![Modern Mixed-Precision Methods: Hardware Perspectives, Algorithms, Kernels, and Solvers](https://img.swapcard.com/?u=https%3A%2F%2Fcdn-api.swapcard.com%2Fpublic%2Fimages%2F75766daf8d824b35bd81cc32ea635f1e.png&q=0.8&m=fit&h=720)
![](https://img.swapcard.com/?u=https%3A%2F%2Fcdn-api.swapcard.com%2Fpublic%2Fimages%2F75766daf8d824b35bd81cc32ea635f1e.png&q=0.8&m=crop&h=720)
Modern Mixed-Precision Methods: Hardware Perspectives, Algorithms, Kernels, and Solvers
Sunday, May 12, 2024 9:00 AM to 1:00 PM · 4 hr. (Europe/Berlin)
Hall Y6 - 2nd floor
Tutorial
Compiler and Tools for Parallel ProgrammingMixed PrecisionNovel AlgorithmsNumerical LibrariesParallel Programming Languages
Information
This tutorial will guide the audience through the rapidly expanding landscape of mixed- and multi- precision methods. The ongoing cross-pollination between high-performance computing (HPC) and machine learning (ML) is leading to intelligent computational steering of large-scale simulations. More importantly for this tutorial, sharing of the hardware platforms and exploiting their wide range of computational modes has led to proliferation of multiple representations of floating-point data. What followed was an increased interest in new methods that exploit them. Against the backdrop of high-performance libraries produced by internet-scale companies, hardware vendors, national laboratories, and academic institutions, we will show the recent algorithmic progress in exploiting multiple precisions for increased efficiency in achieved performance, required communication, and/or optimized storage needs. The techniques presented in the tutorial employ floating-point representations such as limited precision, quantized integers, and modular precision ecosystems, among others. Note that the lossless or lossy compression techniques are not covered as their algorithmic and accuracy considerations are substantially divergent to mixed-precision aspects. A portion of this tutorial will cover the fundamentals of the HPC software development in order to introduce the audience to some of the low-level details of coding for multiple precisions on modern hardware, including accelerators. The hardware focus of the tutorial will feature floating- point representation and performance of the recent supercomputing and industrial computing CPUs and accelerators, as well as less mainstream devices meant for more specific tasks at constraint power envelopes.
Format
On-site
Targeted Audience
Tutorial is targeted towards beginner- and intermediate-level audiences and is appropriate for graduate students, postdocs, and junior researchers. Some portion of the tutorial is devoted to more advanced topics, which may be of interest to seasoned researchers. Domain scientists may learn about recent trends in methodologies and their high-quality implementations.
Beginner Level
30%
Intermediate Level
70%
Prerequisites
The audience is expected to have a laptop connected to the internet to access presentation material. For hands-on exercises, the laptop will need to have development environment with HPC numerical libraries that are available for Linux, macOS and Windows systems. The prerequisites can be installed from scratch on Linux and macOS through Spack but it may take too long so the setup should be performed prior to the tutorial session.