Possible to port on GPU (CUDA)?

I wonder how hard it is to port some of the algorithm to use GPU eg CUDA?
I am a developer and i wonder if someone is already working on this?

We’ve a few GPU capabilities through Eugene DePrince’s (@deprince http://www.chem.fsu.edu/deprince/) dfcc plugin (https://github.com/edeprince3/gpu_dfcc/). If you’d want to try that, prompt us for installation directions.

I found the topic and installed gpu_dfcc. When I run the calculation with input.dat from the repository I got following error message.

Fatal Error: not enough memory (ccsd).
Error occurred in file: /scratch/psilocaluser/conda-builds/gpu_dfcc_1550120194473/work/ccsd.cu on line: 914

My PC has 16GB RAM and GTX1080Ti. Do I need more RAM for run the gpu_dfcc?


How much memory did you specify in the input file? In your output file, you should see a line right before the code failed that prints how much you need to increase the memory by.