These codes illustrate how to use shared memory parallelism with OpenMP. The algorithm used here is the same as that used on the GPU: it divides particles into small tiles and reorders them every time step to avoid data collisions when performing the deposit. Each tile is controlled by a single thread. The algorithm is described in detail in Skeleton Codes Ref. [4].

For the 2D electrostatic code, a typical execution time for the particle part of this code is about 3 ns/particle/time-step with 12 processing cores. For the 2-1/2D electromagnetic code, a typical execution time for the particle part of this code is about 10 ns/particle/time-step with 12 processing cores. For the 2-1/2D Darwin code, a typical execution time for the particle part of this code is about 22 ns/particle/time-step with 12 processing cores. (All timings are on a 2.67GHz Intel Nehalem processor.)

For the 3D electrostatic code, a typical execution time for the particle part of this code is about 6 ns/particle/time-step with 12 processing cores. For the 3D electromagnetic code, a typical execution time for the particle part of this code is about 15 ns/particle/time-step with 12 processing cores. For the 3D Darwin code, a typical execution time for the particle part of this code is about 41 ns/particle/time-step with 12 processing cores. (All timings are on a 2.67GHz Intel Nehalem processor.)

Electrostatic

1. 1D Parallel Electrostatic Spectral code: mpic1

2. 2D Parallel Electrostatic Spectral code: mpic2

3. 3D Parallel Electrostatic Spectral code: mpic3

Electromagnetic

4. 1-2/2D Parallel Electromagnetic Spectral code: mbpic1

5. 2-1/2D Parallel Electromagnetic Spectral code: mbpic2

6. 3D Parallel Electromagnetic Spectral code: mbpic3

Darwin

7. 1-2/2D Parallel Darwin Spectral code: mdpic1

8. 2-1/2D Parallel Darwin Spectral code: mdpic2

9. 3D Parallel Darwin Spectral code: mdpic3

Want to contact the developer? Send mail to Viktor Decyk at decyk@physics.ucla.edu.