3 ms·
Not CUDA itself, but lock-in of various libraries that are CUDA-only. No one stops you from making a library that works with OpenCL or Vulkan for example for co
by shmerl 8y ago
Not CUDA itself, but lock-in of various libraries that are CUDA-only. No one stops you from making a library that works with OpenCL or Vulkan for example for compute purposes.
- alfalfasprout 8y agoNo one stops you but it's a huge time investment. It's not only CUDA... it's the massive amount of optimized libraries NVIDIA provides. For example, CuBLAS (linear algebra libraries), CuDNN (libraries for deep learning). These things really need massive engineering teams to pull off and are very tied to the hardware.
- shmerl 8y agoSo there should be combined effort to make such libraries that are cross hardware. If many people need it, why don't they pool resources and create open libraries like that?