Hello there,

I've spent more time than I'm willing to bluntly admit trying to get eddy_openmp to work properly, and am wondering if anybody else has dealt with (and hopefully overcome) the same issue?

I have multiband, multi-shell DWI data with reverse phase-encoded b0s and b100s, acquired from a Siemens scanner, and being analyzed on an Ubuntu machine.  I used the reverse phase encode data to run topup without issue.  I can run eddy_openmp beautifully on a single shell with 64 A-P volumes for both of my higher b-values, as well as A-P multishell data with 3 shells and a total of 137 volumes.  For some reason, if I add my P-A b0 and 6 volumes of b100 P-A data, I get this error every time:
When I ran "--very_verbose", I also got this error:
On another iteration, the numbers listed above were different, with multiple numbers that had multi-spaces between them, and numbers ranging from 0 - 135132143142, and a different number of total fields/characters.

Here are the things I've altered, checked, and/or swapped:
 - putting my PA vols at the beginning, putting them at the end
 - putting my PA volumes at the end but without the b0
 - ensuring that my bval and bvec files have exactly one space between entries, and that they are always correct
 - condensing my acqparams file to 2 lines (I had multiple because I fed my low bvals into topup with my b0) so the only options in my index file would be "1" and "2"
 - just the PA vols by themselves (get a different error, suspect there are too few volumes though)
 - with and without the "--data_is_shelled" arg
 - with and without the "--dont_peas" arg // + combinations of these options and the options on the previous line
 - using an older version of eddy that was called "eddy_patched" (not sure if I named it or if FSL did) but seemed to multi-thread just like eddy_openmp
 - raw files vs. rescaled files
 - different subjects
 - I've played around with niter and fwhm too because I needed to speed things up
 - I've tried --dont_sep_offs_move

Here are args I've held constant:
 - mb=3
 - repol
 - ol_type=both

Finally, my data runs through dtifit without issues.

Any idea what could be going on here??!  Has anybody successfully used two phase-encode directions of the same b-value with eddy_openmp and had success?  Any and all insight is greatly appreciated.