Computer Organization and Design
Section 6.6
Computer Architecture: A Quantitative Approach
Section 4.4
GPUs are one of the most pervasive accelerators. Not only do all of your computers have a GPU, but even our phones have GPUs. In fact, GPUs are usually one of the largest components on our processors!
So, let’s learn a bit about how GPUs work.
Next, let’s talk about some other accelerators. In this video, I talk about the TPU or tensor processor unit which is a completely novel architecture design from Google and other new architectures. I talk some about the “roofline” model. However, we haven’t covered that this quarter. If you want to find out more about the roofline model, it’s described in Section 6.10 of the book.
Up next… ECS201C the graduate computer architecture course!.