Re: pfConfused about environments - HELP

New Message Reply Date view Thread view Subject view Author view

Scott McMillan (scott++at++ht.com)
Thu, 27 Mar 1997 11:35:25 -0500 (EST)


Rob Jenkins writes
>
> Yikes :-)
>
> You might try running the app with par from the 'good' and 'bad' environments
> respectively, direct output to files and diff them - something must be
> different. Does 'limit' show anything different between the twp $HOMEs ?
>
> What is the problem you see BTW ?
>
> Cheers
> Rob

Simon Bennett writes:
>
> Can you define "unstable" further?
>
> Are both home directories on the same disk - same sort of filesystem? (XFS,
> EFS etc..)
>
> How is the script set up?
>
> If you don't specify -f for [t]csh - then it'll load your $HOME/.[t]cshrc
> before it starts the script....
>
> If it's not already there try adding
>
> #!/usr/bin/tcsh -f
>
> to the the top of your script.

In detail, this is the problem: I do dynamic simulation...that is I develope
and implement involved algorithms to simulate articulated mechanisms (like
robots). To do this I have to detect collisions and compute the forces of
interaction between bodies, and then compute the accelerations of each.
Fixed stepsize numerical integration is used to compute the updated positions
and velocities at each time step. These positions update the DCS nodes in a
scene graph.

By unstable, I mean the animation/simulation is going loopy in the "bad"
environment. I does not look like an integration stepsize problem but rather
bad data is creeping into the simulation from somewhere. It appears to be a
numerical difference that diverges (but I haven't been able to track anything
down yet). I installed all of the C++ patches (except 1787) for the 7.1 IDO,
but, in retrospect, it is a waste of time because it runs fine with HOME set
to my account on the machine in question.

Answers to the suggestions:
 - I haven't tried par yet
 - `limit` shows nothing different
 - both accounts are on (different) XFS disks
 - there is no script - the executable is invoked directly from the command
                        line

We've also been through environment variables in greate detail....nothing is
obviously wrong though.

I don't think this should be happening....It has me VERY worried because it
means that my executable is not portable.

Yikes is right, or ACK!

scott

-- 
  Scott McMillan  |       HT Medical, Inc.      | Developing virtual environ-
   scott++at++ht.com   |      http://www.ht.com      | ment medical and surgical
 Ph: 301-984-3706 | 6001 Montrose Rd., Ste. 902 | simulations and surgery
Fax: 301-984-2104 |     Rockville, MD 20852     | simulation creation tools.

======================================================================= List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/ Submissions: info-performer++at++sgi.com Admin. requests: info-performer-request++at++sgi.com


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:54:58 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.