• 0 Posts
  • 15 Comments
Joined 1 year ago
cake
Cake day: June 11th, 2023

help-circle






  • Yup, been writing a new shader language to replace GLSL and HLSL for Vulkan Compute purposed, but I eventually switch from SPIR-V IR to MLIR and use IREE Compiler which accepts the MLIR and compile it to any of CUDA, ROCm, SPIR-V and so forth.

    A lot of it was because of my unadulterated hatred toward our current Machine Learning Frameworks...

    It’s one of the project that I’ve been working on to outright replace Pytorch/Tensorflow and ban those two framework from my office forever. I got fed up not knowing how much exactly do I need in memory allocation, computational cost, and so forth when running or training neural net models. Plus I want an easier way to split the model across lower-end GPU too that doesn’t rely on Nvidia-only GPU for CUDA code. I also wanted to have SPIR-V as a fallback compute kernel, because if CUDA/ROCm is too new for GPU, you’re SOL, but if you have SPIR-V, chances are, any GPU made in the last 10 years that have a Vulkan Driver, would likely be supported.



  • I agree on avoiding on the idea of avoiding having to make your own parser generator, this is precisely what I’m doing and it’s hell. I assumed that you probably want to pick up some understanding on how parser differs when it come to writing grammars. As for ease of use and requiring the least understanding, using something like Earley parser is probably the easiest, it would be slower than other parser algorithms, but it could handle ambiguous grammars making it ideal for first timers to learn how to write a programming language.


  • Yep, and if open source licensing could be revoked on a whim, you can imagine the chaos that ensued. That would be my understanding as well, old version that have MPL license is perfectly fine to fork off, newer version might not be as it is under a different license. One of the reason why I liked Apache License is that it have make it explicitly clear that it’s irrevocable whereas MPL it is operating on an assumption that it’s not revocable. The most fundamental problem with the legal system in USA is that no law is “set in stone” and leaving things to assumption is open to reinterpretation by the judge who may have sided against you. (Hell, Google vs Oracle on Copyrighted API is still on case-to-case basis, so take it as you will.)

    Disclaimer: I am not a lawyer. I just share what I learned from Legal Eagle youtube and few other sources.








  • I tried to use it, but it have some big issues in reliability, because at the end of the day, despise the dataset it’s trained on, it’s still something I describes as a “language interpolation.”

    It sometime make TERRIBLE recommendations for which tools/libraries I should explore, because it assumes that those libraries might have support. Those libraries never does and so I wasted weeks on it. (It doesn’t help that both code and project are undocumented.)

    So after that experience, I demote ChatGPT usefulness to just “cleaning up pre-written documentation so it sounds better.” That’s it.