CERN openlab Mini-Workshop on Floating-Point Computation

Tuesday 9 May 2017, 09:00 → 17:30 Europe/Zurich

31/3-004 - IT Amphitheatre (CERN)

Luca Atzori (CERN), Omar Awile (CERN)

The workshop

CERN openlab is pleased to announce a mini-workshop on floating-point computation and arithmetic to be offered at CERN on 9 May 2017.

The workshop will be run by Jeff Arnold, former Intel staff software engineer and a long-standing collaborator with CERN openlab. The target audience for this workshop are those who wish to gain a basic understanding of the fundamentals of floating-point arithmetic and computations. While extensive knowledge of floating-point arithmetic is not assumed, experience with scientific programming in C, C++ or Fortran is required.

Topics

The topics to be covered include:

• Basics of floating-point arithmetic on x64 architecture
• IEEE Standard for Floating-Point Arithmetic (IEEE 754-2008)
• Properties of floating-point arithmetic
• Error-Free Transformations and their uses
• Pitfalls in floating-point computations
• Impact of compilers and math libraries

The workshop will include hands-on exercises. Attendees are expected to have access to CERN computing facilities and to bring their own laptop to the workshop.

---

 

FloatingPoint-Exercises.pdf

FloatingPoint.Handout.pdf

1 Lecture

Speaker: Jeffrey Arnold (Fermi National Accelerator Lab. (US))

Recording

10:00 Break

2 Lecture

Speaker: Jeffrey Arnold (openlab/Fermi National Accelerator Lab. (US))

Recording

11:10 Break

3 Lecture and demonstrations

Speaker: Jeffrey Arnold (openlab/Fermi National Accelerator Lab. (US))

Recording

12:30 Lunch

4 Lecture and demonstrations

Speaker: Jeffrey Arnold (openlab/Fermi National Accelerator Lab. (US))

Recording

14:45 Break

5 Exercises

6 Wrap-up