MAX/Mojo to "replace CUDA", Chris Lattner's new project after Swift, LLVM, TPUs

[Dylan Bethune-Waddell]

Hi All,

Thought it worth mentioning this project that could go nicely with BUDA in theory. Video linked below is heavy on the marketing slant, unlike most on their Youtube posts - especially the community meetings which are great - but it does highlight grand promises made for the last few years by the team at Modular being delivered on and AMD jumping on the bandwagon in some official sense. Apparently, they are providing a "free" MI300/325/355 playground where you can sign up by email and test-drive the stack. I sent in the form but haven't heard back yet:
https://youtu.be/TrBXHPGRlnQ?si=KBH9HVjO77_qIcJX

So far, only bleeding edge NVIDIA and AMD cards are officially supported, although 30xx and 40xx ostensibly work anyways. Motivation for WCG was the potential of their relatively lightweight, no-CUDA, same code runs on AMD and NVIDIA GPUs container + BUDA for new projects. Chris Lattner did a series of blog posts on the Modular community forums to outline the case for/intent of the project, which are probably a better read than the video above is a watch. Here is the latest (part 10) with links to all the preceding parts 1-9 as well:
https://forum.modular.com/t/modular-s-bet-to-break-out-of-the-matrix-democratizing-ai-compute-part-10/1405

Best,
Dylan