Eileen Martin
|
Assistant Professor
Virginia Tech |
|
Eileen Martin is an assistant professor at Virginia Tech in the Department of Mathematics and Division of Computational Modeling and Data Analytics. She is a Luther and Alice Hamlett Junior Faculty Fellow in Virginia Tech’s Academy of Integrated Science. Her research focuses on computational science for applications in earth science, energy, and infrastructure. She serves as an associate editor for Computers & Geosciences. She earned her PhD in computational and mathematical engineering at Stanford in 2018, where she was a member of the Stanford Exploration Project seismic imaging group. She holds an MS in geophysics from Stanford, and a BS with a double-major in mathematics and computational physics from UT-Austin.
|
Workshop: Why we love arrays for data science
Prerequisite: Basic flow control (for/while loops, nested loops) and vector addition, inner products/dot products, matrix-vector multiplication
In this tutorial-style workshop, we'll walk through some of the basics of computer architecture and how it affects the performance of our codes for common data analysis techniques. The particular features we will explore using timing and memory profiling are memory hierarchy (cache levels and DRAM), movement of data in cache lines, and data prefetchers. We will develop some basic understanding of computer architecture, and a few strategies to improve the performance of our codes to take advantage of these common architecture features, particularly when working with data in structured arrays.
In this tutorial-style workshop, we'll walk through some of the basics of computer architecture and how it affects the performance of our codes for common data analysis techniques. The particular features we will explore using timing and memory profiling are memory hierarchy (cache levels and DRAM), movement of data in cache lines, and data prefetchers. We will develop some basic understanding of computer architecture, and a few strategies to improve the performance of our codes to take advantage of these common architecture features, particularly when working with data in structured arrays.