JUNE 18–22, 2017
FRANKFURT AM MAIN, GERMANY
FRANKFURT AM MAIN, GERMANY
Presentation Details |
|||||
| Name: |  |
(RP05) Optimizing Smoothed Particle Hydrodynamics Code Phantom on Haswell & KNL | |||
| Time: | |
Tuesday, June 20, 2017 08:35 am - 09:45 am |
|||
| Room: | Substanz 1+2 | ||||
| Breaks: | 07:30 am - 10:00 am Welcome Coffee | ||||
| Presenter: | Charlene Yang, Pawsey Supercomputing Centre | ||||
| Abstract: | This poster presents the optimization work carried out on a Smoothed Particle Hydrodynamics (SPH) code Phantom on Haswell and KNL. With 8 steps of optimization, the performance results show 3x speedup on Haswell and 4x on KNL. A few remarks are provided to highlight the differences on KNL in terms of architecture and optimization strategy. This work will also serve as an example of optimizing other codes, such as those in molecular dynamics, computer graphics and neural networks that also deal with kd-tree, neighbour search, irregular computation and random memory accesses. Authors: Charlene Yang, Pawsey Supercomputing Centre Ralph Christopher Bording, Pawsey Supercomputing Centre Daniel Price, Monash University Rebecca Nealon, Monash University |
||||
| Download | RP05_Yang.pdf (2385 KB) |
||||
| |||||