Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

tomshardware.com

Post reply on HN
Print view View original

81–90 of 380 posts

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#81

Post by latchkey »

PeterisP wrote: post #47
latchkey wrote: post #37

Earlier quoted context omitted.

AMD definitely has and is doubling down on ROCm.

Perhaps they're doubling down, but even doubling down is not enough to say that they're serious about it since they've been so neglectful for so many years - for example, right now they explicitly say that many of AMD GPUs are not supported by ROCm; if they're not willing to put their money where their mouth is and do the legwork to ensure support for powerful cards they sold just a few years ago, how can they say yo…

People argue for ROCm to support older cards because that is all they have accessible to them. AMD has lagged on getting expensive cards into the hands of end users because they've focused only on building super computers.

I predict that access to the newer cards is a more likely scenario. Right now, you can't rent a MI250 or even MI300x, but that is going to change quickly. Azure is going to have them, as well as others (I know this, cause that's what I'm building now).

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#82

Post by latchkey »

bragr wrote: post #34

People don't seem particularly motivated to move away from CUDA to me. I've been poking around various models and tooling over the last couple months, and they pretty much all have something like device = "cuda" if torch.cuda.is_available() else "cpu" and I've yet to see a single one implement the AMD NN middleware: https://www.amd.com/en/developer/zendnn.html

You didn't watch the announcement on the MI300x. You don't have to change that line to use AMD.

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#83

Post by shihab »

lacker wrote: post #66
buildbot wrote: post #33

As another commenter said, it's CUDA. Intel and AMD and whoever can turn out chips reasonably fast, but nobody gets that it's the software and ecosystem. You have to out-compete the ecosystem. You can pick up a used Mi100 that performs almost like an A100 for 5x less money on eBay for example. Why is it 5x less? Because the software incompatibilities mean you'll spend a ton of time getting it to work compared to an N…

The funny thing to me is that so much of the "AI software ecosystem" is just PyTorch. You don't need to develop some new framework and make it popular. You don't need to support a zillion end libraries. Just literally support PyTorch. If PyTorch worked fine on Intel GPUs, a lot of people would be happy to switch.

But you can't support Pytorch without a proper foundation in place. They don't need to support zillion _end_ libraries, sure, but they do need to have at least a very good set of standard libraries, equivalent of Cublas, Curand etc.

And they don't. My work recently had me working with rocRAND (Rocm's answer to Curand). It was frankly pretty bad- the design, performance (50% slower in places that don't make any sense because generating random numbers is not exact that complicated), and documentation (God it was awful).

Now, that's a small slice of the larger pie. But imagine if this trend continues for other libraries.

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#84

Post by mschuetz »

PeterisP wrote: post #50
bragr wrote: post #34

People don't seem particularly motivated to move away from CUDA to me. I've been poking around various models and tooling over the last couple months, and they pretty much all have something like device = "cuda" if torch.cuda.is_available() else "cpu" and I've yet to see a single one implement the AMD NN middleware: https://www.amd.com/en/developer/zendnn.html

This is a bit of a chicken and egg problem, because the developers of these projects probably don't even have relevant AMD devices to test that, because why would they? The only reasonable way to break the cycle would be proactive intervention by AMD to contribute code and testing and easy "works out of the box" installation to all the major popular projects to add AMD support so they can sell more their hardware lat…

CUDA is just way to good in terms of UX with nothing platform-independent coming close to it. I wish there was a competitor, but there simply isn't.

I wish AMD&Intel would extend compute shaders with pointers&pointer casting, arbitrary large buffers instead of just 4GB, device-wide sync, and function pointers. Those are kinda my must-have functionality. Even better, just use C++ for compute shaders.

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#86

Post by mschuetz »

jonahrd wrote: post #51

I absolutely understand why NVIDIA is incentivized to maintain CUDA dominance, and I absolutely understand that within these market parameters, AMD/Intel/others have dropped the ball. However it's worth noting that in the end, it's the consumer that loses out when these technical/capability moats maintain a de facto sort of monopoly on certain use cases.

AMD/Intel/Khronos are free to compete with something better. NVIDIA isn't preventing them from doing so. And in that regard, CUDA is a massive benefit to consumers, because the alternatives are really bad.

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#87

Post by kaelinl »

bragr wrote: post #34

People don't seem particularly motivated to move away from CUDA to me. I've been poking around various models and tooling over the last couple months, and they pretty much all have something like device = "cuda" if torch.cuda.is_available() else "cpu" and I've yet to see a single one implement the AMD NN middleware: https://www.amd.com/en/developer/zendnn.html

I haven't used it personally, but my understanding is that AMD/ROCm-accelerated backends for PyTorch overload the "cuda" device and module so that ROCm shows up as CUDA for feature testing. They want to make a transition seamless, and lots of existing code checks for CUDA, so they do what's necessary to make that existing code run.

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#88

Post by pgeorgi »

ActionHank wrote: post #76
mi_lk wrote: post #26

Earlier quoted context omitted.

You don't know Pat Gelsinger do you?

We're in 2023, he's been in the CEO seat for 2 years already. He's had plenty of time to show the world his intent and where they are going. All that has happened is they launched a very mid GPU and have yielded more ground to AMD. Meanwhile AMD continue to eat away at Intel's talent pool, market share, and still managed to push into the AI space. He should be sweating.

> for 2 years already ... All that has happened is they launched a very mid GPU

Hardware development cycles are closer to 5 years. So while he might have gotten some adjustments done on the designs so far, if he turned the ship around it'll take a while longer to materialize.

The software side is more agile, so any tea leave reading to discern what Gelsinger's strategy looks like is best done over there.

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#89

Post by alecco »

It looks like both Pat Gelsinger and Lisa Su have no clue about software and can't even delegate this. They expect "the community" to do the software for their extremely complicated hardware.

If you see the presentations by Nvidia's Bill Daly [1] it shows they've been evolving hardware and software together. The CUDA programming model was a massive bet a long time ago. And they deservedly won ML/AI.

Unless Intel and AMD do a radical change it's game over for them. They will lose to ARM and Nvidia.

[1] https://www.youtube.com/watch?v=kLiwvnr4L80

Re: Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'

#90

Post by lumost »

shihab wrote: post #83
lacker wrote: post #66

Earlier quoted context omitted.

The funny thing to me is that so much of the "AI software ecosystem" is just PyTorch. You don't need to develop some new framework and make it popular. You don't need to support a zillion end libraries. Just literally support PyTorch. If PyTorch worked fine on Intel GPUs, a lot of people would be happy to switch.

But you can't support Pytorch without a proper foundation in place. They don't need to support zillion _end_ libraries, sure, but they do need to have at least a very good set of standard libraries, equivalent of Cublas, Curand etc. And they don't. My work recently had me working with rocRAND (Rocm's answer to Curand). It was frankly pretty bad- the design, performance (50% slower in places that don't make any sense…

Folks also underestimate how complex these libraries are. There are dozens of projects to make BLAS alternatives which give up after ~3-6 months when they realize that this project will take years to be successful.
Post reply on HN