Submitted by Miki on Sat, 07/28/2012 - 08:26
Hi, when I do two same analyses (e.g. ALFF) with two separate Matlab at the same time in the same user account, I found that it is likely that one of the analysis crashes with the other. I notice that they both produce and erase tmp files with random numbers in the /tmp/ directory constantly. If both analyses produced the same file concidently, e.g. ALFF_3462_username, and one of the analysis deleted it after using it while the other analysis suddenly lose track of such file then it will crash. Is that true?
Now even if one of the analysis seems to finish ALFF first (i.e. it didn't crash), is it possible that such ALFF computation may not be 100% correct due to the interference of the other analysis?
Thanks for your answer!
Submitted by YAN Chao-Gan on Wed, 08/01/2012 - 23:28 Permalink
Re
The current version of REST has some problem in parallel computing, thus you will get error in two separate MATLAB with the same account.
However, we have figured out this issue, and you can enjoy the parallel computing feature in the next release of REST (in August soon).