On Fri, 06 Nov 2015 14:18:18 +0100,
"Martin Patzak (GMX)" martin.pat...@gmx.de wrote:
>> I cannot confirm Lorens numbers, but end up in the ballpark given by
>> Stefano:
>>
>> mnm@vincent:~/M/_Linux/python/_mnms_tests$ python pyownet_timing.py
>> //razmaban:4304
>> pyownet: ver. 0.8.2
>> (/usr/local/lib/python2.7/dist-packages/pyownet/__init__.pyc)
>> proxy_obj: ownet server at ('10.67.90.173', 4304)
>> server info: pid 21863, ver. 2.9p8
>>
>> timeit:
>> statement: proxy_obj.dir("/")
>> number: 20
>> repetitions: 5
>>
>> ** non persistent : 2.40 ms, 2.40 ms, 2.43 ms, 2.44 ms, 2.49 ms,
>> ** persistent : 1.82 ms, 1.82 ms, 1.86 ms, 2.05 ms, 2.12 ms,
>>
>> The test was run on a remote (local LAN) debian machine connecting to
>> an owserver running on a bananapi running debian jessie with prebuilt
>> owfs debian package version 2.9p8.
>> The tested 1wire net consists out of LinkUSB
>> <http://owfs.org/index.php?page=linkusb> as master, 2 DS2408
>> io-modules and 25 DS28B20 temperature sensors.
>>
>> Cheers
>>
>> Martin
>>
> for the sake of science, the same test locally on the bananapi:
>
> mnm@razmaban:~/python$ python pyownet_timing.py
> pyownet: ver. 0.8.2
> (/usr/local/lib/python2.7/dist-packages/pyownet/__init__.pyc)
> proxy_obj: ownet server at ('127.0.0.1', 4304)
> server info: pid 21863, ver. 2.9p8
>
> timeit:
> statement: proxy_obj.dir("/")
> number: 20
> repetitions: 5
>
> ** non persistent : 3.18 ms, 3.24 ms, 3.24 ms, 3.25 ms, 3.47 ms,
> ** persistent : 2.37 ms, 2.39 ms, 2.40 ms, 2.42 ms, 2.58 ms,
>
>
> Loren, there could be something at odds with your setup.
>
> Martin
There could be many things weird about my setup... But I can't imagine what
could cause this one. I have two DS18B20 on a BBB GPIO port via the w1 system.
Ubuntu 14.04 with standard distribution owfs package. But owserver should
isolate this test from any dependencies on the hardware connection - right? I
know reading uncached values on the BBB takes about 1.1 seconds, so that can't
be the problem. I do end up with about 100 port 4304 TIME_WAIT lines in
netstat.
I just watched the connection from Windows to BBB in Wireshark, and can see the
difference between non-persistent and persistent. It takes the BBB about 1 mS
to ACK a request from Windows, and about 40 mS to respond with a reading. All
other times are way less than 1 mS (at least from Windows' point of view).
Windows is connected via Wi-Fi to the router the BBB is hardwired to. Looks
like ping times between them are about 1.1 mS, so that explains the ACK delay
but not the program response timing.
It takes a similar 49 mS for owhttpd to respond to the first GET for its web
page. About 2.2 seconds to GET a temperature sensor, I guess that triggers
uncached readings of my two sensors.
ii owserver 2.8p15-1ubuntu4 armhf Backend server for 1-Wire control
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 22210 0.0 0.2 37352 1108 ? Ssl Nov05 0:28
/usr/bin/owserver -c /etc/owfs.conf --pid-file /var/run/owfs/owserver.pid
A minimal search suggests BananaPi is not significantly faster than BBB, so
what could be going on here?
>>>> Somehow your figures are consistent with this interpretation:
>>>> non-persistent overhead is a constant small delta with respect to
>>>> persistent connection. Since your owserver is apparently much slower
>>>> than the one which I run my tests, you have a smaller relative
>>>> overhead.
Looks like the open/close of the network connection is sub-millisecond even on
my BBB. It is only when I ask for a value that I hit the 40 to 50 mS delay.
| Loren Amelang | lo...@pacific.net |
------------------------------------------------------------------------------
_______________________________________________
Owfs-developers mailing list
Owfs-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/owfs-developers
