To really modify deep learning models, I’ve decided to take up Coursera’s Introduction To Concurrent Programming course. This will hopefully be a solid introduction to CUDA and give me a better sense of what operations I can do efficiently.

I’ve finished the first two modules (“Course Overview” and “Core Principles of Parallel Programming on CPUs and GPUs”). So far the course has some rough edges (unclear instructions, a link that didn’t work) but I think they start at a good place for making sure everyone is on the same page, know how to use the course’s code editor, and are going to use the same verbage.

I loved learning about Flynn’s Dichotomy. This was the first really new material for me and it feels like a useful structure to say “how complex is this current code, how can we make it less complex?”

Aiming to have this course done in the next 7 days. Could speed through it but I want to push forward on other parts of my job search and set myself up to use CUDA on my own machine after this course is done. Will make a post for each of the remaining modules. Wish me luck!