Wondering if we can impl this:
https://developer.nvidia.com/cuda/tile
https://docs.nvidia.com/cuda/tile-ir/
Seems a lot of people were interested in cuTile
https://developer.nvidia.com/blog/focus-on-your-algorithm-nvidia-cuda-tile-handles-the-hardware/
https://docs.nvidia.com/cuda/cutile-python/quickstart.html
https://github.com/NVIDIA/cutile-python
https://github.com/NVIDIA/TileGym
Industry now seems very interested in the productivity from this new kind of programming paradigm, which would be parallel with SIMT style in industry production for a while.
Early stage involvement would be excited.
Wondering if we can impl this:
https://developer.nvidia.com/cuda/tile
https://docs.nvidia.com/cuda/tile-ir/
Seems a lot of people were interested in cuTile
https://developer.nvidia.com/blog/focus-on-your-algorithm-nvidia-cuda-tile-handles-the-hardware/
https://docs.nvidia.com/cuda/cutile-python/quickstart.html
https://github.com/NVIDIA/cutile-python
https://github.com/NVIDIA/TileGym
Industry now seems very interested in the productivity from this new kind of programming paradigm, which would be parallel with SIMT style in industry production for a while.
Early stage involvement would be excited.