Which setup is more efficient for running localcolabfold: personal computer with GPU or CPU-based server?

[ShailNair]

Hi, thank you for making AlphaFold easy to run locally.

My office has a Linux server with 105 cores and 500 GB of RAM, but no GPU. My personal computer contains WSL2 Ubuntu, 16 cores, 50 GB of RAM, and an NVIDIA RTX 3090 (24 GB of VRAM). I am wondering which setup would run faster for localcolabfold when processing 600 proteins (150-500 aa in length), the server or my personal computer?

Currently, my WSL2 setup with the RTX 3090 processes 3-5 proteins per day.