Don't worry worry about it...
Changes in compiler and GPU architecture change orders of operation and
since NVIDIA updates its compiler every couple months it's nearly
impossible to keep up.
Worry when the same run produces different results on the same
configuration when run several times...
On Fri, Sep 7, 2012 at 10:13 AM, Jonathan Gough
<jonathan.d.gough.gmail.com>wrote:
> Hey All,
>
> Just got my new system up and running:
>
> Ubuntu 12.04
> gnu fortran etc.
> mpich2
> Nvidia Driver 295.49
> Nvidia GeForce 690 (has 2 cores)
>
>
> Everything compiles fine (serial, parallel, cuda, cuda parallel)
>
> But with make test on the cuda and cuda.MPI install I get errors that are
> there, but small. Need I be concerned? If so, what suggestions are there
> out there?
>
> I know the differences are small, but I figured I better check.
>
> logs/test_amber_cuda/2012-09-05_17-33-35.diff
>
> possible FAILURE: check nucleosome_md2_ntt0.out.dif
> /home/jonathan/amber12/test/cuda/nucleosome
> 261c261
> < Etot = 3.4246 EKtot = 62.6287 EPtot =
> 65.3240
> > Etot = 3.4252 EKtot = 62.6282 EPtot =
> 65.3240
> ### Maximum absolute error in matching lines = 6.00e-04 at line 261 field 3
> ### Maximum relative error in matching lines = 1.75e-04 at line 261 field 3
> ---------------------------------------
> possible FAILURE: check jac.out.dif
> /home/jonathan/amber12/test/cuda/jac
> 112c112
> < Etot = 0.2318 EKtot = 47.0779 EPtot =
> 47.0435
> > Etot = 0.2324 EKtot = 47.0785 EPtot =
> 47.0434
> ### Maximum absolute error in matching lines = 6.00e-04 at line 112 field 3
> ### Maximum relative error in matching lines = 2.59e-03 at line 112 field 3
> ---------------------------------------
> possible FAILURE: check mdout.dhfr.dif
> /home/jonathan/amber12/test/cuda/dhfr
> 206c206
> < Etot = 0.1629 EKtot = 54.6594 EPtot =
> 54.6895
> > Etot = 0.1628 EKtot = 54.6593 EPtot =
> 54.6895
> ### Maximum absolute error in matching lines = 1.00e-04 at line 206 field 3
> ### Maximum relative error in matching lines = 6.14e-04 at line 206 field 3
> ---------------------------------------
> possible FAILURE: check mdout.dhfr.ntr1.dif
> /home/jonathan/amber12/test/cuda/dhfr
> 267c267
> < Etot = 0.2226 EKtot = 63.7687 EPtot =
> 63.8464
> > Etot = 0.2224 EKtot = 63.7686 EPtot =
> 63.8464
> ### Maximum absolute error in matching lines = 2.00e-04 at line 267 field 3
> ### Maximum relative error in matching lines = 8.99e-04 at line 267 field 3
> ---------------------------------------
> possible FAILURE: check mdout.ips.amd1.dif
> /home/jonathan/amber12/test/cuda/amd/gact_ips
> 252c252
> < Etot = 0.2870 EKtot = 113.9213 EPtot =
> 113.9790
> > Etot = 0.2868 EKtot = 113.9212 EPtot =
> 113.9790
> 256c256
> < EKCMT = 31.7648 VIRIAL = 314.2456 VOLUME =
> 0.2383
> > EKCMT = 31.7648 VIRIAL = 314.2456 VOLUME =
> 0.2384
> ### Maximum absolute error in matching lines = 2.00e-04 at line 252 field 3
> ### Maximum relative error in matching lines = 6.97e-04 at line 252 field 3
>
>
> logs/test_amber_cuda_parallel/2012-09-07_12-50-05.diff
>
> possible FAILURE: check nucleosome_md2_ntt0.out.dif
> /home/jonathan/amber12/test/cuda/nucleosome
> 261c261
> < Etot = 3.4246 EKtot = 62.6287 EPtot =
> 65.3240
> > Etot = 3.4252 EKtot = 62.6282 EPtot =
> 65.3240
> ### Maximum absolute error in matching lines = 6.00e-04 at line 261 field 3
> ### Maximum relative error in matching lines = 1.75e-04 at line 261 field 3
> ---------------------------------------
> possible FAILURE: check jac.out.dif
> /home/jonathan/amber12/test/cuda/jac
> 112c112
> < Etot = 0.2318 EKtot = 47.0779 EPtot =
> 47.0435
> > Etot = 0.2328 EKtot = 47.0780 EPtot =
> 47.0434
> ### Maximum absolute error in matching lines = 1.00e-03 at line 112 field 3
> ### Maximum relative error in matching lines = 4.31e-03 at line 112 field 3
> ---------------------------------------
> possible FAILURE: check mdout.dhfr.ntr1.dif
> /home/jonathan/amber12/test/cuda/dhfr
> 267c267
> < Etot = 0.2226 EKtot = 63.7687 EPtot =
> 63.8464
> > Etot = 0.2224 EKtot = 63.7689 EPtot =
> 63.8464
> ### Maximum absolute error in matching lines = 2.00e-04 at line 267 field 3
> ### Maximum relative error in matching lines = 8.99e-04 at line 267 field 3
> ---------------------------------------
> possible FAILURE: check mdout.ips.amd1.dif
> /home/jonathan/amber12/test/cuda/amd/gact_ips
> 252c252
> < Etot = 0.2870 EKtot = 113.9213 EPtot =
> 113.9790
> > Etot = 0.2868 EKtot = 113.9212 EPtot =
> 113.9790
> ### Maximum absolute error in matching lines = 2.00e-04 at line 252 field 3
> ### Maximum relative error in matching lines = 6.97e-04 at line 252 field 3
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Sep 07 2012 - 10:30:04 PDT
