[Dockdev] Re: Re: opteron error

[DOCK]

Dear Tan,

   I apologize that you are having this problem.  We are aware of the fact 
that the random number generator is system-dependent and have actually 
rewritten this portion of the code completely along with the minimizer code 
that calls it.  It will be included in the next release (DOCK 5.2) which is 
due out by the end of the month.  Unfortunately, we do not have a quick 
alternative fix to this issue, but you should be notified the day we do the 
release.

Sincerely,
Terry


At 04:46 PM 1/20/2005, Tan Guangming wrote:
>DOCK
>Hello£¡
>
>         I find that there are two functions for generating a random 
> number. I use the other random number generating function to get the 
> results on Opteron and Xeon because it works well, but give different results.
> >float Random_Number_Generator::get_random_number() {
>
> > float num;
> >  if (generator_choice == SYSTEM_RANDOM_NUMBER_GENERATOR)
> >    num = (float) rand() / (float) RAND_MAX;
> >  else
> >    num = independent_rand();
> >  return num;
> >}
>         I didn't change the parameter file "mtx.in".
>         Thanks!
>
>======= 2005-01-21 02:44:45 ÄúÔÚÀ´ÐÅÖÐдµÀ£º=======
>
> >Dear Tan,
> >
> >   For the segmentation fault in the demo, the first thing to check is that
> >you have that you have a functioning version of python.  If you do not have
> >this working, then dock will seg fault.
> >   For the second question, a)  If you can't get the code to work, where
> >are the numbers from? and b) did you change something in the parameters
> >because only 231 orientations passed for the Xenon whereas 233 passed for
> >the Opteron?  If you could answer these questions, I may be able to help
> >you more.
> >
> >Sincerely,
> >Terry
> >
> >
> >At 05:51 AM 1/20/2005, you wrote:
> >>Hello!
> >>I compiled dock 5.1.1 on Opteron processor Linux platform, it shows the
> >>SEGMENTATION FAULT error message when I run the demo. I traced the program
> >>and found that one instruction j=iy/NDIV in the function
> >>Random_Number_Generator::independent_rand() gived the error,i.e. iy =
> >>19371779413, NDIV=67108862, however, j = 8156!
> >>Besides, the final results on AMD Opteron is different from that on Intel
> >>Xeon!
> >>Xeon:
> >>  .  ...
> >>         Orientations Passed : 231
> >>  .....
> >>         Energy score: -40.376114
> >>                         vdw:  -19.698305
> >>                         es:       -20.677811
> >>Opteron:
> >>.  ...
> >>         Orientations Passed : 233
> >>  .....
> >>         Energy score: -50.252425
> >>                         vdw:  -23.808451
> >>                         es:       -21.243562
> >>Why?
> >>
> >>
> >>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
> >>
> >>
> >>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Tan Guangming
> >>                            National Research Center for Intelligent
> >> Computing Systems,
> >>                            Institute of Computing Technology,
> >>                            Chinese Academy of Sciences
> >>
> >>                            6#, South Road, Kexueyuan, Zhongguancun,
> >>                        Beijing, P.R.China
> >>                        P.O.Box 2704   Zip Code 100080
> >>
> >>                Phone: 86-10-62565533 ext. 5730
> >>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡tgm at ncic.ac.cn
> >>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2005-01-20
>
>= = = = = = = = = = = = = = = = = = = =
>
>
>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡ÖÂ
>Àñ£¡
>
>
>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡Tan Guangming
>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡tgm at ncic.ac.cn
>¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡2005-01-21