Re: Year 68 Virus


Subject: Re: Year 68 Virus
From: Chad Schafer (cschafer@stat.Berkeley.EDU)
Date: Thu Sep 10 1998 - 17:28:10 MDT


Hello.

It was also some time ago when I dealt with this, and the problem
is indeed that the number of seconds gets to be too large. Assuming
that this portion of the code has not changed greatly in the past
few years, the problem can be seen by going into the procedure
caldyi.F; the variable which gets too large is "nsecs". Let me
know and I can send you a copy of the fix I made. I handled it
by storing nsecs as two integers. Of course, all of the points
where calculations involved nsecs needed altering, but as I recall
the only changes needed were in caldyi.F.

Chad Schafer

On Thu, 10 Sep 1998, Andy Mai wrote:

> On Thu, 10 Sep 1998, Kevin Werner wrote:
>
> > Hello.
> >
> > I have been running the CCM at t31x15 with 18 levels. I am running the
> > model on a SUN machine with multiple processors (although the result is
> > invariant to the number of processors). The model has consistently crashed
> > after 68 years and one month at timestep 1789568. I have encountered the
> > same result with both versions 3.2 and 3.6. Is anyone aware of a solution
> > to this problem? Or ideas on how to go about fixing it?
> >
> > Thanks,
> > Kevin Werner
> >
> > o__ Kevin Werner INTERNET: werner@atmos.washington.edu
> > _,>/'_ 685-8438 UUCP: uw-beaver!atmos.washington.edu!werner
> > (_) \(_)......... WEB: http://www.atmos.washington.edu/~werner/
> > Dept of Atmospheric Science, Box 351640, University of Washington
>
>
>
> Ah, a fellow cyclist...
>
> This is the same problem as Unix running out of dates in the year 2038.
>
> Yes, I've had this problem, too, but it was a long time ago. It happens because
> at some point in the model a calculation is done which involves the number of
> seconds since the beginning of the run. So after 2**31-1 seconds =~ 68 years,
> things go haywire. If someone else doesn't point out the relevant spot in
> the code for you, send me another message and I'll try to locate it. BTW,
> which version of CCM are we talking about here?
>
> Andy Mai
> 303-497-1391
>
>
>
> --- __o Andrew W. Mai
> ---- _`\<,_ National Center for Atmospheric Research
> --- (*)/ (*) Boulder, CO 80307 USA mai@ucar.edu
>
>



This archive was generated by hypermail 2b27 : Thu Jun 01 2000 - 09:06:52 MDT