2

Has anyone come across the "Bad version or endian-key" error in Matlab? A quick Google search hasn't yielded any results nor a visit to the Mathworks site.

I'm using R2013b and came across the error while running a compiled version of the program i.e. MCR version 8.2. The error came about while running a particularly large number of Monte Carlo simulations with large cell arrays used to store results. The program also uses the parallel computation toolbox and a Java based progress bar suited to parfor loops.

user2848074
  • 55
  • 6

1 Answers1

2

My recollection is that that particular error can occur if you try to load corrupt data. In parallel computing, this can also occur if the transfer from workers back to the client is corrupt or truncated. (That message has message ID MATLAB:Deserialize which indicates it's to do with loading data).

Edric
  • 23,676
  • 2
  • 38
  • 40
  • Think it's worth noting that the test environment is a virtual setup that was originally allocated 4 cores and was recently increased to 8. The error occurred when trying to use all 8 cores. So there's definitely potential for erroneous transfer from the workers to the client. – user2848074 Feb 12 '14 at 20:21
  • Exactly, I encounter this error in parfor. `Error using parallel.internal.pool.deserialize (line 9) Bad version or endian-key Error in distcomp.remoteparfor/getCompleteIntervals (line 141) origErr = parallel.internal.pool.deserialize(intervalError);` Do you know how to solve it? – Sibbs Gambling Jul 06 '14 at 03:58
  • 1
    http://stackoverflow.com/questions/24592602/bad-version-or-endian-key-in-matlab-parfor Thank you so so much sir! – Sibbs Gambling Jul 06 '14 at 05:15