Tutorial on Floating-Point Analysis and Reproducibility Tools for Scientific Software

Los Alamos National Laboratory

January, 2020
Time: 9:00am - 12:15pm (half-day tutorial)

Part of the this tutorial was presented at SC19 (see here)

Presenters

Ignacio Laguna (organizer), Lawrence Livermore National Laboratory
Michael Bentley, University of Utah
Ian Briggs, University of Utah
Ganesh Gopalakrishnan, University of Utah

Schedule

Time	Module	Presenter	Slides
9:00 - 9:10am	Introduction (housekeeping)	Ignacio	slides
9:10 - 9:30am	Floating-point background	Ganesh	slides
9:30 - 10:00am	FPChecker: floating-point exceptions, GPUs, CUDA	Ignacio	slides
10:00 - 10:15am	Break or Q&A
10:15 - 10:45am	ARCHER: data races, OpenMP	Ian	slides
10:45 - 11:30am	FLiT: floating-point variability, compiler optimizations	Ian	slides
11:30 - 12:00pm	ReMPI: MPI, floating-point variability	Ignacio, Ian	slides
12:00 - 12:15pm	Q&A

Access to AWS Instances

We provide exercises for each module in AWS instances. You can get a username, password, along with the IP address of an instance here. You can access the instance via ssh as follows:

ssh [USERNAME]@[IP ADDRESS]

Directory Structure

The directory structure in your home directory will look like this:

/home/user1/
   |---Module-TOOL1
      |---exercise-1
      |---exercise-2
      |---exercise-3
   |---Module-TOOL2
      |---exercise-1
      |---exercise-2
      |---exercise-3
   ...

Each module directory contains directories for each exercise. Please run the exercises in each of them as indicated by the speaker.

Tutorial on Floating-Point Analysis Tools