Are you really surprised? You can't even really blame their GPU division, they're like children claiming they're the greatest artist ever after presenting you their crayon painting and the kids parents go along with it because why not?
No one professional uses amd shit. They arent even close to equivalent. And it has nothing to do with the hardware, its all software and qc (or lack thereof)
It's hardware, software and qc. To act like AMD is even a competitor to Nvidia when it comes to ML or any other professional use case is beyond delusional.
>And it has nothing to do with the hardware
I've always had my suspicions thay AMD ships lackluster silicon.
It's why some people can have zero trouble and others have things be a nightmare.
>And it has nothing to do with the hardware, its all software
What language do you code in for your gpu work then? I'm going to say none because these threads are obvious shill bait and you people should be killed.
Does anyone actually use rocm for AI development? Or are all their 'partners' the equivalent of microsoft 'investing' 1 million dollars into Rust: a dusty server sitting in a corner just so they can say AMD is somehow still relevant.
I've tried installing rocm for my 6600xt for so long I don't even remember. Even tried the pre-made docker container, didn't work.
Got it working on my 7600 just fine. Used the arch packages.
May install arch just for it. Thanks
Friendly dev more tools for AMD
AMD can say literally anything and no one will ever call them on their BS
Are you really surprised? You can't even really blame their GPU division, they're like children claiming they're the greatest artist ever after presenting you their crayon painting and the kids parents go along with it because why not?
it is probably true, it's just developers not using their tools and only focusing in nvidia toolsets because team green has a way bigger market share.
>it is probably true
It's a fricking gpu you morons they're all the same at the hardware level just buy the fastest one you can afford.
No one professional uses amd shit. They arent even close to equivalent. And it has nothing to do with the hardware, its all software and qc (or lack thereof)
It's hardware, software and qc. To act like AMD is even a competitor to Nvidia when it comes to ML or any other professional use case is beyond delusional.
>And it has nothing to do with the hardware
I've always had my suspicions thay AMD ships lackluster silicon.
It's why some people can have zero trouble and others have things be a nightmare.
>And it has nothing to do with the hardware, its all software
What language do you code in for your gpu work then? I'm going to say none because these threads are obvious shill bait and you people should be killed.
read: "please write an AMD competitor to CUDA for us for free you stupid goy"
hipify+rocm lets it run 99% of nvidia exclusive cuda shit
WOULD!
Does rocm only still only support vega?
>Absolute frickabout to use AMD for AI on Linux
>Pretty crap even if you can get it working
Fricking hell AMD -_-
Does anyone actually use rocm for AI development? Or are all their 'partners' the equivalent of microsoft 'investing' 1 million dollars into Rust: a dusty server sitting in a corner just so they can say AMD is somehow still relevant.