Simulink rapid acceleration creates temp files which fill drive

조회 수: 7 (최근 30일)
schapplm
schapplm 2013년 12월 3일
답변: schapplm 2013년 12월 5일
I set up a parfor-loop which simulates a Simulink-model, like in this tutorial: http://www.mathworks.de/de/help/simulink/examples/rapid-accelerator-simulations-using-parfor.html
It works great and fast, but each Sim Command executes a exe-File which writes the Simulation output (tout, yout) to a temp file like "oup5ec41494_5f3b_4a75_8e18_0978da1bc809.mat" in the "slprj\raccel\modelname"-folder.
The file is being deleted after evaluation, but only to windows trash, so after thousands of iterations (I use the parfor loop in a genetic algorithm), my hard disk drive is full (after creating 160GB temp files). Is there a possibility, to make matlab delete the temp files without trash?
I am using Matlab 2012b with Windows 7 x64.

채택된 답변

schapplm
schapplm 2013년 12월 5일
In Matlab 2013a, the problem seems to be solved.

추가 답변 (1개)

Ryan G
Ryan G 2013년 12월 4일
What's your parfor loop look like? You might be able to do something like
parfor i = 1:100
eval(['myOutput' num2str(i) ' = sim(gcs)'])
%Not sure exact syntax, try not to use eval if possible
% Process algorithm here
delete(['myOutput' num2str(i)])
end
  댓글 수: 1
schapplm
schapplm 2013년 12월 5일
using the eval function does not help, because the temp file is generated outside matlab in the model-.exe-file that is generated by the rapid accelerator. So i cannot delete the temp-file in Matlab either, since it is deleted to recycle bin by the model.exe.
% Build the Rapid Accelerator target (create .exe-file)
rtp = Simulink.BlockDiagram.buildRapidAcceleratorTarget(sl_Modellname);
parfor i=1:100
parameterSet = Simulink.BlockDiagram.modifyTunableParameters(rtp, ...
'EKF_Q', EKF_Q, 'EKF_P0', EKF_P0); % modify some parameters
simOut= sim(sl_Modellname, ...
'SimulationMode', 'rapid', ...
'RapidAcceleratorUpToDateCheck', 'off', ...
'RapidAcceleratorParameterSets', parameterSet);
% calculations with simOut
% ...
end

댓글을 달려면 로그인하십시오.

카테고리

Help CenterFile Exchange에서 Multicore Processor Targets에 대해 자세히 알아보기

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by