Hacker News new | ask | show | jobs
by 26fingies 991 days ago
very nice but i assume it’s going to be limited to cuda as a backend because nvidia
3 comments

Currently we support both CUDA and CPU to some extent. CPU is done through standard C++ (and soon stdpar). Obviously standard C++ is problematic since it doesn't include everything we support (FFTs, matrix multiplies, etc). One option is to use open-source libraries that do these, but then it ends up being a lot of dependencies that are hard to manage. We have plans to improve CPU support soon, so stay tuned.
Vulkan Compute would be nice.
I don't actually know a lot about massively parallel libraries like CUDA. Does AMD have an equivalent technology associate with their GPUs? It feels like it should be fairly straightforward to create some kind of high level library that just uses CUDA or whatever AMD has on the back end.
Traditionally OpenCL was the alternative to CUDA. Recently AMD has been pushing their ROCm platform.
Others are free to provide proper C++ support on their GPUs.
They're also free to write math libraries that don't depend on GPUs.
Hi, MatX currently has partial support for CPUs too. Please see this comment:

https://news.ycombinator.com/item?id=37758635

The performance boost you get from using a GPU is incredible.

You might as well say they're free to make monitors that don't rely on color.

GPU performance per dollar is only competitive for specific workloads. For extremely large scale compute, getting enough data center GPUs can also be challenging.