The idea is to look at some technologies:

  • C++17 parallel execution
  • TBB (used by g++)
  • Intel AVX-512

Some example programs should be written to demonstrate the possible speed improvement using each technology and even combinations of the technologies.

Looking for hackers with the skills:

Nothing? Add some keywords!

This project is part of:

Hack Week 18

Activity

  • over 4 years ago: a_faerber liked this project.
  • almost 5 years ago: aschnell started this project.
  • almost 5 years ago: aschnell liked this project.
  • almost 5 years ago: aschnell originated this project.

  • Comments

    • aschnell
      over 4 years ago by aschnell | Reply

      I used the well-known mandelbrot set as a test program using parallel code and/or AVX-512 instructions. The speedup for parallel code is factor 6.7, for AVX-512 it is 7.6 and for both it is 49.8. For AVX it is important to know that the CPU frequency drops from 3.7 to 3.4 GHz.

      The CPU has 4 cores and hyper-threading. With AVX-512 eight double operations can be done with one instruction.

    Similar Projects

    This project is one of its kind!