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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

optimizer turns out to be the fastest, the concurrent optimizer always returns a basic solutionat optimality.The concurrent optimizer requires more memory than any individual optimizer, and ofcourse it adds system load by consuming more aggregate CPU time than the fastestindividual optimizer would alone. But the advantages offered in terms of robust solution ofmodels, and assurance in most cases of the minimum solution time, will make it attractive inmany situations.Parallel MIP OptimizerBy default, <strong>ILOG</strong> <strong>CPLEX</strong> uses the deterministic parallel MIP optimizer to solve a mixedinteger programming problem. In doing so, it exploits parallel computations while it solvesnodes of the MIP branch & cut tree. It also executes strong branching computations inparallel.Besides these applications of parallel processes, there may be additional possibilities forparallel computation at the root node. However, these additional possibilities are notdeterministic; they are known as opportunistic. In order to exploit such opportunities, yourapplication must allow <strong>ILOG</strong> <strong>CPLEX</strong> to run opportunistically in parallel. To do so, eitheryou can explicitly set the parallel mode parameter (ParallelMode,CPX_PARAM_PARALLELMODE) to the value -1 (minus one), or you can implicitly allowopportunistic parallel optimization by setting the threads parameter (Threads,CPX_PARAM_THREADS) to a value strictly greater than one.As explained in Determinism of Results on page 493, you may find it advantageous todevelop your application in deterministic parallel mode, where you can rely on theinvariance and repeatability of the search path and results to evaluate the correctness of yourmodel and solutions. After you are convinced of correctness of your model, there are twodifferent approaches you can take in deployment of your application. If performance iscritical, consider deploying in opportunistic parallel mode. While faster performance inopportunistic mode cannot be guaranteed, it does generally out-perform deterministic mode.On the other hand, if performance is not critical, you may prefer to deploy in deterministicmode to retain the possibility of reproducing any problems that your end-user may encounterduring deployment. In summary, you need to evaluate for your model and application whichmode is more appropriate.The following sections offer further insight into managing MIP optimization in parallel.◆ Root Relaxation and Parallel MIP Processing on page 498◆ Memory Considerations and the Parallel MIP Optimizer on page 498◆ Output from the Parallel MIP Optimizer on page 498<strong>ILOG</strong> <strong>CPLEX</strong> <strong>11.0</strong> — USER’ S MANUAL 497

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

Saved successfully!

Ooh no, something went wrong!