Menu
▾
▴
Re: [Owfs-developers] Owserver performance with i2c master
|
From: Paul A. <pau...@gm...> - 2011-02-11 03:23:42
|
Looking at the log, it's a little hard to tell. The DS2482 can only perform 1 byte at a time. If you run a newer version (2.8p6) and compile with ./configure --enable-owtraffic && sudo make install you will be better able to see the actual data requests and responses. Paul Alfille On Thu, Feb 10, 2011 at 5:23 PM, "Roman Jałoza - TechBase Sp. z o.o." <r.j...@a2...> wrote: > I will attach output from debug mode with writing/reading PIO from ds2408 > and reading temperature from ds18b20 sensor. > There is a lot of repeated command, is it right? > > Thanks a lot > > Roman > > W dniu 2011-02-10 23:00, Pascal Baerten pisze: > > as the i²c commands overhead is little, 100kHz i²c is fast enough for 1wire > communication, which tops at 15kHz in standard speed. Also the ds2482 does > not support 1000kHz, max 400kHz > > I use owserver with both owfs and owhttpd. But for testing performance > issues, I avoid accessing the bus from http which access every properties of > the browsed device and produce lot of debugging output. > > accessing the file via owfs (uncached) directory (without autocompletition > that produce bus access) allows to get debug output only on the part you > want to analyse. > putting a digital scope on the DQ line may help to better understand what's > going wrong. > > I don't see anything wrong on your owserver configuration. > > > pascal > > > 2011/2/10 "Roman Jałoza - TechBase Sp. z o.o." <r.j...@a2...> >> >> One more thing... >> >> You are using owserver alone or with owfs? >> I'm running owserver with configuration: >> >> device = /dev/i2c-0 >> Celsius >> timeout_volatile = 60 >> timeout_stable = 60 >> timeout_directory = 30 >> timeout_presence = 60 >> timeout_network = 0 >> pid_file = /mnt/mtd/owpid >> foreground >> error_print = 1 >> error_level = 3 >> timeout_persistent_low = 600 >> timeout_persistent_high = 3600 >> format = fic >> cache_size = 0 >> >> Maybe there is something wrong? >> >> >> W dniu 2011-02-04 07:22, Pascal Baerten pisze: >> >> here the output of a simple test in php (using ownet.php) via owserver on >> debian nslu2 owfs 2.8p3 and busmaster is 2482-800 >> reading pio is ~50ms >> reading temperature is ~770ms >> reading cached temp (frm owfs memory) is 15ms >> >> >>> measuring timming for temperature and pio >>> connect to ownet duration=3ms >>> temperature for /bus.0/uncached/28.0521AD010000/temperature = >>> 17.6875 duration=771ms >>> waiting 5s duration=5006ms >>> set /bus.0/12.5B1E6D000000/PIO.B=1 duration=118ms >>> read /uncached/bus.0/12.5B1E6D000000/sensed.B=0 duration=49ms >>> set /bus.0/12.5B1E6D000000/PIO.B=0 duration=85ms >>> read /uncached/bus.0/12.5B1E6D000000/sensed.B=1 duration=46ms >>> waiting 1s duration=1002ms >>> temperature for /bus.0/28.0521AD010000/temperature = >>> 17.6875 duration=15ms >>> done >>> >> >> >> >> 2011/2/3 Roman Jałoza <r.j...@a2...> >>> >>> How looks your timings? How fast can you read temperature or PIO? >>> Are these timings are similar to my? (650ms to PIO and 1-1,2 s for temp >>> without converT command ) >>> >>> Roman >>> >>> W dniu 03.02.2011 21:44, Pascal Baerten pisze: >>> >>> Hi Roman, >>> >>> I'm using owfs with ds2482-800 and found the communication slower than >>> other bus masters solutions. >>> However not that slow than you describe. >>> >>> Analysing 1-Wire communication with a digital scope shown that an idle >>> period is visible between each byte and "tri-bits". >>> This idle period is around 1,3ms for my system (nslu2 with debian), this >>> reduce the effective throughput to 4kbit/s instead of 15kbit/s. >>> >>> Do you have an hadware assited i2c controller embedded or are you using >>> software i2c? >>> I think that my behaviour come from software i2c that does not buffer >>> transactions and require attention from owfs between each byte and tri-bits. >>> >>> Regarding the temp reading, I think you better to read 'temperature' >>> property after a simultaneous converT command instead of 'fasttemp' which is >>> not prepared by the simultaneou command. - Paul correct me if not true - >>> >>> You may also find usefull information by starting your owfs daemon with >>> '--debug' param. Reading temp involve a blocking transaction that wait until >>> conversion is finished by polling 1-wire or for fixed time depending on >>> method used and parasite power mode. >>> >>> Pascal >>> >>> >>> >>> 2011/2/3 "Roman Jałoza - TechBase Sp. z o.o." <r.j...@a2...> >>>> >>>> Hi, >>>> >>>> I'm developing 1-wire communication at embedded computer with ARM9 and >>>> ds2482-100 master chip. >>>> Is anyone test owfs performance via i2c master? >>>> >>>> I'm reading temperature from ds18b20 sensors and it took a long time >>>> (from 1,5s to 8s) with fasttemp read. It should be much faster. I've made >>>> some tests with writing simultaneous/temperature bit but performance was >>>> still the same - means poor. >>>> >>>> Reading PIO states from ds2408 device took about 600ms (but sometimes >>>> 1,5s). >>>> >>>> Devices are powered (reading with parasite mode took 5-8s). >>>> >>>> Is anyone have some idea? some solution? >>>> >>>> -- >>>> >>>> Best Regards >>>> >>>> Roman Jałoza >>>> >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)! >>>> Finally, a world-class log management solution at an even better >>>> price-free! >>>> Download using promo code Free_Logger_4_Dev2Dev. Offer expires >>>> February 28th, so secure your free ArcSight Logger TODAY! >>>> http://p.sf.net/sfu/arcsight-sfd2d >>>> _______________________________________________ >>>> Owfs-developers mailing list >>>> Owf...@li... >>>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>>> >>> >>> >>> >>> -- >>> Pascal >>> www.brain4home.eu >>> to subscribe, send a mail to <lis...@br...> >>> >>> >>> ------------------------------------------------------------------------------ >>> The modern datacenter depends on network connectivity to access resources >>> and provide services. The best practices for maximizing a physical >>> server's >>> connectivity to a physical network are well understood - see how these >>> rules translate into the virtual world? >>> http://p.sf.net/sfu/oracle-sfdevnlfb >>> >>> _______________________________________________ >>> Owfs-developers mailing list >>> Owf...@li... >>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>> >>> >>> ------------------------------------------------------------------------------ >>> The modern datacenter depends on network connectivity to access resources >>> and provide services. The best practices for maximizing a physical >>> server's >>> connectivity to a physical network are well understood - see how these >>> rules translate into the virtual world? >>> http://p.sf.net/sfu/oracle-sfdevnlfb >>> _______________________________________________ >>> Owfs-developers mailing list >>> Owf...@li... >>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>> >> >> >> >> -- >> Pascal >> www.brain4home.eu >> to subscribe, send a mail to <lis...@br...> >> >> >> ------------------------------------------------------------------------------ >> The modern datacenter depends on network connectivity to access resources >> and provide services. The best practices for maximizing a physical >> server's >> connectivity to a physical network are well understood - see how these >> rules translate into the virtual world? >> http://p.sf.net/sfu/oracle-sfdevnlfb >> >> _______________________________________________ >> Owfs-developers mailing list >> Owf...@li... >> https://lists.sourceforge.net/lists/listinfo/owfs-developers >> >> >> ------------------------------------------------------------------------------ >> The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE: >> Pinpoint memory and threading errors before they happen. >> Find and fix more than 250 security defects in the development cycle. >> Locate bottlenecks in serial and parallel code that limit performance. >> http://p.sf.net/sfu/intel-dev2devfeb >> _______________________________________________ >> Owfs-developers mailing list >> Owf...@li... >> https://lists.sourceforge.net/lists/listinfo/owfs-developers >> > > > > -- > Pascal > www.brain4home.eu > to subscribe, send a mail to <lis...@br...> > > ------------------------------------------------------------------------------ > The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE: > Pinpoint memory and threading errors before they happen. > Find and fix more than 250 security defects in the development cycle. > Locate bottlenecks in serial and parallel code that limit performance. > http://p.sf.net/sfu/intel-dev2devfeb > > _______________________________________________ > Owfs-developers mailing list > Owf...@li... > https://lists.sourceforge.net/lists/listinfo/owfs-developers > > ------------------------------------------------------------------------------ > The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE: > Pinpoint memory and threading errors before they happen. > Find and fix more than 250 security defects in the development cycle. > Locate bottlenecks in serial and parallel code that limit performance. > http://p.sf.net/sfu/intel-dev2devfeb > _______________________________________________ > Owfs-developers mailing list > Owf...@li... > https://lists.sourceforge.net/lists/listinfo/owfs-developers > > |