OpenCL-Wrapper VS CUDA.jl

In case you go with OpenCL, start here: https://github.com/ProjectPhysX/OpenCL-Wrapper

I've also open-sourced an OpenCL-Wrapper to eliminate all of the boilerplate code that otherwise comes with the OpenCL C++ bindings and lower the entry barrier. Especially for larger projects, the biolerplate code becomes really offputting, and I solved it entirely.

Somewhat unusual these days, but I mainly use OpenCL C. It's seems cumbersome and hard to learn at first, but becomes much more easy to use with the right tools. Once you master it, it whipes the floor with CPU programming; it's not unusual to see 100x speedup on a GPU compared to multithreaded CPU code at the same energy consumption. It's just as fast as CUDA - as efficient as the microarchitecture allows - but compatible with literally all GPU/CPU hardware of the last decade. No need to waste time on code porting if the next supercomputer has GPUs from a different vendor, it just runs out-of-the-box. Ideal for scientific compute!

I added this to my OpenCL-Wrapper in this commit, so anything built on top of it, such as FluidX3D, works on Arc out-of-the-box. Additionally, I fixed Intel's wrong VRAM capacity reporting on Arc in this patch.

Try OpenCL. You only need to implement the code once (in a vectorized form) and it works cross-platform on all GPUs and all CPUs, even on FPGAs. Performance is exactly as good as CUDA. There is still no rivaling framework today, although SYCL is starting to become a viable alternative.

I have found that OpenCL-Wrapper from PhysX has a great solution to this : https://github.com/ProjectPhysX/OpenCL-Wrapper/

It would also mean learning Julia, but you can write GPU kernels in Julia and then compile for NVidia CUDA, AMD ROCm or IBM oneAPI.

https://juliagpu.org/

I've written CUDA kernels and I knew nothing about it going in.

See https://juliagpu.org/

If you look at Julia open source projects you'll see that the projects tend to have a lot more contributors than the Python counterparts, even over smaller time periods. A package for defining statistical distributions has had 202 contributors (https://github.com/JuliaStats/Distributions.jl), etc. Julia Base even has had over 1,300 contributors (https://github.com/JuliaLang/julia) which is quite a lot for a core language, and that's mostly because the majority of the core is in Julia itself.

This is one of the things that was noted quite a bit at this SIAM CSE conference, that Julia development tends to have a lot more code reuse than other ecosystems like Python. For example, the various machine learning libraries like Flux.jl and Lux.jl share a lot of layer intrinsics in NNlib.jl (https://github.com/FluxML/NNlib.jl), the same GPU libraries (https://github.com/JuliaGPU/CUDA.jl), the same automatic differentiation library (https://github.com/FluxML/Zygote.jl), and of course the same JIT compiler (Julia itself). These two libraries are far enough apart that people say "Flux is to PyTorch as Lux is to JAX/flax", but while in the Python world those share almost 0 code or implementation, in the Julia world they share >90% of the core internals but have different higher levels APIs.

If one hasn't participated in this space it's a bit hard to fathom how much code reuse goes on and how that is influenced by the design of multiple dispatch. This is one of the reasons there is so much cohesion in the community since it doesn't matter if one person is an ecologist and the other is a financial engineer, you may both be contributing to the same library like Distances.jl just adding a distance function which is then used in thousands of places. With the Python ecosystem you tend to have a lot more "megapackages", PyTorch, SciPy, etc. where the barrier to entry is generally a lot higher (and sometimes requires handling the build systems, fun times). But in the Julia ecosystem you have a lot of core development happening in "small" but central libraries, like Distances.jl or Distributions.jl, which are simple enough for an undergrad to get productive in a week but is then used everywhere (Distributions.jl for example is used in every statistics package, and definitions of prior distributions for Turing.jl's probabilistic programming language, etc.).

If you just want to do some numerical code that requires linear algebra and GPU, your best bet would be Julia or Python+JAX.

GitHub - JuliaGPU/CUDA.jl: CUDA programming in Julia.

Should have (2017) in the title.

Indeed cool to program julia directly on the GPU and Julia on GPU and this has further evolved since then, see https://juliagpu.org/

I'm excited to eventually see something like JuliaGPU with support for multiple backends.

That's really cool! Have you looked at CUDA.jl for the Julia language? Maybe you could take some ideas from there. I am pretty sure it does the same thing you do here, and they support any arbitrary code with the limitations that you cannot allocate memory, I/O is disallowed, and badly-typed code(dynamic) will not compile.