GAUSS 11: Time Trials

"I love the speed of GAUSS 11!"

Performance data listed below was gathered on a Linux server running SUSE 11.0 with a quadcore 2.66 Ghz Intel Xeon CPU, model number E5430 and 8 GB of RAM.

GAUSS 11 will automatically run with as many threads as can be used efficiently based upon system resources and the code to be run. This is by default and requires no user input. Computation times for GAUSS 11 with 1, 2 and 4 threads were obtained by setting a maximum number of threads that could be created.

All times are in seconds.

*random data from function rndn.
Function Input* GAUSS 10 GAUSS 11
(Number of threads)  
1
1
2
4
LU factorization
x = 2000x2000
18.17
0.79
0.59
0.43
Cholesky factorization
x = 2500x2500
7.58
0.69
0.39
0.20
Matrix Multiply: x*y
x = 2000x2000
y = 2000x2000
9.54
1.64
0.85
0.46
Linear Solve: y/x
x = 1500x1500
y = 1500x1500
65.1
1.06
0.62
0.37
Matrix Inversion: inv(x)
x = 1500x1500
3.26
0.9
0.53
0.38
Eigenvalues and Eigenvectors: eigv(x)
x = 800 x 800
10.71
2.04
1.93
1.91

Performance data listed below was gathered on a machine running Windows 7 64-bit with a quadcore 3.3 Ghz Intel i5 CPU, model number 2600K and 8 GB of RAM.

All times are in seconds.

*random data from function rndn.
Function Input* GAUSS 10 GAUSS 11
(Number of threads)  
1
4
LU factorization
x = 2000x2000
16.1
0.24
Cholesky factorization
x = 2500x2500
1.96
0.14
Matrix Multiply: x*y
x = 2000x2000
y = 2000x2000
6.43
0.33
Linear Solve: y/x
x = 1500x1500
y = 1500x1500
33.4
0.31
Matrix Inversion: inv(x)
x = 1500x1500
2.40
0.27
Singular Values: lapSvdcusv
x = 500x500
1.5
0.12