Newsgroups: comp.parallel.pvm
From: heller@nirvana.imo.physik.uni-muenchen.de (Helmut Heller)
Subject: PVM on SGI Challenge??
Organization: Leibniz-Rechenzentrum, Muenchen (Germany)
Date: 14 Sep 1994 09:25:39 GMT
Message-ID: <356fij$r69@sunserver.lrz-muenchen.de>


Hello,

since several months I am trying to run my application, a large molecular  
dynamics program named EGO, under PVM on the SGI Challenge (shared memory  
version of PVM), but so far without success. Needless to say that my  
program works just fine on regular workstation clusters (NeXT, SUN, SGI,  
RS6000, HP), the Cray T3D, and the IBM SP1, all under PVM, version 3.2.3.

I have had contact with Weicheng (wjiang@cs.utk.edu), the author of this  
PVM version, but he left several months ago for China and I could not get  
back in contact with him since. 

I am using version 3.3.0b09 of PVM. I already changed my program to deal  
with some idiosyncrasies of the SGIMP-PVM (have to manually free all  
buffers after usage, cannot swap buffers, problem with returning task  
IDs...). Here is the description of my current error:

isere(2:169)> more pvml.211
[t80040000] ready  3.3.0b09   Fri Jul  8 22:55:35 1994
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad pa
[t80040000] [t4000c] rameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]:
[t80040000] [t4000c] pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t
[t80040000] [t4000c] 4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]:
[t80040000] [t4000c] pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble()
[t80040000] [t4000c] : Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]: pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] libpvm [t4000c]:
[t80040000] [t4000c] pvm_upkdouble(): Bad parameter
[t80040000] [t4000c] 262156: Out of memory! Program stops.
[t80040000] dm_halt() from (isere), halting...
[t80040000] work() pvmd halting
[t80040000] pvmbailout(0)


one of the worker nodes gives this "Bad parameter" message  (do you have  
any idea what that might mean?) and then runs out of memory! Since the test  
molecule is VERY small, under normal conditions this CANNOT be (and does  
not happen during test runs with other PVMs on other machines). I suspect  
that the out-of-memory is a result of some corrupted data due to the "Bad  
parameter" error before.

What experience do other people have with PVM on the SGI Challenge? Is  
3.3.0b09 the latest version of PVM for that machine? Can someone help??

Thanks a lot for your help,
bye,
Helmut



--
Servus, Helmut  (DH0MAD)   ______________NeXT-mail accepted________________
Phone: +49-89/2394-4565    "Knowledge must be gathered and cannot be given"
heller@nirvana.imo.physik.uni-muenchen.de             ZEN, one of BLAKES7
FAX:   +49-89/2394-4607    ------------------------------------------------
Helmut Heller, Ludwig Maximilians University, Institute for Medical Optics
Theoretical Biophysics Group, Room 230

