Mdrun -rerun output file sizes do not make sense

GROMACS version: 2022
GROMACS modification: No

I’m trying to calculate potential energies from an .xtc file and the original .tpr file. Since I don’t have an .edr file, I’m doing gmx mdrun -rerun. The .xtc file has ~500 snapshots that are spaced 1ns apart. With these files, when I do the rerun, it says I get the following warning:

WARNING: This run will generate roughly 3411371164616200192 Mb of data

This must be a bug, right?

If I do run it, it seems to only process every 1ns, as expected. However, the .trr file it produces is 1.1G, compared to the .xtc file that was 127M. Does this make sense? What could the .trr have that the .xtc file didn’t? Neighbor list information, etc.?

Yes, mdrun -rerun does not know at startup how many frames it will read.

The trr write full precision binary output, whereas xtc is compressed.