System Setup for GPU Performance Measurements
In this document, I outline the steps to set up a system to take NVIDIA GPU performance measurements. This could be required, for example, to analyze and improve the performance of an LLM.
This document is the converse of a friction log. I ran into many obstacles getting this to work, so I have noted here a least-friction, paved path to profiling NVIDIA GPUs.
High Level Outline First, we use a remote VM in Google Cloud that is attached to an NVIDIA GPU.
[Read More]