Week 11b: NumPy to CuPy

This past week we were able to successfully implement all of the processing with CuPy. While our previous workarounds for implementing the function argwhere were unsuccessful, we were able to pull the official version from GitHub and the CuPy team. However, through various tests, we found that the framing section of the code ran slower via the GPU in comparison to the CPU. Because this surprised us, we have opted to keep framing run by the CPU and continue with the other processes onto the GPU.

We are now in the process of working on the visualization code, however difficulties with mayavi prevent us from using the saved waveform and running the code. We essentially turned off visualization on our personal computer so that the code could be run and tested as we worked through processing. We’ll be looking to use the remote login to resolve this issue as that desktop has it functioning correctly.