Dear lucky K20/K20x user. Several Amber users recently encountered
some issues with Titan GPUs during Amber calculations, please see the
actual discussion thread:
"experiences with EVGA GTX TITAN Superclocked - memtestG80 - UNDERclocking
in Linux ?"
Here some problems were reported also with finishing/reproducibility of
Amber benchmark
tests (the whole suite is available for download here:
http://ambermd.org/gpus/benchmarks.htm ).
We would kindly appreciate if you could do at least 100K steps long above
mentioned
benchmark tests (those PME are enough) twice (using the same random seed )
and report
here your results i.e. if all the tests were successfully finished and if
they were 100% reproducible
(e.g. obtaining exactly the same final energies).
Thanks a lot in advance !
M.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jun 01 2013 - 14:00:02 PDT
