[gmx-users] g_energy floating exception

[Lynne E. Bilston]

Sergio and Erik,

I think I reliably got the error irrespective of the energy file size.

I'm using C compiler Compaq C V6.1-011. Bert was using DEC C V5.2-033 on 
Digital UNIX V4.0 (Rev. 564.32)

so maybe it's a problem with the libraries on the v6 compilers? 
Incidentally, I tried Bert's Binary, and got the same error.

This was the debug I got:

Reading frame      0 time    0.000           signal Floating point 
exception at   [do_enx:296 +0x14,0x12001def4]        tmp1 = fr->ener[i].eav;
(dbx) where
 >  0 do_enx(fp = 0, fr = 0x14003d360) ["enxio.c":296, 0x12001def4]
    1 main(argc = 1, argv = 0x11fffc018) ["g_energy.c":922, 0x12001182c]

(dbx) list 295,300
    295
 >* 296      tmp1 = fr->ener[i].eav;
    297      if((tmp1/(fr->step+1))<GMX_REAL_EPS)
    298        tmp1=0;
    299      bOK = bOK && do_real(tmp1);
    300      fr->ener[i].eav = tmp1;

(dbx) print tmp1
0.000000
(dbx) print fr->ener[i].eav
Denormalized number 0xc0087c30

Cheers,

Lynne


At 07:44 PM 17/04/2002 -0300, Sergio Oyama Junior wrote:
>Hi Erik and Lynne,
>
>Erik, thank you for your response. I'm using the Compaq C compiler v6.4-214.
>Yes, you are right. According to my observation, the occurrence of this
>exception is dependent on the energy file size.
>Considering Lynne's comments, do you think it is still worth to send you the
>file?
>
>Lynne, thank you for your advice. I will try the 3.0 version. I think I
>still have an old binary somewhere here!
>
>Best regards,
>
>Sergio
>
> > Hi,
> >
> > If you can create a not-to-large energy file that still crashes and send
> > it to me I'll have a look. What's the version of your compiler?
> >
> > Cheers,
> >
> > Erik
>
>
> > Sergio,
> >
> > No fix, but g_energy from v 3.0 works fine with .edr files from the newer
> > versions of mdrun. Let me know if you need the old binary, or just
> > recompile it yourself.
> >
> > Erik,
> >
> > I sent a small energy file to David, and debug output as I recall, but I
> > don't think he could replicate it. Certainly Bert de Groot said he wasn't
> > having problems with g_energy on his alpha machines.
> >
> > -Lynne

________________________________________________________________________________
Lynne E. Bilston, PhD
Senior Scientist, Prince of Wales Medical Research Institute, and
Conjoint Associate Professor, Faculty of Medicine, University of New South 
Wales
Barker St, Randwick, NSW 2031 Australia
Tel: +61-2-9382-7924	Fax: +61-2-9382-2643
________________________________________________________________________________