fp.

Including stuff like Fortran, Haskell, Java, .NET via PTX, Python JIT, IDE tooling integration with major IDEs, graphical GPU debugging and profiling, libraries and co?

Then I guess all the best.

zorked•14m ago

This post has some serious peanut-gallery vibes.

embedding-shape•42m ago

Ambitions but neat, good luck if nothing else :)

If you were to guess, when do you think your Nsight Compute alternative might be ready with your own toolchain?

maxloh•26m ago

There is also ZLUDA, which is open source and works on pre-compiled binaries.

https://github.com/vosen/ZLUDA

DiabloD3•15m ago

Its easier to just get rid of your legacy code entirely and use Vulkan for compute, or have your compiler emit SPIR-V directly.

No reason to tie yourself to Nvidia's moat.

swerner•9m ago

Unfortunately, Vulkan Compute doesn’t to all the things that OpenCL, SYCL, HIP or CUDA do.

lulzx•14m ago

I have been trying for cuda -> metal, to run it on mac, https://github.com/lulzx/cuda-metal

luciana1u•9m ago

every CUDA alternative follows the same arc: bold launch, works for 3 operations, then a Discord server where the last message is 'any updates?' from 2024

woctordho•9m ago

There's nothing wrong to run CUDA on non-Nvidia hardware. CUDA has an interface that is reasonably well-designed, well-documented/reverse-engineered, and battle-tested for decades. What we need is not to invent another interface just under the name of 'open standard', but to implement the same interface. ROCm is exactly doing this, and so are other hardware SDKs such as MooreThread and Alibaba T-Head.