I have 2 EPICS IOCs running reliably for several years now at 500Hz
(MVME3431). The load of the CPU induced by this high clock rate seems
negligeable. For my application, this rate is useful for ADC periodic
scanning but has nothing to do with the periodic DB scanning. I tried 1000Hz
but I had unattended hangs of the system without any clear explanation.
J.F. Gournay
CEA Saclay
DAPNIA/SIS
tel +33 (0)1 69 08 70 32
fax +33 (0)1 69 08 63 01
-----Message d'origine-----
De : Steven Hartman [mailto:hartman@fel.duke.edu]
Envoyé : lundi 17 novembre 2003 17:16
À : EPICS tech-talk
Objet : Re: EPICS periodic scan and sysClkRateSet()
On Mon, 17 Nov 2003, David Maden wrote:
> If you change the clock rate, be careful with the "delay" function of
> the sequencer, at least with version 1.9.2.Beta12. The argument is no
> longer seconds if the tick rate is not 60 Hz.
Thanks. We are using sequencer-1.9.5 and it looks like this problem has been
fixed by this version. In seq-1.9.5, all the delays are relative to
sysClkRateGet(), so your conversion factor is no longer needed.
Note that this same problem does appear in some of the old device and driver
support which is included with base in 3.13. Other places in base
(3.13.7) where I have found taskDelay() using a constant include:
ca/vxWorks_depen.c
db/dbCa.c
rec/recWaitCa.c
--
Steve Hartman
hartman@fel.duke.edu | 919-660-2650
Duke Free Electron Laser Laboratory
- Navigate by Date:
- Prev:
RE: EPICS periodic scan and sysClkRateSet() Jeff Hill
- Next:
bug in RULES.Host R3.13.* Dirk Zimoch
- Index:
1994
1995
1996
1997
1998
1999
2000
2001
2002
<2003>
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
- Navigate by Thread:
- Prev:
ask the ext_pref-edm.tar file guobao shen
- Next:
bug in RULES.Host R3.13.* Dirk Zimoch
- Index:
1994
1995
1996
1997
1998
1999
2000
2001
2002
<2003>
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
|