Live data from Hacker News

AMD reckons it can win the high-end GPU battle

techradar.com

31–40 of 160 posts

Re: AMD reckons it can win the high-end GPU battle

#31

Earlier quoted context omitted.

OpenCL is all you need, and for the sake of a healthy (read: non-monopoly) GPU computing ecosystem you must insist on good OpenCL support! Having an additional CUDA backend is fine if you want, but AMD GPUs are often really strong in a level playing field, which is exactly what you want as a consumer. Basically no consumer ever benefitted from a monopoly, and GPU computing is very important. I've developed (as lead G…

It's not, though. The blas and fft libraries are not only much slower (less efficient for the claimed peak of the cards), but they're missing features. For example, gemms don't work on complex half precision in any opencl library I've seen. Batched modes are missing for some types, etc. For basic operations, they're sufficient. But it's hard to get behind libraries that AMD doesn't really control. The ship has sailed…

In principle there's nothing stopping people from developing similarly great libraries for OpenCL, which I would on the one hand expect to happen eventually, but on the other hand it's impossible to deny the existing achievements of Nvidia's software team.

What I'm hoping is that people in a similar position to mine, where I could easily choose OpenCL since I was coding everything from scratch in basically C, choose OpenCL instead of CUDA.

It's completely possible in many cases, and I'm not the world's greatest developer or anything; twice I chose OpenCL over CUDA and both turned out great. Particularly the JIT model is very useful. Judge for yourself:

https://indigorenderer.com

https://chaoticafractals.com

It's also very useful for cross-vendor GPU benchmarking, see https://www.indigorenderer.com/benchmark-results?filter=sing...

Re: AMD reckons it can win the high-end GPU battle

#34
post #27

Earlier quoted context omitted.

I am stiiiill mostly reading/researching but I'm pretty sure that CUDA is crucial and OpenCL just won't cut it for ML. That said, AMD had a CUDA clone/conversion system a while back. I'm not sure where that's gone. The ideal would be to leapfrog CUDA and present something more usable and straightforward, especially something where the memory and threading model is really clear. CUDA already seem pretty good but there…

well, there is SYCL

I believe it's going to get a lot of support from Intel as well for their forthcoming GPU efforts.

Re: AMD reckons it can win the high-end GPU battle

#35
post #18

Earlier quoted context omitted.

Agree, I dropped Linux due bad graphics drivers (AMD)

How long ago did you try it? It was definitely bad in the fglrx days!

My move happened few months ago. I used Linux for years (and still use it in VMs/servers), however after changing GPU it was a compatibility nightmare. I chose AMD partly because I heard that drivers are good on Linux. Not sure how many hours I spend on it.

Re: AMD reckons it can win the high-end GPU battle

#36

Earlier quoted context omitted.

Agree, I dropped Linux due bad graphics drivers (AMD)

Are you really saying that AMD is worse than Nvidia on Linux ??? I have a Nvidia GPU (on my laptop), and I've had a truly horrible experience on Linux with Nvidia proprietary drivers. I was actually considering ditching my laptop, and buying a new one with AMD chipsets, due to the issues I've had on Linux due to the presence of the Nvidia graphics card. I occasionally play games on my Linux laptop, using Proton (a St…

I am not saying that. I just had terrible experience on a laptop with AMD GPU + Linux (rip battery) and on a desktop with discrete AMD GPU (suspend and whatnot errors) + Linux. Friends and coworkers with NVIDIA didn't seem to have those problems

Re: AMD reckons it can win the high-end GPU battle

#37
post #3

They aren't going to win anything until they fix their laughably bad drivers and keep them fixed. I have several Ryzens, but there's no way I'd go anywhere near their graphics cards.

We bought a few Ryzens with Vega8 for our devs on linux and it was a mistake. Unfortunately integrated graphics glitched horribly and it required us to find right set of kernel flags and features to disable. After encountering bug with amdgpu module randomly giving up, we just used discrete gpu. I don't feel comfortable relying on their gpus after this experience but cpus are undeniably good.

Re: AMD reckons it can win the high-end GPU battle

#38
post #3

They aren't going to win anything until they fix their laughably bad drivers and keep them fixed. I have several Ryzens, but there's no way I'd go anywhere near their graphics cards.

The drivers are perfectly fine, as long as you don't have a new 5xxx card. Those seem to have ongoing issues.

But my experience with AMD cards on Windows and Linux has been overall pretty great.

Re: AMD reckons it can win the high-end GPU battle

#39
post #18

Earlier quoted context omitted.

How long ago did you try it? It was definitely bad in the fglrx days!

My move happened few months ago. I used Linux for years (and still use it in VMs/servers), however after changing GPU it was a compatibility nightmare. I chose AMD partly because I heard that drivers are good on Linux. Not sure how many hours I spend on it.

Which card is this?

Re: AMD reckons it can win the high-end GPU battle

#40

Their software things (drivers, compilers, apis) are laughable.

Agreed! I posted this link here because I thought it was laughable, too. We do GPGPU programming (signal processing) and there's really no choice but Nvidia. I'd love there to be some competition, but AMDs developer tools and support are nothing compared to Nvidia. Plus Nvidia architecture is more "general purpose" than AMDs is for general computation
Post reply on HN