Continue to next iteration in for-loop while saving data
Mostrar comentarios más antiguos
I have a for-loop and in each iteration it produces about 1Gb data. Since it takes some time to save this data to hard drive, I want to use a different worker to save the data and let the for-loop continue to the next iteration at the same time on the original worker. Is there any way to do this? Thank you.
Edit: The calculations in each iteration before saving data takes longer time than saving to disk, so the current worker won't reach the save command in the next iteration before the other worker finishes save for the current iteration.
3 comentarios
Image Analyst
el 20 de Abr. de 2021
Wouldn't you still be limited by your disk I/O controller rather than your CPU? I mean you can only write data through the cable to the disk up to a certain speed. It just won't go any faster. Have you checked your disk access with Windows Task Manager? Is your disk at 100% during a GB write?
Weixin Wang
el 20 de Abr. de 2021
messaoudi nada
el 24 de Sept. de 2021
dear @Image Analyst , i need your help please! , im working about machine learning models , so after extracting features from images i have a 12 structure data ,i want to save the data in a reference matrix and test matrix ,i tried this loop but it didn't work it just save the features of the last image in structure ! i don't know what's wrong heare ! hope u can give help to me !
thanks in advance
for i=1:12
for iref=1:10
for jref=1:40
F1=data1_rf.(['p',num2str(iref)]).(['image',num2str(jref)]).Features;
for itst=1:2
for jtst=1:40
TSF=data1_T.(['p',num2str(itst)]).(['image',num2str(jtst)]).Features;
end
end
end
end
end
Respuestas (1)
Thomas Falch
el 1 de Oct. de 2021
0 votos
You can use parfeval to run save in the background on a worker like this:
pool = parpool('local', 1) % Start pool with one worker
for i = 1:100
data = generateData()
parfeval(@() save(data), 0); % Sends data to worker to be saved in the background
end
Categorías
Más información sobre Parallel for-Loops (parfor) en Centro de ayuda y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!