Re: [AMBER] svdcmp in sander/ncsu-rmsd.F90 (AmberTools15) is sensitive to higher optimizations, suggest using lapack routine instead

From: Åke Sandgren <>
Date: Thu, 16 Jul 2015 19:05:29 +0200

If i get time i can take a look at the other places a svdcmp routine is
being called.

On 07/16/2015 06:30 PM, David A Case wrote:
> On Tue, Jun 30, 2015, Åke Sandgren wrote:
>> svdcmp in sander/ncsu-rmsd.F90 (AmberTools15) have numerical problems at
>> higher optimization levels.
>> The ncsu/smd test fails with intel (15.0) at -O2.
>> Attached is a patch that replaces the call to svdcmp in dstmev with the
>> corresponding lapack routine (xGESVD) making the test work at full
>> optimization.
>> I've only replaced this one call at the moment since i didn't want to
>> spend too much time doing it in case it gets rejected.
> Sorry to be slow in replying: summer travels take up a lot of time.
> What you did looks OK to me (not extensively tested....); this patch will
> be in the next release of AmberTools. If you changes to other library
> calls, please let us know. I never know how much people are still using
> the ncsu stuff; PLUMED seems to a very popular alternative.
> Thanks for sending this!....regards...dave case

Ake Sandgren, HPC2N, Umea University, S-90187 Umea, Sweden
Internet:   Phone: +46 90 7866134 Fax: +46 90-580 14
Mobile: +46 70 7716134 WWW:
AMBER mailing list
Received on Thu Jul 16 2015 - 10:30:02 PDT
Custom Search