From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <45C1C335.5070007@domain.hid> Date: Thu, 01 Feb 2007 11:38:45 +0100 From: Wolfgang Grandegger MIME-Version: 1.0 Subject: Re: [Adeos-main] switchbench tests with IPIPE tracer disabled References: <70651.51913.qm@domain.hid> In-Reply-To: <70651.51913.qm@domain.hid> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit List-Id: General discussion about Adeos List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: poornima r Cc: adeos-main@gna.org poornima r wrote: > Hi, > > Thanks for the reply. > > Linux version:linux-2.6.18 > Xenomai: xenomai-2.3.0 (Stable version) > adeos patch: adeos-ipipe-2.6.18-ppc-1.5-01.patch OK, I'm curious, did you use the vanilla kernel from kernel.org? More comments below. > The tests were run as follows: > 1)The sampling period in the code for latency and > switchbench was changed to 1000000000ns(to remove > overrun error) > 2)switchtest was run with -n5 option > 3)cyclictest was run with -t5 option(5 threads > were created.) > 4)cyclictest was terminated with Illegal instruction > (after creating 5 threads) with IPIPE tracer enabled. > > These were the results without I-PIPE Tracer option: > (All the tests were run without any load) > 1)LATENCY TEST:- > User mode:- > /mnt/out_xen/bin# ./latency -t0 > == Sampling period: 1000000 us > == Test mode: periodic user-mode task > == All results in microseconds > warming up... > RTT| 00:00:01 (periodic user-mode task, 1000000 us > period, priority 99) > RTH|-----lat min|-----lat avg|-----lat > max|-overrun|----lat best|---lat worst > RTD| 167.000| 167.000| 167.000| 0| > 167.000| 167.000 > RTD| 176.000| 176.000| 176.000| 0| > 167.000| 176.000 > RTD| 168.000| 168.000| 168.000| 0| > 167.000| 176.000 > RTD| 171.000| 171.000| 171.000| 0| > 167.000| 176.000 > > Kernel mode:- > root@domain.hid# ./latency -t1 > == Sampling period: 1000000 us > == Test mode: in-kernel periodic task > == All results in microseconds > warming up... > RTT| 00:00:00 (in-kernel periodic task, 1000000 us > period, priority 99) > RTH|-----lat min|-----lat avg|-----lat > max|-overrun|----lat best|---lat worst > RTD| 123.000| 123.000| 123.000| 0| > 123.000| 123.000 > RTD| 125.000| 125.000| 125.000| 0| > 123.000| 125.000 > RTD| 128.333| 128.333| 128.333| 0| > 123.000| 128.333 > RTD| 127.000| 127.000| 127.000| 0| > 123.000| 128.333 > > Interrupt mode:- > root@domain.hid# ./latency -t2 > == Sampling period: 1000000 us > == Test mode: in-kernel timer handler > == All results in microseconds > warming up... > RTT| 00:00:01 (in-kernel timer handler, 1000000 us > period, priority 99) > RTH|-----lat min|-----lat avg|-----lat > max|-overrun|----lat best|---lat worst > RTD| 45.334| 45.334| 45.334| 0| > 45.334| 45.334 > RTD| 45.000| 45.000| 45.000| 0| > 45.000| 45.334 > RTD| 46.000| 46.000| 46.000| 0| > 45.000| 46.000 > RTD| 47.334| 47.334| 47.334| 0| > 45.000| 47.334 > RTD| 46.334| 46.334| 46.334| 0| > 45.000| 47.334 I remember similar figures from measurements under 2.4. I guess you tested without load!? Nevertheless, most of the latency is due to code execution time because the system is very slow and the caches are small. The sampling period is 1 second. I think "-p500" should already work. > 2)CYCLICTEST RESULTS:- > root@domain.hid# ./cyclictest -t5 > 5.14 3.71 1.72 6/31 216 > > T: 0 ( 0) P:99 I: 1000 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 1 ( 0) P:98 I: 1500 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 2 ( 212) P:97 I: 2000 C: 8112 Min: 169 > Act: 189 Avg: 204 Max: 288 > T: 3 ( 0) P:96 I: 2500 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 4 ( 216) P:95 I: 3000 C: 21596 Min: 180 > Act: 1279 Avg: 702 Max: 1336 Hm, this looks bogus. What returns "./cyclictest" without options (or -t1)? > 3)SWITCHBENCH TEST RESULTS:- > root@domain.hid# ./switchbench -n5 > == Sampling period: 1000000 us > == Do not interrupt this program > RTH| lat min| lat avg| lat max| > lost > RTD| 229.333| 45.666| 229.333| > 0 > > Test results with IPIPE tracer enabled > 1)LATENCY TEST RESULTS:- > User mode:- > root@domain.hid# ./latency -t0 > == Sampling period: 1000000 us > == Test mode: periodic user-mode task > == All results in microseconds > warming up... > RTT| 00:00:01 (periodic user-mode task, 1000000 us > period, priority 99) > RTH|-----lat min|-----lat avg|-----lat > max|-overrun|----lat best|---lat worst > RTD| 340.000| 340.000| 340.000| 0| > 340.000| 340.000 > RTD| 338.666| 338.666| 338.666| 0| > 338.666| 340.000 > RTD| 341.000| 341.000| 341.000| 0| > 338.666| 341.000 > RTD| 342.000| 342.000| 342.000| 0| > 338.666| 342.000 > > 2)kernel mode:- > root@domain.hid# ./latency -t1 > == Sampling period: 1000000 us > == Test mode: in-kernel periodic task > == All results in microseconds > warming up... > RTT| 00:00:00 (in-kernel periodic task, 1000000 us > period, priority 99) > RTH|-----lat min|-----lat avg|-----lat > max|-overrun|----lat best|---lat worst > RTD| 303.333| 303.333| 303.333| 0| > 303.333| 303.333 > RTD| 309.666| 309.666| 309.666| 0| > 303.333| 309.666 > RTD| 325.000| 325.000| 325.000| 0| > 303.333| 325.000 > RTD| 306.333| 306.333| 306.333| 0| > 303.333| 325.000 > > Interrupt mode:- > root@domain.hid# ./latency -t2 > == Sampling period: 1000000 us > == Test mode: in-kernel timer handler > == All results in microseconds > warming up... > RTT| 00:00:01 (in-kernel timer handler, 1000000 us > period, priority 99) > RTH|-----lat min|-----lat avg|-----lat > max|-overrun|----lat best|---lat worst > RTD| 153.334| 153.334| 153.334| 0| > 153.334| 153.334 > RTD| 154.667| 154.667| 154.667| 0| > 153.334| 154.667 > RTD| 164.334| 164.334| 164.334| 0| > 153.334| 164.334 > RTD| 154.667| 154.667| 154.667| 0| > 153.334| 164.334 > RTD| 163.667| 163.667| 163.667| 0| > 153.334| 164.334 The IPIPE tracer seems to add quit some code, Jan, does this look OK for you? > 2)CYCLICTEST RESULTS:- > root@domain.hid# ./cyclictest -t5 > 0.18 0.15 0.09 3/26 194 > > T: 0 ( 0) P:99 I: 1000 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 1 ( 0) P:98 I: 1500 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 2 ( 190) P:97 I: 2000 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 3 ( 0) P:96 I: 2500 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > T: 4 ( 193) P:95 I: 3000 C: 0 Min: 1000000 > Act: 0 Avg: 0 Max:-1000000 > Illegal instruction Can you post the output of /proc/xenomai/faults after the illegal instruction. > 3)SWITCHBENCHTEST RESULTS:- > root@domain.hid# ./switchbench -n5 > == Sampling period: 1000000 us > == Do not interrupt this program > RTH| lat min| lat avg| lat max| > lost > RTD| 667.333| 133.333| 667.333| > 0 > > > SWITCHTEST:- > Since FPU is not enabled for ppc, I was not able to > get proper results for switchtest. Please mail the > results of switchtest. "switchtest -n" will not use FPU instructions. Does this work? Thanks. Wolfgang. > Thanks, > Poornima > > > --- Wolfgang Grandegger wrote: > >> poornima r wrote: >>> Hi, >>> >>> I was able to run latency and switchbench tests on >>> MPC860 board with IPIPE tracer feature >>> enabled/disabled. >> Nice, what versions of Linux and Xenomai did you use >> for your tests? It >> is known that Linux 2.6 does not yet work properly >> for 8xx. It would >> also be nice if you could briefly post your results >> on this list as well. >> >>> Interrupt & scheduling Latencies and switching >>> latencies (with IPIPE tracer option disabled) were >>> having lower values when compared to latency >> values >>> with IPIPE tracer enabled. >> The ipipe tracer introduces some overhead (extra >> code) and as this CPU >> is very slow, you will notice it. >> >>> cyclictest test results were same with or without >>> IPIPE tracer enabled. >> Your results would really be of interest. >> >>> Is this reduced context switching latency in >>> switchbench test due to timer interrupt? >> Likely it also due to overhead introduced by the >> ipipe tracer. >> >> BTW: the switchtest should run as well with the >> option "-n" (for not >> using the FPU), >> >> Wolfgang. >> >> > > > > > ____________________________________________________________________________________ > 8:00? 8:25? 8:40? Find a flick in no time > with the Yahoo! Search movie showtime shortcut. > http://tools.search.yahoo.com/shortcuts/#news > >