LANL is hosting an Accelerated Python Tutorial, presented by Scot Halverson from Nvidia. This event is open to NERSC users.
Description
In this interactive tutorial, users will learn tools for accelerated computing in Python. This will include measuring, understanding, and improving performance of their python applications, including ML workflows using PyTorch and Tensorflow. We’ll cover native Python tools, PyTorch and Tensorflow specific tools, and NVIDIA provided tools. We’ll also cover tools for leveraging NVIDIA GPUs with traditional compute intensive applications including Numba and CuPy.
Topics Covered
- Basic Python Profiling
- PyTorch and Tensorflow Profiling
- NVIDIA Nsight Systems and Nsight Compute Profilers
- ML-focused optimization strategies
- Prerequisites
Familiarity with Python and basic familiarity with either PyTorch or Tensorflow.
Speaker Bio
Scot Halverson is a Senior Solutions architect for HPC and AI at NVIDIA. Scot has focused his career on accelerated computing, most recently as a member of the Supercomputing team working closely with national labs within the DOE and other HPC centers. Prior to joining NVIDIA, Scot spent nearly 10 years at Los Alamos National laboratory across a number of Divisions including X Computational Physics (XCP), Computer and Computational Sciences (CCS), and Analytics Intelligence and Technology (A) where he applied parallel and accelerated computing methods to a variety of LANL’s computational workloads. Scot has a Masters degree in Computer Science, as well as undergraduate degrees in computer science and Chemistry from the University of Minnesota.
Registration
Please register now. A calendar invitation will be sent in advance of the training event.