Re: [Wien] Problem with dstart in new version 17.1

2017-08-14 Thread Fecher, Gerhard
et: Montag, 14. August 2017 15:46 An: A Mailing list for WIEN2k users Betreff: Re: [Wien] Problem with dstart in new version 17.1 You probably have to drop/change the optimization level and/or use -ftz (perhaps -noftz). I have seen related issues in mixer when small numbers are added leading to N

Re: [Wien] Problem with dstart in new version 17.1

2017-08-14 Thread Laurence Marks
You probably have to drop/change the optimization level and/or use -ftz (perhaps -noftz). I have seen related issues in mixer when small numbers are added leading to NaN's being output. I blame this on Intel, although I may be wrong. On Mon, Aug 14, 2017 at 7:49 AM, Fecher, Gerhard

[Wien] Problem with dstart in new version 17.1

2017-08-14 Thread Fecher, Gerhard
I experienced a problem with dstart of version 17.1, an error appears as given below when running a calculation for the attached structure (initialisation with defaults). A closer look, however, shows that this error does not result from lapw2 but from dstart. (I found NANs in some files used