court for the final decision:

* https://arstechnica.com/tech-policy/2021/04/how-the-supreme-...

* https://en.wikipedia.org/wiki/Google_LLC_v._Oracle_America,_....