Building a Personal Computer for AI and Data Science

Your build guide for constructing your personal home computer for AI and Data Science.There are already a lot of build guides out there for deep learning..And some here and there for reinforcement learning..But very very few for all of them.That’s why I wrote this guide, for people who were, or may at some point be, interested in all of these fields!How do we optimise our build for all areas of AI and Data Science?With the whole deep learning buzz lately, everyone’s been talking about how important GPUs are..It’s true, GPUs are the major force behind deep learning and are absolutely necessary for doing work in the field.Unfortunately though, people have pushed the CPU, RAM, motherboard, storage, and case selections off to the side..That’s not too bad for deep learning specifically if that’s all you’re doing, but if you’re also interested in other fields, you’ll also need to consider the other PC components!In this build guide, we’re going to walk through building a personal computer that can do well with deep learning, data science, reinforcement learning, and all fields of AI pretty darn well..It comes with 11GB of vRAM and usually runs between 1600~1700mHz.A lot of people in the deep learning community advocate for the Titan xp because it comes with 12GB of vRAM..People say it’s fine since if you’re only doing deep learning, most of the processing is happening on the GPU anyways..But for reinforcement learning and data science, much of the processing happens on the CPU!.This is great for pre-processing as well as if you go with multiple GPUs for deep learning..People in the deep learning community often go with “2 times your GPU memory!”..I don’t really work with data sets where I have to load 32GB into memory all at once..Does that mean our board and CPU will never work if we want 2 GPUs….Not at all!X99 MotherboardMost people really over think the PCIe lanes..It’s simply not worth it.In addition, most people will hardly need multiple GPUs.. More details

Leave a Reply