home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.vms
- Path: sparky!uunet!cis.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!agate!linus!philabs!crpmks!billb
- From: billb@crpmks.uucp (Bill Bochnik (Info Systems))
- Subject: Re: TUNING - QUANTUM SYSGEM parameter
- Reply-To: billb@crpmks.UUCP (Bill Bochnik (Info Systems))
- Organization: Ciba-Geigy Corporation
- Date: Fri, 13 Nov 1992 01:48:46 GMT
- Message-ID: <1992Nov13.014846.5404@crpmks.uucp>
- References: <1992Nov12.144951.485@hhcs.gov.au>
- Lines: 30
-
- In article <1992Nov12.144951.485@hhcs.gov.au> pihlab@hhcs.gov.au writes:
- >
- >First some preamble...
- >
- >Have you noticed the SYSGEN parameter QUANTUM.
- >
- >By default it is set to 20 (units of 10ms each).
- >
- >This means that if 5 processes each use their entire CPU timeslice (QUANTUM)
- >then 1 second has elapsed before another process can have some CPU time.
- >
- >A QUANTUM of 100 equals 1 full second. Minimum setting is 2.
- >
- >I bring this up because we have a large range of different capacity VAX boxes
- >and all of them use the DEFAULT QUANTUM value of 20 and all of them have
- >at least one CPU intensive process hogging the machine to the detriment of
- >all other users. If you have several high CPU users then your average
- >response time per user becomes unbareable.
-
- From what I understand (and I may be wrong), the faster machines can handle
- the QUANTUM being lower, as the overhead on thode machines is less (9000's,
- 7000's, 6000's; not recommended on 8000's!).
-
-
-
- --
- Bill Bochnik | It's hard to be a James Bond in an Abbott
- Systems Analyst | and Costello world.
- Ciba-Geigy Corporation | Signed and sealed, they deliver oblivion.
- philabs!crpmks!billb | It's worse than that, he's dead Jim.
-