Experimental Physics and
Industrial Control System

"Bob Dalesio" <[email protected]> · Sun, 20 Jun 2004 07:11:21 -0600

Hi James,

4K loops per second would mean that you could run 4K
instances of an analog input, PID record, and an analog
output record at once per second - or 400 of them at 10
Hz. The 68020 was a fairly slow processor. The current
power PCs are processing about 100K records per second,
leaving about 50% of the processor available.

These bench marks do not include the time to handle the
I/O - so any driver/hardware combination will need to be
added.

40% of the processor is left to allow the channel access
server time to communicate the values to channel access
clients around the network. This is a dynamic load that
does not show up at boot time.

The foot print question is best answered by others. The
last numbers I had in my head was: 500KB for vxWorks,
500KB for EPICS and space for the queues that channel
access will use. Consider this a preliminary answer.

Bob
On Sat, 19 Jun 2004 22:34:33 -0400 (EDT)
 "James A. Smith" <[email protected]> wrote:

Hi,
I'm new to the list and am trying to learn more about
EPICS after running

across a paper called "EPICS Architecture" by Dalesio,
Kraimer and

Kozubal. The paper discusses benchmarks for 68020
hardware, and I had a

question about that. The authors state
"The benchmarks for the database scan tasks show the
ability to close

4,000 analog loops per second leaving 40% of a 68020
available for the

sequencer and channel access services. The maximum
periodic repetition

rate currently available is 60Hz. Using interrupt on
end-of-conversion

hardware allows for rates higher than 60Hz."
The way I read the first sentence, it seems that a
(low-level) control

loop with analog input and analog output can run at 4kHz.
(Please correct

me if I'm off-base) I'm not sure what a "periodic
repetition rate" is,

though. Would someone be able to clarify that for me?
Oh, while I'm at it, does anyone know what the smallest
memory footprint

that EPICS requires?
thanks! :-)
James

Bob Dalesio
[email protected]
410 557 0297 Maryland
505 699 1632 Cell Phone

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System