MATLAB: PCT and GA optimizer

genetic algorithmParallel Computing Toolbox

Hi, I try to use parallel computing toolbox to decrease genetic algorithm optimizer run time. all of workers will call an external solver using one premade input file.(call MSC Nastran bdf is made in MATLAB). the thing is each time external solver is called some files are created.to avoid conflict between workers three options came to my mind. 1) use of resource lock to avoid conflict but this will bring problems since other workers solver dont wait and will fail. 2) tell PCT to create directory for each worker and copy necessary files into each directory. 3) in objective function get process id of each worker and put an extension in input file made by objective function for external solver therefore each workers will have different output and input file name.

the last option looks like the easiest one. what do you guys think ? and btw I dont know method 2 and 3. need some help 🙂 thanks in advance

Best Answer

If you use the Vectorize option, you can have GA pass the entire population to your objective function in each iteration. Within your objective function, you can then implement your own parallel splitting of the computation using Parallel Computing Toolbox functions.

For example, you can evaluate different population chunks in parallel in an SPMD...END block and use the labindex() command to create a worker-dependent name for your files,

     spmd
      ...
      filename=[filename numstr(labindex)];
     end

Related Solutions

MATLAB: Does the batch job run on a single worker if the code batched doesn’t contain any parallel for loop

Your local machine is making use of maxNumCompThreads, which is set to the number of physical cores on your machine. The 'parallelism' is threads, not processes, that are spawned by eigs onto those cores.

When you call batch, you can pass it a pool argument, specifying how many workers (processes) should start up. batch will request one additional worker to act as your proxy MATLAB client. For example,

j = batch(...,'Pool',3);

will reqest 4 workers run on your cluster. As you've noticed, you don't need additional workers for your code. You just need the single worker that batch will request, which in turn needs to spawn threads (across several cores).

Your question then is, why does eigs run with multiple threads locally, but only a single thread on the cluster? By default, MATLAB will only request a single thread per worker. To increase this, set the number of threads per worker, as such (8 is only an example)

c = parcluster(name-of-cluster-profile);
c.NumThreads = 8;
j = c.batch('myfile');

If, for some reason, that doesn't work, you can call maxNumCompThreads in myfile. The caveat is that setting NumThreads should request additional cores (needed for the additional threads). If setting NumThreads didn't work, you'll only get a single core. Then, calling maxNumCompThreads may have one of the following consequences

Starting threads on cores you don't own
Starting all on the same core, if cgroups are implemented

Neither is desireable.

MATLAB: Parallel pool in batch processing

Until R2016b, all workers received a single thread. The relevant R2016b release note is:

"Multithreaded Workers: Use multiple computational threads on your MATLAB workers

MATLAB workers used to run in single-threaded mode. Now you can control the number of computational threads so that workers can run in multithreaded mode and use all the cores on your cluster. This enables you to increase the number of computational threads, NumThreads, on each worker, without increasing the number of workers, NumWorkers. If you have more cores available, increase NumThreads to take full advantage of the built-in parallelism provided by the multithreaded nature of many of the underlying MATLAB libraries. For more information, see Create and Modify Cluster Profiles."

The documentation for batch shows examples of creating a batch without opening a parpool, and an example of requesting a pool. No explicit "parpool" statement is used.

If you are running on a cluster, you might need to parcluster() and batch() against that cluster.

Best Answer

Related Solutions

MATLAB: Does the batch job run on a single worker if the code batched doesn’t contain any parallel for loop

MATLAB: Parallel pool in batch processing

Related Question