Is it possible to load and modify the same .mat file in two instances of matlab?
3 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Margaret
am 17 Jan. 2014
Kommentiert: Walter Roberson
am 18 Jan. 2014
I am attempting to parallelize my Matlab code without using the parallel computing toolbox, since parfor cannot be compiled using mcc, as far as I know.
Would I run into any issues if I ran the following code simultaneously on two or more Matlab instances?
function simulate(input_data,index,foo,bar)
load(foo);
load(bar);
tempdata = subfunction(input_data(index,:));
if all(tempdata(:,2) <10)
foo_data= vertcat(foo_data,[tempdata,input_data(index,end)]);
else
bar_data = vertcat(bar_data,[tempdata,input_data(index,end)]);
end
save(foo,'foo_data');
save(bar,'bar_data');
end
(where subfunction is another .m file in the current working directory)
Each instance of matlab would have a different index input.
Thanks in advance!
2 Kommentare
Akzeptierte Antwort
Bruno Pop-Stefanov
am 17 Jan. 2014
Bearbeitet: Bruno Pop-Stefanov
am 17 Jan. 2014
I don't know if your code would work but trying to access a single resource from two different processes sounds like a bad idea. I can't speak for MATLAB, but in general, for safety reasons, when a process opens a file (a resource), the file is locked until the process releases it. Assuming that you could access your data in two instances of MATLAB, you don't know how and in which order each unit of data is accessed at a lower, byte level. You could end up with deleted segments of your data, or with duplicated data. If you really want to run two instances of MATLAB, I would recommend to split your data in two and give one side each.
4 Kommentare
Walter Roberson
am 18 Jan. 2014
I do seem to recall that the parts of a .mat file can be concatenated, but possibly not the header as well... that part I do not recall. Also if a variable occurs in multiple places, the last version of it would hold; they would not be concatenated together.
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Parallel Computing Toolbox finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!