12.07.2015 Views

ILOG CPLEX 11.0 User's Manual

ILOG CPLEX 11.0 User's Manual

ILOG CPLEX 11.0 User's Manual

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

0 0 3383.7784 16 4505.0000 Cuts: 35 114 24.89%0 0 3489.7887 13 4505.0000 Cuts: 17 134 22.54%0 0 3531.4512 16 4505.0000 Cuts: 11 149 21.61%* 0+ 0 4501.0000 3531.4512 149 21.54%0 0 3552.4474 19 4501.0000 Cuts: 9 159 21.07%0 0 3588.7513 19 4501.0000 Cuts: 8 178 20.27%0 0 3597.3630 20 4501.0000 Cuts: 5 185 20.08%0 0 3598.6200 21 4501.0000 Cuts: 6 190 20.05%0 0 3601.2921 22 4501.0000 Cuts: 3 194 19.99%* 0+ 0 4457.0000 3601.2921 194 19.20%0 0 3602.5189 23 4457.0000 Cuts: 3 197 19.17%0 0 3633.5846 20 4457.0000 Cuts: 2 204 18.47%0 0 3638.8299 22 4457.0000 Covers: 1 206 18.36%0 0 3649.9331 22 4457.0000 Cuts: 3 212 18.11%0 0 3652.9758 28 4457.0000 Cuts: 5 218 18.04%0 0 3656.7563 24 4457.0000 Cuts: 3 227 17.95%0 0 3659.1865 29 4457.0000 Cuts: 4 232 17.90%0 0 3664.7373 26 4457.0000 Covers: 1 236 17.78%0 0 3676.2923 33 4457.0000 Covers: 5 244 17.52%0 0 3676.7448 31 4457.0000 Flowcuts: 2 250 17.51%0 0 3676.9154 33 4457.0000 Flowcuts: 2 254 17.50%* 0+ 0 3994.0000 3676.9154 254 7.94%* 0+ 0 3990.0000 3676.9154 254 7.85%* 0+ 0 3985.0000 3676.9154 254 7.73%0 2 3680.6918 22 3985.0000 3676.9154 254 7.73%* 22 5 integral 0 3983.0000 3770.3216 358 5.34%Root node processing (before b&c):Real time = 1.39Parallel b&c, 2 threads:Real time = 0.24Sync time (average) = 0.08Wait time (average) = 0.03-------Total (root+branch&cut) = 1.63 sec.Cover cuts applied: 13Flow cuts applied: 40Gomory fractional cuts applied: 6Solution pool: 8 solutions savedMIP - Integer optimal solution: Objective = 3.9830000000e+03Solution time = 1.63 sec. Iterations = 426 Nodes = 40The sample reports that the processors spent an average of 0.03 of a second of real timewaiting for other processors to provide nodes to be processed. The Sync time is the averagetime a processor spends trying to synchronize by accessing globally shared data. Becauseonly one processor at a time can access such data, the amount of time spent insynchronization is really crucial: any other processor that tries to access this region mustwait, thus sitting idle, and this idle time is counted separately from the Wait time.There is another difference in the way logging occurs in the parallel MIP optimizer. Whenthis optimizer is called, it makes a number of copies of the problem. These copies are knownas clones. The parallel MIP optimizer creates as many clones as there are threads available toit. When the optimizer exits, these clones and the memory they used are discarded.<strong>ILOG</strong> <strong>CPLEX</strong> <strong>11.0</strong> — USER’ S MANUAL 499

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!