Updated 2021-06-17

Optimization 101

What: Benchmarking, Profiling, and Optimization a serial linear algebra problem.

When: As posted below.

Where: Teams Link send before class.

Registration: Here

Materials available at GitHub

Warning

To fully benefit from this course, users should have an active PACE account, and either the Windows Subsystem for Linux or WinSCP prior to attendance. Please contact PACE support (pace-support@oit.gatech.edu) with any questions.

Presentation topics covered include:

  • What is optimization?
  • Benchmarking
  • Profiling using
  • GDB
  • Intel Amplifier
  • DDT
  • TAU
  • Optimization using compiler options
  • Optimization using algorithm changes
  • Using Libraries

Format: 105 minutes of presentation and hands-on demonstrations. This course will use a problem-directed approach, focusing on matrix multiplication in the context of systems of linear equations solvers.

The course is recommended for researchers with a background in compiled languages such as C/C++ or Fortran. Familiarity with a programming language such the attendee will be able to read follow code with guidance. Great familiarity is encourages, but not required.

Please contact PACE support (pcae-support@oit.gatech.edu) if you have any questions or concerns.

We are hoping to see you there!

Upcoming Course Offerings

  • Friday, June, 18th 2021 from 2:00-3:45PM via Bluejeans. Register here. Registrants will be sent a link to the Teams meeting.
  • Friday, July, 16th 2021 from 2:00-3:45PM via Bluejeans. Register here. Registrants will be sent a link to the Teams meeting.

Note

Some sessions may be canceled if the attendance is insufficient. If that happens, registered users will be moved automatically to the next session to guarantee a seat (and will be notified via email).