From the course: Google Cloud Professional Machine Learning Engineer Cert Prep

Unlock this course with a free trial

Join today to access over 25,300 courses taught by industry experts.

Building a CUDA GPU stress test

Building a CUDA GPU stress test - Google Cloud Platform Tutorial

From the course: Google Cloud Professional Machine Learning Engineer Cert Prep

Start my 1-month free trial Buy for my team

Building a CUDA GPU stress test

“

One of the more powerful ways to use Rust is to build a systems tool that talks to a GPU. Fortunately, because PyTorch bindings work so well with Rust, what I'm going to do is piece together a tool that can not only talk to the CPU and saturated via PyTorch, but also can talk to a CUDA enabled GPU saturated with PyTorch and then use some of the advantages of Rust, which is the true cause that allows you to spawn a pool of threads and then send data into a GPU to try to get the most out of it in terms of a stress test. Let's go ahead and build that tool in just a few seconds. Let's take a look at the architecture of how you could build a stress test tool for a CUDA enabled GPU by using the systems programming capabilities of Rust and the helpful rust-pytorch bindings. First up, we have to have access to a CUDA enabled GPU with GitHub Codespaces. That's one way to do it. You could also be a AWS instance or GCP instance or an Azure…

Contents