Deviations in the optimized lattice constants of TiO2 supercells

Dear Abinit Community.

I am attempting to investigate the formation energy and lattice structure of substitutional rutile lattices.

The computation time of the system grows quite large. So I have started with a 6 atom and 12 atom simulation. The 6 atom simulation results are in good agreement with the literature values. However the 12 atom simulation reports the expected double energy but the lattice constants are substantially different. Attempts to alter the variables optcell, ionmov, rprim, or ngkpt only cause the program to crash. I have attached the files input and output for the 6 atom and 12 atom simulations. The pseudopotentials used are paw type (JTH v1.1) xc (PBEsol) accuracy (standard) format (xml). The files attached contain the input and output of both the 6 atom and 12 simulations.
TiO2_bulk.abi (1.4 KB)
TiO2_bulk_12_atom.abi (1.8 KB)
TiO2_bulk_12_atom.abo (1.7 MB)
TiO2_bulk.abo (455.6 KB)