It’s no surprise that NVIDIA is gradually dropping support for older videocards, with the Pascal (GTX 10xx) GPUs most recently getting axed. What’s more surprising is the terrible way t…
When people switch to Linux they don’t do a lot of research beforehand. I, for one, didn’t know that Nvidia doesn’t work well with it until I had been using it for years.
To be fair, Nvidia supports their newer GPUs well enough, so you may not have any problems for a while. But once they decide to end support for a product line, it’s basically a death sentence for that hardware. That’s what happened to me recently with the 470 driver. Older GPU worked fine until a kernel update broke the driver. There’s nobody fixing it anymore, and they won’t open-source even obsolete drivers.
I JUST ran into this issue myself. I’m running Proxmox on an old Laptop and wanted to use its 750M…. Which is one of those legacy cards now that I guess means I’d need to downgrade the kernel to use?
I’m not knowledgeable enough to know the risks or work I’d be looking at to get it working so for now, it’s on hiatus.
Even now, CUDA is gold standard for data science / ML / AI related research and development. AMD is slowly brining around their ROCm platform, and Vulcan is gaining steam in that area. I’d love to ditch my nvidia cards and go exclusively AMD but nvidia supporting CUDA on consumer cards was a seriously smart move that AMD needs to catch up with.
CUDA is an Nvidia technology and they’ve gone out of their way to make it difficult for a competitor to come up with a compatible implementation. With cross-vendor alternatives like OpenCL and compute shaders, they’ve not put resources into achieving performance parity, so if you write something in both CUDA and OpenCL, and run them both on an Nvidia card, the CUDA-based implementation will go way faster. Most projects prioritise the need to go fast above the need to work on hardware from more than one vendor. Fifteen years ago, an OpenCL-based compute application would run faster on an AMD card than a CUDA-based one would run on an Nvidia card, even if the Nvidia card was a chunk faster in gaming, so it’s not that CUDA’s inherently loads faster. That didn’t give AMD a huge advantage in market share as not very much was going on that cared significantly about GPU compute.
Also, Nvidia have put a lot of resources over the last fifteen years into adding CUDA support to other people’s projects, so when things did start springing up that needed GPU compute, a lot of them already worked on Nvidia cards.
People buy Nvidia for different reasons, but not everyone faces any issues with it in Linux, and so they see no reason to change what they’re already familiar with.
I’m with you, I know we’ve had a lot of recent Linux converts, but I don’t get why so many who’ve used Linux for years still buy Nvidia.
Like yeah, there’s going to be some cool stuff, but it’s going to be clunky and temporary.
When people switch to Linux they don’t do a lot of research beforehand. I, for one, didn’t know that Nvidia doesn’t work well with it until I had been using it for years.
It’s a good way for people to learn about fully hostile companies to the linux ecosystem.
To be fair, Nvidia supports their newer GPUs well enough, so you may not have any problems for a while. But once they decide to end support for a product line, it’s basically a death sentence for that hardware. That’s what happened to me recently with the 470 driver. Older GPU worked fine until a kernel update broke the driver. There’s nobody fixing it anymore, and they won’t open-source even obsolete drivers.
I JUST ran into this issue myself. I’m running Proxmox on an old Laptop and wanted to use its 750M…. Which is one of those legacy cards now that I guess means I’d need to downgrade the kernel to use?
I’m not knowledgeable enough to know the risks or work I’d be looking at to get it working so for now, it’s on hiatus.
You might be able to use the Nouveau driver with the 750M. Performance won’t be great, but might be sufficient if it’s just for server admin.
Even now, CUDA is gold standard for data science / ML / AI related research and development. AMD is slowly brining around their ROCm platform, and Vulcan is gaining steam in that area. I’d love to ditch my nvidia cards and go exclusively AMD but nvidia supporting CUDA on consumer cards was a seriously smart move that AMD needs to catch up with.
Sorry for prying for details, but why exactly do you need NVIDIA?
CUDA is an Nvidia technology and they’ve gone out of their way to make it difficult for a competitor to come up with a compatible implementation. With cross-vendor alternatives like OpenCL and compute shaders, they’ve not put resources into achieving performance parity, so if you write something in both CUDA and OpenCL, and run them both on an Nvidia card, the CUDA-based implementation will go way faster. Most projects prioritise the need to go fast above the need to work on hardware from more than one vendor. Fifteen years ago, an OpenCL-based compute application would run faster on an AMD card than a CUDA-based one would run on an Nvidia card, even if the Nvidia card was a chunk faster in gaming, so it’s not that CUDA’s inherently loads faster. That didn’t give AMD a huge advantage in market share as not very much was going on that cared significantly about GPU compute.
Also, Nvidia have put a lot of resources over the last fifteen years into adding CUDA support to other people’s projects, so when things did start springing up that needed GPU compute, a lot of them already worked on Nvidia cards.
People buy Nvidia for different reasons, but not everyone faces any issues with it in Linux, and so they see no reason to change what they’re already familiar with.