Hello, Despite the claim, dating back to the NEi days, that the program had no problem size limitations, I pointed out in posts made here in 2015 that the scratch disk was limited to 16GB and that this effectively limited the size of models that could be run. There are some workarounds to get larger models to run within this limitation, such as turning off hexinode, using specific solvers etc. but the workarounds all have some disadvantages. As of mid 2017, the 16GB scratch file size limit remains. We live in a world where disks that have terabytes of space are available for a $200. While very large models are impractical for some solutions such as those that are time- or load- stepped, today's machines make solving huge models practical for other solution types. I would much rather run one 30M DOF static stress model **overnight** than spend **my working hours** doing a multi-pass superelement analysis on the same problem because of the 16GB scratch file limit. Yes that's a brute force approach. It's also the approach that makes money! It is frustrating to come to a machine in the morning and learn that the job failed because the scratch file exceeded 16GB. Windows doesn't have a 16GB limit on file size. This is either in the program code, or a limit of the compiler you're using. I understand that this would mean a re-compile and re-validation and is not as simple as it sounds. But it's needed - and the problem was identified years ago at this point. Help! We need this sort of thing to enable us to compete... which means getting the jobs out with less of our time needed! Thank you for considering my request. Regards, Andrew
Show More
