Parfor code runs slower on 16 core, 2.3 GHz AMD processor than 4 core, 3.3 GHz Intel processor
2 views (last 30 days)
Show older comments
Hello everyone,
For some reason, the following code runs about twice as fast on a desktop with a 4 core, 3.3 GHz Intel processor than on a node on a computing cluster with a 16 core, 2.3 GHz AMD processor (don't know the full specs for either). Either way, a parfor loop is faster on both computers than a regular for loop, but I can't see how 16 cores could be slower than 4 cores. I am compiling the code using mcc beforehand in each case, and the correct number of workers are used.
Thank you in advance for your help!
parfor ii = 1:indexer
[output1,output2] = simulate_ssa((initial_condition(ii,:)), timestep, final_time_step,ii);
%Writes a file out to the current working directly (using the index ii to avoid rewriting over a previous file) and runs a unix command on that file
if output1 == 2
finaloutputA = vertcat(finaloutputA,[output2,initial_condition(ii,end)]);
else
finaloutputB = vertcat(finaloutputB,[output2,initial_condition(ii,end)]);
end
end
0 Comments
Answers (0)
See Also
Categories
Find more on Cluster Configuration in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!