Difference between revisions of "Eigen paralelization"
From Medusa: Coordinate Free Mehless Method implementation
| (One intermediate revision by one other user not shown) | |||
| Line 4: | Line 4: | ||
Exectuion on PC with 4 cores and 8 hyperthreads: | Exectuion on PC with 4 cores and 8 hyperthreads: | ||
| + | |||
[[File:speedups_Ax.png|800px]] | [[File:speedups_Ax.png|800px]] | ||
[[File:l2hit.png|800px]] | [[File:l2hit.png|800px]] | ||
[[File:l3hit.png|800px]] | [[File:l3hit.png|800px]] | ||
| − | Execution on Intel | + | Execution on Intel CPU E5-2620: |
| + | |||
[[File:speedups_phi.png|800px]] | [[File:speedups_phi.png|800px]] | ||
[[File:l2hit_phi.png|800px]] | [[File:l2hit_phi.png|800px]] | ||
[[File:l3hit_phi.png|800px]] | [[File:l3hit_phi.png|800px]] | ||
Latest revision as of 11:20, 28 March 2017
General Eigen parallelism: https://eigen.tuxfamily.org/dox/TopicMultiThreading.html
Eigen uses number of threads specified my OpenMP, unless Eigen::setNbThreads(n); was called.
Exectuion on PC with 4 cores and 8 hyperthreads:
Execution on Intel CPU E5-2620:
