Skip to main content

Table 2 OpenMP Multithreading Speedup

From: Multilevel Parallelization of AutoDock 4.2

Threads

Speedup

% Ideal

serial

1.0×

100

1

0.9×

88

2

1.5×

75

4

3.0×

74

8

5.8×

72

16

11.3×

71

32

22.3×

70

  1. The speedup using OpenMP threads was calculated for 1 to 32 threads on a 3.3 GHz POWER7 p755 32-core server. A reference speed was calculated using mpAD4 compiled without OpenMP (serial). The numbers are scaled to a single serial docking run with the default benchmark parameters with modifications: ga_runs = 128, ga_num_evals = 5,000,000.
Back to article page

Journal of Cheminformatics

ISSN: 1758-2946

Contact us