›
This is the forum archive of Homey. For more information about Homey, visit the Official Homey website.

The Homey Community has been moved to https://community.athom.com.

This forum is now read-only for archive purposes.

Very bad response time

MennoVanDerAatMennoVanDerAat Member
in Archive
I was wondering is it just me or.............
Homey is not doing anything at the moment also after reboot same result
other wireless devices in my house have a response of <1ms (like it should be)
Only thing I tried is installing IFTTT by following the instruction movie




«1

Comments

  • brot84brot84 Member
    Maybe I dont understand your Question not correct but IFTTT is a time Critical Process. Sometimes its very bad response time
  • Fire69Fire69 Member
    The problem is not IFTTT, but his Homey that seems to respond very slowly to network pings.

    Tried it here from work over a VPN and my response times are better then yours.
    Sure it's not a problem with your network?
  • spamenigmaspamenigma Member
    Same for me, I do have similar looking response times compared to 2-3ms to my phone.



    Apps added (for comparison) but not yet using since not much works:
    Belkin, Wikipedia, HTTP request actions, SoundCloud, IFTTT.
  • JeroenKleijnJeroenKleijn Member
    I have the same.  It is possible since lowest is 14ms. Although i also have a max of 805ms..

  • jordenjorden Member
    I hope that doesn't mean that the CPU of the Homey is already overloaded...
  • Fire69Fire69 Member
    jorden said:
    I hope that doesn't mean that the CPU of the Homey is already overloaded...

    I was worried about that too.
    If you look in the Stuff for geeks, my loadavg is near or over 1 lots of times, even while doing nothing special.
    Google tells me that's not really good.
  • spamenigmaspamenigma Member
    loadavg was around 1.1, I just did a reboot and after it restarted its at 2.2 gradually going down to 1.1 and now hovering around 1.3.

    Pings during reboot are:


    So certainly appears cpu related. 
  • jordenjorden Member
    Fire69 said:
    jorden said:
    I hope that doesn't mean that the CPU of the Homey is already overloaded...

    I was worried about that too.
    If you look in the Stuff for geeks, my loadavg is near or over 1 lots of times, even while doing nothing special.
    Google tells me that's not really good.
    1 doesn't have to be problematic, depending on how the CPU/cores etc. but I think it is a high load in this case.

    However, that doesn't have to be a problem either, as long as Homey still responds quickly in terms of 'feeling'
  • Fire69Fire69 Member
    jorden said:
    Fire69 said:
    jorden said:
    I hope that doesn't mean that the CPU of the Homey is already overloaded...

    I was worried about that too.
    If you look in the Stuff for geeks, my loadavg is near or over 1 lots of times, even while doing nothing special.
    Google tells me that's not really good.
    1 doesn't have to be problematic, depending on how the CPU/cores etc. but I think it is a high load in this case.

    However, that doesn't have to be a problem either, as long as Homey still responds quickly in terms of 'feeling'
    The question is how it will scale.  My Homey is just sitting there, not doing much yet.
    If I add my entire z-wave network, and start making flows for it, will it be able to keep up? I hope so.
  • MarcoFMarcoF Member
    edited February 2016
    The Homey SOC should be more powerful then Rpi SOC, but my Rpi response much faster then the figures posted from Homey.

    Ping my Rpi2 over VPN and this is the result:



    And i'm beating the heart of this thing with;
    MySQL, SQLite3, multiple PHPfpm and nginx workers, Wordpress, Logging Camera motion, running WebSolarLog and logging my heat pump and PV (save >30 records a minute), etc

    top figures:
load average: 0.50, 0.53, 0.51

  • Fire69Fire69 Member
    I don't think Homey's SoC is more powerful than the one in the Rpi2.  It was faster than the Rpi1 (they were going to use the Compute module at first), but then changed to a faster SoC. This was way before the Rpi2 was introduced.

    Still, it shouldn't have such a high load now. Unless they have a lot of debugging options enabled for the moment or something.
  • aloftaloft Member
    I have the same response times on my Homey, but a load average of 0.40 ( 5 minute average ).  Just a couple of devices  (mostly 433 Mhz and some HUE devices) and just testing with flows, so no real active use yet. 

    Keep in mind that many network implementations have a low priority for responding to ping requests. So it might not mean anything in terms of load.
  • MarcoFMarcoF Member
    edited February 2016
    OK maybe you have a point.

    Rpi2         has 1gb memory, the quadcore   runs at 900mhz  and its an cortex A7
    Hummingboard has 1gb memory, the singlecore runs at 1000mhz and its an cortex A9

    I think the can compete.

    This is what i found quickly;
    http://www.androidauthority.com/raspberry-pi-2-vs-odroid-c1-vs-hummingboard-vs-mips-creator-ci20-599418/

    and here's the spec sheet of the hummingboard C1000S-D512-FE-3 which is used in Homey;
    https://www.solid-run.com/download/pub/solidrun/SR-uSOM-mx6/Bochure_MicroSOM_06-11.pdf

  • MennoVanDerAatMennoVanDerAat Member



    system information after full reset with clearing all settings and nothing installed


  • aloftaloft Member
    Update on the load on my Homey: the previous 0.40 load average was without Microphone enabled. With microphone enabled, I get 0.70-0.80 . 
  • Fire69Fire69 Member
    MarcoF said:
    OK maybe you have a point.

    Rpi2         has 1gb memory, the quadcore   runs at 900mhz  and its an cortex A7
    Hummingboard has 1gb memory, the singlecore runs at 1000mhz and its an cortex A9

    I think the can compete.

    This is what i found quickly;
    http://www.androidauthority.com/raspberry-pi-2-vs-odroid-c1-vs-hummingboard-vs-mips-creator-ci20-599418/

    and here's the spec sheet of the hummingboard C1000S-D512-FE-3 which is used in Homey;
    https://www.solid-run.com/download/pub/solidrun/SR-uSOM-mx6/Bochure_MicroSOM_06-11.pdf

    Is it a single or dual core? The D512 is single core. 
    If that's the case, the rpi2 can run circles around it without even breaking a sweat! 
  • MarcoFMarcoF Member
    edited February 2016
    Rpi2 has 4 cores of the A7 type at 900mhz.
    HumB has 1 core of the A9 type at 1000mhz.

    So the HumB has a 12% higher clock speed and also the performance of the A9 is 15% better to the A7. 

    So every Core is slower and threads take longer. Multi core is only good if the software is using all the cores as efficient as possible. If multi core code its wrongly implemented, it will be against you.
  • kriebelkouskriebelkous Member
    Just throwing this out there. I am in no way an expert. 
    But could this be interference on the wifi module. Homey had 7 wireless trans\receivers. I would imagine they had to perform some tricks to make them all work. Causing a little lag but making it more stable?
  • MarcoFMarcoF Member
    There you could have a point!
    I also thought about it and it wouldn't surprise me if there's some sort of generic queue where are signals are wireless commands are queued and for every command the specific radio is turned on for xxxx ms.

    But now i'm righting this, what to do with incoming signals.... The radio then should aways be on, else you loos data.
  • kriebelkouskriebelkous Member
    @marcof Did we just agree on something?  :D
  • MarcoFMarcoF Member
    "I Love Forums" :heart: 
    :wink: 

    I'm positive I will receive a Homey, but i'm a bit skeptical about some choice and thing are being done. I hope things change and i'm a big worried about the future though.
  • AthomeyAthomey Member
    I don't think it can be interference. Homey is not transmitting constantly, it's listening most of the time. Listening cannot interfere with WIFI. So if this interference was the case, much more responsetimes would be below a few tens of ms's.
  • kriebelkouskriebelkous Member
    Athomey said:
    I don't think it can be interference. Homey is not transmitting constantly, it's listening most of the time. Listening cannot interfere with WIFI. So if this interference was the case, much more responsetimes would be below a few tens of ms's.
    It looks like there is a wave pattern in the ping response. Maybe Homey is throttling somehow? Again i am not an expert so do'nt shoot me down if i say anything stupid here. If i had to build something that has 2 (or more) parts that could interfere a mechanism that turns down the signal for part A and turns it up for part B in rotation seems like a way to achieve it. That would (in my logic) explain the heartbeat like ping responses. I was not talking about interference, but a mechanism that could prevent it. Maybe @emile can shed some light on this. I kind of want to know how wrong i am :smile:
  • JohanKusterJohanKuster Member
    Not always bad. Average 26ms during 100 pings


  • JohanKusterJohanKuster Member
    and the Raspberry Pi2 only running Node-Red:


  • Fire69Fire69 Member


    Just did a full reset.
    Did nothing special yet, no devices installed, just a few apps.

    Loadavg was about 0.6 at the moment of pinging.
  • AthomeyAthomey Member
    That RaspPi is on a cable, right? <1ms doesn't look like WIFI. Are you comparing apples with oranges? :smile: 
  • JohanKusterJohanKuster Member
    Athomey said:
    That RaspPi is on a cable, right? <1ms doesn't look like WIFI. Are you comparing apples with oranges? :smile: 
    I knew there must be a reason why I like cables soo much. Haha sorry, you are right.
  • MarcoFMarcoF Member
    My ping results of the Rpi2 are also cabled... 
  • tunftunf Member
    edited February 2016
    can you do a pathping instead of ping command and post it here?
    So pathping 1.1.1.1. This will collect more info and shows all systems in between.
    (hint, you can also use this to athom servers)
Sign In or Register to comment.