Group Specific Component

CuTile

Our ideas were:

trying to reproduce the paper https://www.mdpi.com/2079-9292/15/5/1034 (speedups of kernel fusing)
trying to implement + benchmark some ideas of this https://developer.download.nvidia.com/video/gputechconf/gtc/2019/presentation/s9593-cutensor-high-performance-tensor-operations-in-cuda-v2.pdf (e.g. more tiling for reuse on different hardware levels)