162
Nvidia bans using translation layers for CUDA software
(www.tomshardware.com)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 05 Mar 2024
162 points (97.1% liked)
Programming
17314 readers
347 users here now
Welcome to the main community in programming.dev! Feel free to post anything relating to programming here!
Cross posting is strongly encouraged in the instance. If you feel your post or another person's post makes sense in another community cross post into it.
Hope you enjoy the instance!
Rules
Rules
- Follow the programming.dev instance rules
- Keep content related to programming in some way
- If you're posting long videos try to add in some form of tldr for those who don't want to watch videos
Wormhole
Follow the wormhole through a path of communities !webdev@programming.dev
founded 1 year ago
MODERATORS
Jesus, they really are one of the most egregiously lock-in focused and monopolistic companies around. It saddens me deeply that consumers (gamers) just don't give a flying fuck about this and continues to pay a premium for Nvidia cards. 90% market dominance in gaming and probably at least that in GPGPU workloads.
All the while AMD tries to sell their cards on supporting / creating open standards like Freesync, FSR and Vulkan but because they don't have CUDA (since it's proprietary) they virtually can't be bought by prosumers that want to do some GPGPU stuff as a hobby and gamers buy Nvidia for brand recognition, Ray tracing which they are stronger in (but I argue isn't really all that outside a few notable exceptions like Alan Wake 2) and DLSS being ahead of FSR. But look at non-RT $/FPS and AMD wins easy at all price points and they don't shaft the people who bought their cards by not giving them the new version of DLSS like Nvidia do. It's just sad.
Vote with your wallet they scream, while everyone votes for the alternative that openly wants to squeeze every penny out of them because they are slightly better...
Vote with your wallet
I did.
I bought my 3080 back in 2020, because I knew AI was the future of graphics, based on all the R&D and white papers nVidia was pumping out up to that point.
No regrets.
Not my problem nVidia was the only one to invest in the tech, while AMD relied solely on TSMC to shrink their dies.
It has nothing to do with brand loyalty or fanboys or any of that shit.
It's just straight up better tech.
If you train AI models then you probably rely on CUDA and you're really left without any meaningful choice. It also wouldn't matter if AMD jumped 100% on AI even 5 years ago because CUDA has been so intensely adopted by the industry and AMD would need to do something completely novel and extremely impressive to have any chance of making a meaningful dent in just 5 years time.
As such I don't really blame you, as I said in my above post as well. I blame the gamers, the people that don't use CUDA and just play video games, the people complaining about how expensive GPUs have become while still fucking buying nVidia cards. The fact that AMD can deliver a product that costs less at the same performance point (without RT) is pretty impressive given their miniscule volumes compared to nVidia.