AMD Quietly Funded A Drop-In CUDA Implementation Built On ROCm: It's Now Open-Source
Open link in next tab
AMD Quietly Funded A Drop-In CUDA Implementation Built On ROCm: It's Now Open-Source
https://www.phoronix.com/review/radeon-cuda-zluda
Nvidia is not gonna be happy lol they developed cuda to corner the market and build a monopoly based on their own framework get fucked nvidia lol.
Why is this project suddenly back after 3 years? What happened to Intel GPU support?
In 2021 I was contacted by Intel about the development od ZLUDA. I was an Intel employee at the time. While we were building a case for ZLUDA internally, I was asked for a far-reaching discretion: not to advertise the fact that Intel was evaluating ZLUDA and definitely not to make any commits to the public ZLUDA repo. After some deliberation, Intel decided that there is no business case for running CUDA applications on Intel GPUs.Shortly thereafter I got in contact with AMD and in early 2022 I have left Intel and signed a ZLUDA development contract with AMD. Once again I was asked for a far-reaching discretion: not to advertise the fact that AMD is evaluating ZLUDA and definitely not to make any commits to the public ZLUDA repo. After two years of development and some deliberation, AMD decided that there is no business case for running CUDA applications on AMD GPUs. One of the terms of my contract with AMD was that if AMD did not find it fit for further development, I could release it. Which brings us to today.
If you use CUDA, you can now use AMD GPUs instead of only being able to use Nvidia. Otherwise, it doesn’t mean much.
Yes I understand that part. I'm just wondering how. Will AMD release drivers? Third party drivers? No drivers, only software? What about software that just looks for Nvidia GPUs to use CUDA? Etc etc.
This is the best summary I could come up with:
While there have been efforts by AMD over the years to make it easier to port codebases targeting NVIDIA's CUDA API to run atop HIP/ROCm, it still requires work on the part of developers.
The tooling has improved such as with HIPIFY to help in auto-generating but it isn't any simple, instant, and guaranteed solution -- especially if striving for optimal performance.
In practice for many real-world workloads, it's a solution for end-users to run CUDA-enabled software without any developer intervention.
Here is more information on this "skunkworks" project that is now available as open-source along with some of my own testing and performance benchmarks of this CUDA implementation built for Radeon GPUs.
For reasons unknown to me, AMD decided this year to discontinue funding the effort and not release it as any software product.
Andrzej Janik reached out and provided access to the new ZLUDA implementation for AMD ROCm to allow me to test it out and benchmark it in advance of today's planned public announcement.
The original article contains 617 words, the summary contains 167 words. Saved 73%. I'm a bot and I'm open source!
In the context of the documentation "supported" means AMD will dedicate engineering resource to solving your issues, and that only applies to Instinct products.
"Unsupported" doesn't mean "it doesn't run". It means your bug reports are at the same level as every other commoner and could be ignored.
Does ROCm run on gfx1010 (5700 XT)? I believe so. Will you get any traction on bug reports? Probably not.