The plot thickens....
Following David's suggestion I took out all the baseline regressors. Everything ran fine. As a test I started putting the regressors back in.
It seems that in all cases the final regressor, dL (*.dfile[6]), is what is causing the problems. Take out that regressor all of them run fine.
In answer to Rick's questions, I am running /bin/bash as the shell and I do just download the gcc32 binaries, I do not compile my own. Both systems with the failure are running current Slackware distributions.
I have looked at that timeseries (*.dfile[6]) in the ones that fail as well as the ones that work. Other than the actual values, of course, they look very similar. They have the same trends, same approximate values as the ones that run.
I'm really stumped why they would seg fault. Especially since it doesn't happen for you there. Do you suppsose it's something with Slackware? Though that would be odd that most run, but not these two subjects.
iggy