Well, you posted the question on a national holiday, so answers tend to be slow.
The message means that the matrix computed at the extreme autocorrelation parameter (rho=0.8) was not positive definite. This can arise if you have a peculiar set of regressors, due to tiny numerical errors. You can try using the option '-CORcut 0.0001' to see if this fixes the problem (pushes the numerical errors out farther). If this does not solve the issue, then I'll need access to your script and .xmat.1D file in order to see what's happening.