Re: interrupt latency

yodaiken@chelm.cs.nmt.edu
Wed, 12 Aug 1998 03:53:39 -0600

On Wed, Aug 12, 1998 at 12:35:16AM -0700, Linus Torvalds wrote:
> However, usually when people are interested in interrupt latency, they
> aren't so much interested in how long an interrupt takes to dispatch, but
> how long interrupts may be disabled by whatever code that needs to do some
> critical region or whatever.
>
> And that is never going to be tested by a software interrupt. The whole
> point of that kind of interrupt latency is to see how quickly the OS can
> respond to an _unexpected_ and asynchronous thing.

I don't see how Larry can do it in lmbench. It would not be a complex
kernel patch in vanilla Linux and it could be done in a module in RTLinux,
but how then to get the numbers for the feeble OSs?

Unsolicited testimonial on RTL mailing list:

"The Controller now runs safely a longer time ( 8-10 seconds we need)
with 4000 Hz on a Pentium 120 and does it's job very well (realize: in
the Windows-World you reach at this point of time 200 Hz (No, not 2000
Hz 200 Hz!) on a 300 Mhz Pentium with something like DT-Vee and a 8000
Deutsch Marks A/D converter card! They tried to sell us a DSP card for
15000 Deutsch Marks without software! We only smiled and showed them our
Linux box. Imagine their faces :-) "

-- 


---------------------------------
Victor Yodaiken
Department of Computer Science
New Mexico Institute of Mining and Technology
Socorro NM 87801
Homepage		http://www.cs.nmt.edu/~yodaiken
PowerPC Linux page	http://linuxppc.cs.nmt.edu
Real-Time Page		http://rtlinux.org




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html