Re: [rtl] *Fast* IRQ response ( 2.08us )

[Paul Koning <pkoning@xedia.com>]

>>>>> "1698" == 1698  <C.> writes:

 1698> Greetings,

 1698> I just measured the time from the leading edge of the IRQ
 1698> signal to the leading edge of a pulse generated by my rt-linux
 1698> handler at 2.08us most of the time and sometimes just under
 1698> 4us.  The measurements made using a Lecroy 9384 1gHz digital
 1698> scope with one probe on the irq signal and the other on the LPT
 1698> port I'm pulsing.

 1698> The system is 2.0.29 rtl 0.6, 200mhz p5 with 64mb ram, 3-4gb
 1698> scsi drives, 500ms/s Gage PCI waveform digitizer, and an ISA
 1698> board full of specialized PIC chips.

 1698> Looking at the signals on the scope, they appear virtually
 1698> jitter free and very clean.  Even when running " find / -name
 1698> dlfkldf -print " the signals remain stable.  This is just the
 1698> opposite of how windoze reacts of course.

Neat.

It's interesting to compare that with a so-called "real time system"
that is being marketed quite a lot right now.

>From "Embedded Systems Development" magazine (first issue, 3/98), page 
10: 
	"...Microsoft claims an interrupt response time of 0.5 ms..."

So RTLinux is about a factor 100 better.

	paul
--- [rtl] ---
For more information on Real-Time Linux see:
http://rtlinux.cs.nmt.edu/