# Difference between revisions of "R Benchmark 2.5"

Jump to navigation
Jump to search

Moskalenko (talk | contribs) |
(We now have a c6420 with Skylake processors. Do we want to do benchmarks on it?) |
||

(One intermediate revision by the same user not shown) | |||

Line 1: | Line 1: | ||

[[Category:Benchmarking]] | [[Category:Benchmarking]] | ||

+ | [[Category:AlexReview]] | ||

+ | |||

;Processor type: | ;Processor type: | ||

: HPG1 = AMD Opteron 6378 @ 2.4GHz | : HPG1 = AMD Opteron 6378 @ 2.4GHz |

## Latest revision as of 21:14, 5 April 2018

- Processor type
- HPG1 = AMD Opteron 6378 @ 2.4GHz
- HPG2 = Intel Haswell Intel Xeon E5-2698 v3 @ 2.30GHz (testing)

Benchmark Name | Time, sec | |||
---|---|---|---|---|

R version | 3.0.2 | 3.2.0 | ||

Processor type | HPG1 | HPG2 | HPG1 | HPG2 |

I. Matrix calculation | ||||

Creation, transp., deformation of a 2500x2500 matrix | 1.69 | 1.20 | 1.58 | 1.07 |

2400x2400 normal distributed random matrix ^1000 | 2.70 | 1.30 | 2.69 | 1.32 |

Sorting of 7,000,000 random values | 1.32 | 0.81 | 1.36 | 0.87 |

2800x2800 cross-product matrix (b = a' * a) | 25.20 | 13.18 | 2.73 | 0.79 |

Linear regr. over a 3000x3000 matrix (c = a \ b') | 12.57 | 5.92 | 1.37 | 0.41 |

Trimmed geom. mean (2 extremes eliminated) | 3.86 | 2.10 | 1.80 | 0.90 |

II. Matrix functions | ||||

FFT over 2,400,000 random values | 0.62 | 0.32 | 0.64 | 0.29 |

Eigenvalues of a 640x640 random matrix | 1.75 | 0.93 | 0.81 | 0.35 |

Determinant of a 2500x2500 random matrix | 9.04 | 4.29 | 1.19 | 0.39 |

Cholesky decomposition of a 3000x3000 matrix | 9.27 | 5.07 | 0.99 | 0.40 |

Inverse of a 1600x1600 random matrix | 9.14 | 3.37 | 1.18 | 0.31 |

Trimmed geom. mean (2 extremes eliminated) | 5.25 | 2.38 | 0.98 | 0.35 |

III. Programmation | ||||

3,500,000 Fibonacci numbers calculation (vector calc)(sec) | 2.97 | 1.47 | 2.70 | 1.35 |

Creation of a 3000x3000 Hilbert matrix (matrix calc) (sec) | 0.62 | 0.43 | 0.44 | 0.32 |

Grand common divisors of 400,000 pairs (recursion) | 1.72 | 1.05 | 0.86 | 0.49 |

Creation of a 500x500 Toeplitz matrix (loops) | 1.21 | 0.52 | 0.74 | 0.34 |

Escoufier's method on a 45x45 matrix (mixed) | 0.88 | 0.62 | 0.62 | 0.40 |

Trimmed geom. mean (2 extremes eliminated) | 1.22 | 0.70 | 0.74 | 0.41 |

Total time for all 15 tests | 80.71 | 40.48 | 19.90 | 9.10 |

Overall mean (sum of I, II and III trimmed means/3)_ (sec) | 2.91 | 1.51 | 1.09 | 0.50 |