Performance of Linear Programming Library GIPALS32

Version 3.4 (January 8, 2011)

Optimalon Software provides the performance tests for GIPALS32 using well-known LP test problems from the NETLIB library. The tests can be freely downloaded in MPS format from NETLIB website.

All tests have been performed on Intel Core2 CPU 2.4 GHz, 2 GB of RAM under Microsoft Windows XP Professional operating system. The results are summarized in the following table along with LP problem name, size and the objective function value.

Small and medium size Linear Problems:

LP Problem Variables Count Constraints Count Obj. Function Value Solution Time, sec
25fv47 1,571 821 5501.84 0.438
80bau3b 9,799 2,262 987224.24 1.141
bnl2 3,489 2,324 1811.23 1.453
cycle 2,857 1,903 -5.22639 0.672
d2q06c 5,167 2,171 122784.21 2.172
d6cube 6,184 415 315.49 0.578
dfl001 12,230 6,071 11266397 83.453
maros-r7 9,408 3,136 1497185.16 6.391
pilot 3652 1441 -557.489 4.219
pilot87 4,883 2,030 301.71 9.844
stocfor3 15,695 16,675 -39976.78 5.157
wood1p 2,594 244 1.44290 0.375
woodw 8,405 1,098 1.304476 0.578

Large size Linear Problems:

LP Problem Variables Count Constraints Count Obj. Function Value Solution Time, sec
pds-10 48,763 16,558 26727119679 52.593
pds-20 105,728 33,874 23821705237 477.907
pds-30 154,998 49,944 21385565848 2164.562
watson_1 383,927 201,155 -1693.47 119.484
watson_2 671,861 352,013 -7544.81 202.797
rail2586 920,683 2,586 935.97 694.265