The granularity of the I/O Event scanning has always been in
the hands of the programmer.
While it is true that you could do make a seperate scan list
for each bit and that this would minimize the record
processing, you are trading this off against the length of
the scan list on the I/O even scanner and the checking of
the bits in the interrupt service routine. As the context
switch to the I/O event scanner is the real overhead, not
the record processing, it seems fine to put the granularity
equal to the granularity of the interrupt. I believe that
the binary records on a power PC take less than 1 usec each.
The context switch take over 30 usecs. It seems that the
code in the interrupt service routine would also be clearer
in the case where the events are posted once per card.
On the other hand, if the record processing chains that are
fired off for any of these bits is really horrendous, you
may want to have a seperate event for each bit.
Bob
- References:
- IOSCANPVT Jim Thomas
- Navigate by Date:
- Prev:
IOSCANPVT Jim Thomas
- Next:
Re: IOSCANPVT Marty Kraimer
- Index:
1994
1995
1996
1997
1998
1999
2000
<2001>
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
- Navigate by Thread:
- Prev:
IOSCANPVT Jim Thomas
- Next:
Re: IOSCANPVT Marty Kraimer
- Index:
1994
1995
1996
1997
1998
1999
2000
<2001>
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
|