Re: 2.12, cufft problems on a GTX 1070 ? [SOLVED]

From: Nicholas M. Glykos (glykos_at_mbg.duth.gr)
Date: Wed Jan 11 2017 - 12:51:54 CST

Dear All,

Please ignore my previous post concerning putative cufft() problems
on a GTX 1070, they were not reproducible (and definitely not NAMD-related).

Apologies for the email traffic,
Kind regards,
Nicholas

>> On Sun, 1 Jan 2017, Nicholas M Glykos wrote:
>>>
>>> Dear All,
>>>
>>> Happy new year. Any thoughts on how to confirm putative problems with
>>> cufft on namd 2.12 ? I have been doing some tests with a small peptide
>>> system using adaptive tempering and I was getting unrealistic results (the
>>> temperature 'stuck' at the high end of the adaptive tempering range).
>>> The problem appears to go away after setting 'usePMECUDA no' but to be on
>>> the safe side I would like to independently confirm that there is indeed a
>>> problem with cufft (and not something else that have escaped my notice).
>>>
>>> All the best,
>>> Nicholas
>>>
>>> ps. The machine is a stand-alone node with an i7, a GTX 1070 and the
>>> 375.26 driver. The NAMD version is the official 2.12 multicore+CUDA
>>> binary.
>>>

-- 
             Nicholas M. Glykos, Department of Molecular Biology
      and Genetics, Democritus University of Thrace, University Campus,
   Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620,
     Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/glykos/

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:01 CST