Re: PREEMPT_RT vs ADEOS: the numbers, part 1

From: Ingo Molnar
Date: Sat Jun 11 2005 - 14:30:22 EST



a couple of other wrokloads that are easy to measure and are useful for
triggering worst-case latencies:

hackbench:

http://developer.osdl.org/craiger/hackbench/
http://developer.osdl.org/craiger/hackbench/src/hackbench.c

it creates tons of threads and does message-passing between them. E.g.
"hackbench 50" or "hackbench 100" is a pretty good test.

i use 40 copies of LTP running in parallel:

while true; do ./runalltests.sh -x 40; done

this is good at triggering worst-case latencies too. Plus dbench is good
too:

http://samba.org/ftp/tridge/dbench/
http://samba.org/ftp/tridge/dbench/dbench-3.03.tar.gz

# dbench-3.03> ./dbench 50 -c ./client.txt

also, there's a very good on-host IRQ-latency measurement tool as well:

http://www.affenbande.org/~tapas/wiki/index.php?rtc_wakeup

this uses TSC timestamping to detect wall-clock delays of the RTC
interrupt. The tools jumps through lots of hoops to make sure the
numbers are reliable. If you run it under the -RT kernel then run the
RTC IRQ thread at a higher-than-all-other-threads priority:

chrt -f 95 -p `pidof 'IRQ 8'`
./rtc_wakeup -f 1024 -t 100000

(when you run it, and if you also have RTC_HISTOGRAM enabled in your
.config, then the kernel will print a histogram when you stop
rtc_wakeup, so you've got two sources of information.)

Ingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/