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

From: Åke Sandgren <ake.sandgren.hpc2n.umu.se>
Date: Tue, 30 Jun 2015 08:41:35 +0200

Hi!

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.

-- 
Ake Sandgren, HPC2N, Umea University, S-90187 Umea, Sweden
Internet: ake.hpc2n.umu.se   Phone: +46 90 7866134 Fax: +46 90-580 14
Mobile: +46 70 7716134 WWW: http://www.hpc2n.umu.se



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber


Received on Tue Jun 30 2015 - 00:00:02 PDT
Custom Search