Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

A lone developer just open sourced a tool that could bring an end to Nvidia's AI hegemony — AMD financed it for months but abruptly ended its support. Nobody knows why

Feb 24, 2024 - techradar.com
Andrzej Janik, a developer who created a tool allowing Nvidia's CUDA code to run on AMD and Intel GPUs, has open-sourced his project, ZLUDA, after both AMD and Intel withdrew their support. The tool, which was updated to version 3 last week, was initially developed in 2020 to enable Intel GPUs to run CUDA, a proprietary hardware-software ecosystem that supports Nvidia's dominance. However, Intel decided there was no business case for running CUDA applications on its GPUs and requested the project be taken down.

After leaving Intel, Janik was contracted by AMD to continue ZLUDA development. However, AMD also concluded there was no business case for running CUDA applications on its GPUs and ended its support for the project. The current version of ZLUDA, now based on AMD's ROCm solution and only supporting Radeon GPUs, has been released back to the public domain. Despite the lack of corporate backing, ZLUDA has shown promise in testing, with many CUDA software able to run on HIP/ROCm without modifications.

Key takeaways:

  • Developer Andrzej Janik has open sourced his tool, ZLUDA, that allows Nvidia's CUDA code to run on AMD and Intel GPUs without modifications.
  • Both AMD and Intel have dropped support for the project, citing no business case for running CUDA applications on their GPUs.
  • ZLUDA is now based on AMD's ROCm solution and only supports Radeon GPUs, with Janik stating that the project is more or less complete.
  • Despite lack of corporate backing, ZLUDA has shown promise in testing, with many CUDA software able to run on HIP/ROCm without any modifications.
View Full Article

Comments (0)

Be the first to comment!