Summary: Queuing Systems

[Wolfgang Sauer <sauer ( ( at ) ) organik.uni-erlangen.de>]

 : Subject: Summary: Queuing Systems.
 Sorry to keep you waiting, but I have been sick last week.
 Besides several requests for the summary and a few announcements like
 "... just installing XXX, report on first experiences will follow ..."
 (keep them coming, and preferably some words on WHY you choose XXX)
 there was very little resonance. Thanks to all who responded or will do so
 in the future.
 Anyway, here it is:
 Original request:
 < Dear Netters,
 <
 < As clusters of networked workstations are becoming ever more popular in the
 < Comp Chem community, so do queuing systems to utilize their power such as
 < NQS (public domain and proprietory versions), DQS, ...
 < I'd like to hear about the experiences people have made with these systems,
 < advantages and disadvantages of the respective systems (ease of use,
 < performance ...).
 apparently I forgot to include "ease of installation and managment".
 Brian Salter-Duke (B_DUKE ( ( at ) ) DARWIN.NTU.EDU.AU) wrote:
 >
 > I have just installed the public domain NQS on a stand alone small
 > IBM RISC 6000 - a M2A (like M20). It is too early to give opinions, as
 > I clearly havent understood it fully yet or perhaps configered it properly.
 >
 > What I really need is a simpler queueing system for a mixed job lot -
 > Gaussian, Mopac, gamess ets running onew job at a time form several users.
 > NQS is over-kill.
 Wolfgang Ksoll <woks4000 ( ( at ) ) mailszrz.zrz.TU-Berlin.DE> gave me a
 lead to his
 (very recommended) paper "Parallel und Batch Scientific Computing".
 You can
 get it by anon FTP from ftp.zrz.tu-berlin.de:/pub/aix/parallel/para.ps
 > On 16 pages it gives an introductory view on parallel computing from the
 point
 > of a chemist. In his chapter "Queuing Systems" he deals with
 >	NQS (PD, Convex, Cray)
 >	CERN/NQS
 >	NQS/Exec
 >	DQS
 >	DNQS
 >	VQS
 >	CONDOR
 >	IBM LoadLever
 > together with a short description of each, he gives references and sources.
 > The paper is in German.
 > In his mail he also mentioned the DQS-based
 >	CODINE
 > and that he preferred DQS for the well advanced functionality.
 Someone here mentioned yet another system:
 >
 > BATCH3.3 by Scott Presnell, Dept. Pharm. Chem., UCSF
 === End of summary ===