[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
Re: [rtl] Release 9F Problems...
On Thu, 1 Oct 1998, Kurt Rosenfeld wrote:
> Wow. A picture is worth a thousand words. I understand the blue trace
> but what is the pink trace in the bit-flipping example? Are you using the
> make_periodic function or is your task a handler(responding to whtever
> generated the pink trace)? Can you do an FFT or a histogram on the
> bit-flipping case? Under different conditions, like running the same
> code under the regular Linux scheduler? It would be so interesting! I
> have always looked at the trace of the bit-flipping on my scope but could
> never see it in the frequency/PDF domain. Thanks for posting your data.
> -kurt
>
The pink trace is the current trace, the live signal, the blue traces are
stored traces - so the signal is jumping around (offset) in time - I
believe this shows what real real-time engineers call "jitter" (I'm a
simple chemist so I cannot say for sure :).
This is a make_periodic task at 40 us on a Pentium 166MHz (ide disk, 64 MB
DRAM, PCI tulip fast ethernet card, PCI video card (but no monitor,
vnc is used to access the box), ISA CIO-DAS16/Jr DAQ card (not used), and
a National Instruments' PCI DIO96 digital input/output card). The task
shown is from a single line of a 32 line write to the PCI DIO card (the
task runs one bit pattern in the first period and another bit pattern on
the second period - the line monitored on the scope is a line which
changes between the two). While the task is running the box is being
accessed via the fast ethernet card through the vnc server (this is an X
server hooked to a frame buffer server which communicates with a "thin"
vnc client on an NT4 box - it is really very cool, unlike regular X the
client is stateless!).
I will try to provide an FFT shortly - hopefully later today. I don't
know how good the FFT gizmo is on this scope, I may need a real analyzer
to do this - as the digital storage shows the trace is not really
stationary - won't that mess up an FFT?
I will also try to rig up a regular-vs-rt linux comparison, but this would
probably be better done with parallel or serial port lines as opposed to
this special i/o card (that way other people could use it too on different
systems). I think there is an example app suitable for this. The way
things are going right now it may be a while before I get to this.
Thanks for taking a peek at the traces,
-Don
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail majordomo@rtlinux.cs.nmt.edu OR
echo "unsubscribe rtl <Your_email>" | mail majordomo@rtlinux.cs.nmt.edu
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/