4 ms·

Umm, why not? We wrote entire NVIDIA, AMD, and QCOM drivers in that style. https://github.com/tinygrad/tinygrad/blob/master/tinygrad/runtime/ops_nv.py https:/

Umm, why not? We wrote entire NVIDIA, AMD, and QCOM drivers in that style. https://github.com/tinygrad/tinygrad/blob/master/tinygrad/runtime/ops_nv.py https://github.com/tinygrad/tinygrad/blob/master/tinygrad/ru... https://github.com/tinygrad/tinygrad/blob/master/tinygrad/runtime/ops_amd.py https://github.com/tinygrad/tinygrad/blob/master/tinygrad/ru... https://github.com/tinygrad/tinygrad/blob/master/tinygrad/runtime/ops_qcom.py https://github.com/tinygrad/tinygrad/blob/master/tinygrad/ru...

almostgotcaught 2y ago
Because it's slow duh
- georgehotz 2y ago
  This sounds like prejudice. Have you benchmarked it?
  - almostgotcaught 2y ago
    Yes I literally duplicated your approach for my driver stack last week and surprise surprise the FFI overhead into libc is too high.
    - georgehotz 2y ago
      FFI? This isn't how GPUs work...they are MMIO (mostly) Those drivers are faster than anything else when used to run fixed command queues (what neural network runs are)