(08-15-2016, 12:11 PM)MarkHaysHarris777 Wrote: [ -> ] (08-15-2016, 09:10 AM)amc2012 Wrote: [ -> ]It's been a month since your last update. Any progress?
And if not, are we at the point where the PINE folks will start swapping out any bad boards? I don't mind waiting a little longer if there's an actual fix, but at this point it seems like you're going down the hardware route. Would be nice if PINE made good with its most ardent supporters.
amc2012, yes we had some significant progress last night. irc member Needles swapped out his switch for a new
TEG S82G 8 port switch; also, he is using cat6 cable. he is now getting 850mb/s without packet loss.
This is an issue that is definitely related to environment (even tolerances in the switch) but apparently is also related to auto negotiation (in the master slave concept who ends up master, effectively winning the negotiation session). It seems that GbE is too new for even the switches to have it all figured out yet. At any rate, Needles has a significant improvement by replacing the switch; which tells me that GbE switches are also morphing (getting better) and that this IS NOT a PineA64 hardware problem per se.
Still looking into this.
Thanks, Mark,
I'm still interested in feedback from
on this, as I exclusively use Cat6 cable and have now tried three different GbE switches with the same results - almost zero throughput when the adapter is set to GbE.
Not sure what you mean by GbE being "too new." Having been around for the better part of two decades, and having been RAPIDLY deployed even when it first emerged, GbE is actually one of the more mature ethernet technologies out there. Check out the
Ethernet Alliance Roadmap. I work in IT, and we've been relying on GbE for a long long time now. I think it's probably safe to say that the Cisco and Netgear routers I use at home have "figured it out."
I have *never* run into any kind of issue like this in my home or work environment. I have seen faulty ethernet ports, bad cables, bad switches, and bad ports in switches. I have effectively eliminated all these possibilities from my PINE environment through extensive testing. This is a pretty simple environment we're dealing with here, and any issues I've seen even occur when I'm not connected to anything but two machines with a switch between them (trying to access the PINE via a putty session).
I have heard that it's environment related, just based on the fact that waldo sent in his non-functioning PINE and it worked in longsleep's environment - I still am not sure if he sent in the same power supply he was using at home or not? To me, that would be the only reasonable "environment" factor here, since many of us have already gone through the requisite network-related troubleshooting procedures.
(08-15-2016, 02:20 PM)amc2012 Wrote: [ -> ] (08-15-2016, 12:11 PM)MarkHaysHarris777 Wrote: [ -> ] (08-15-2016, 09:10 AM)amc2012 Wrote: [ -> ]It's been a month since your last update. Any progress?
And if not, are we at the point where the PINE folks will start swapping out any bad boards? I don't mind waiting a little longer if there's an actual fix, but at this point it seems like you're going down the hardware route. Would be nice if PINE made good with its most ardent supporters.
amc2012, yes we had some significant progress last night. irc member Needles swapped out his switch for a new
TEG S82G 8 port switch; also, he is using cat6 cable. he is now getting 850mb/s without packet loss.
This is an issue that is definitely related to environment (even tolerances in the switch) but apparently is also related to auto negotiation (in the master slave concept who ends up master, effectively winning the negotiation session). It seems that GbE is too new for even the switches to have it all figured out yet. At any rate, Needles has a significant improvement by replacing the switch; which tells me that GbE switches are also morphing (getting better) and that this IS NOT a PineA64 hardware problem per se.
Still looking into this.
Thanks, Mark,
I'm still interested in feedback from on this, as I exclusively use Cat6 cable and have now tried three different GbE switches with the same results - almost zero throughput when the adapter is set to GbE.
Not sure what you mean by GbE being "too new." Having been around for the better part of two decades, and having been RAPIDLY deployed even when it first emerged, GbE is actually one of the more mature ethernet technologies out there. Check out the Ethernet Alliance Roadmap. I work in IT, and we've been relying on GbE for a long long time now. I think it's probably safe to say that the Cisco and Netgear routers I use at home have "figured it out."
I have *never* run into any kind of issue like this in my home or work environment. I have seen faulty ethernet ports, bad cables, bad switches, and bad ports in switches. I have effectively eliminated all these possibilities from my PINE environment through extensive testing. This is a pretty simple environment we're dealing with here, and any issues I've seen even occur when I'm not connected to anything but two machines with a switch between them (trying to access the PINE via a putty session).
I have heard that it's environment related, just based on the fact that waldo sent in his non-functioning PINE and it worked in longsleep's environment - I still am not sure if he sent in the same power supply he was using at home or not? To me, that would be the only reasonable "environment" factor here, since many of us have already gone through the requisite network-related troubleshooting procedures.
Is your switch managed? If so, try forcing gbe on your switch. I did it here, and it worked fine.
Are you referring to Ethernet running on Android and Remix OS, or talking about Linux????
(08-15-2016, 02:27 PM)yusijs Wrote: [ -> ] (08-15-2016, 02:20 PM)amc2012 Wrote: [ -> ] (08-15-2016, 12:11 PM)MarkHaysHarris777 Wrote: [ -> ] (08-15-2016, 09:10 AM)amc2012 Wrote: [ -> ]It's been a month since your last update. Any progress?
And if not, are we at the point where the PINE folks will start swapping out any bad boards? I don't mind waiting a little longer if there's an actual fix, but at this point it seems like you're going down the hardware route. Would be nice if PINE made good with its most ardent supporters.
amc2012, yes we had some significant progress last night. irc member Needles swapped out his switch for a new
TEG S82G 8 port switch; also, he is using cat6 cable. he is now getting 850mb/s without packet loss.
This is an issue that is definitely related to environment (even tolerances in the switch) but apparently is also related to auto negotiation (in the master slave concept who ends up master, effectively winning the negotiation session). It seems that GbE is too new for even the switches to have it all figured out yet. At any rate, Needles has a significant improvement by replacing the switch; which tells me that GbE switches are also morphing (getting better) and that this IS NOT a PineA64 hardware problem per se.
Still looking into this.
Thanks, Mark,
I'm still interested in feedback from on this, as I exclusively use Cat6 cable and have now tried three different GbE switches with the same results - almost zero throughput when the adapter is set to GbE.
Not sure what you mean by GbE being "too new." Having been around for the better part of two decades, and having been RAPIDLY deployed even when it first emerged, GbE is actually one of the more mature ethernet technologies out there. Check out the Ethernet Alliance Roadmap. I work in IT, and we've been relying on GbE for a long long time now. I think it's probably safe to say that the Cisco and Netgear routers I use at home have "figured it out."
I have *never* run into any kind of issue like this in my home or work environment. I have seen faulty ethernet ports, bad cables, bad switches, and bad ports in switches. I have effectively eliminated all these possibilities from my PINE environment through extensive testing. This is a pretty simple environment we're dealing with here, and any issues I've seen even occur when I'm not connected to anything but two machines with a switch between them (trying to access the PINE via a putty session).
I have heard that it's environment related, just based on the fact that waldo sent in his non-functioning PINE and it worked in longsleep's environment - I still am not sure if he sent in the same power supply he was using at home or not? To me, that would be the only reasonable "environment" factor here, since many of us have already gone through the requisite network-related troubleshooting procedures.
Is your switch managed? If so, try forcing gbe on your switch. I did it here, and it worked fine.
It is not managed. So you're saying you had this issue (where GbE was non functional) and you put it on a managed switch with a port hard-set to GbE and that eliminated the issue? Because THAT would be a solution I could live with if that's the case. Would be easy to just buy a managed switch.
Yes. It seems very much like an issue with auto neg. Basically I set it to force 1000m Full Duplex on that port. If I didnt set anything, I got 0.2-1 mbps at best. This was on ubuntu, pretty sure it's the same for android though.
(08-15-2016, 02:44 PM)yusijs Wrote: [ -> ]Yes. It seems very much like an issue with auto neg. Basically I set it to force 1000m Full Duplex on that port. If I didnt set anything, I got 0.2-1 mbps at best. This was on ubuntu, pretty sure it's the same for android though.
This is the most helpful/hopeful thing anyone has suggested so far. I now have a 5-port managed switch coming Saturday. Guess it makes sense it would be an auto-negotiate issue. If so, I still have hopes there is some kind of firmware fix that could happen for our boards. But this is a reasonable workaround if it works. Will report back this weekend.
Cheers.
Hopefully it works for you to! Mine is a Cisco SG-200 8 port, and works like a charm!
(08-15-2016, 02:20 PM)amc2012 Wrote: [ -> ]Not sure what you mean by GbE being "too new." Having been around for the better part of two decades, and having been RAPIDLY deployed even when it first emerged, GbE is actually one of the more mature ethernet technologies out there. Check out the Ethernet Alliance Roadmap. I work in IT, and we've been relying on GbE for a long long time now. I think it's probably safe to say that the Cisco and Netgear routers I use at home have "figured it out." I have *never* run into any kind of issue like this in my home or work environment.
I have never run a GbE network at home; most people have not. My entire network is 10/100 as are most people's home networks; why--? Because once a network is setup (mine years and years ago) why setup another one. All of my devices are 100 now. None of my devices are GbE.
it takes a full twenty-five years to fully adopt a new technology to the 'common place' status. GbE is not there yet. And, as you can see not only by the varying opinions on the subject but also by the varying status reports, GbE is not the standard some of you have been hoping it would be.
This issue seems to be a 'tolerance' timing environment specific problem that can be resolved with the right hardware and configuration. There have been some recent developments in that direction. I hope your situation may be corrected also. Obviously packets are going to be lost and through-put is going to be low if auto negotiation occurs every couple of seconds.
Best of luck to you.
(08-15-2016, 03:35 PM)MarkHaysHarris777 Wrote: [ -> ]I have never run a GbE network at home; most people have not. My entire network is 10/100 as are most people's home networks; why--? Because once a network is setup (mine years and years ago) why setup another one. All of my devices are 100 now. None of my devices are GbE.
it takes a full twenty-five years to fully adopt a new technology to the 'common place' status. GbE is not there yet. And, as you can see not only by the varying opinions on the subject but also by the varying status reports, GbE is not the standard some of you have been hoping it would be.
This issue seems to be a 'tolerance' timing environment specific problem that can be resolved with the right hardware and configuration. There have been some recent developments in that direction. I hope your situation may be corrected also. Obviously packets are going to be lost and through-put is going to be low if auto negotiation occurs every couple of seconds.
Best of luck to you.
I think with the arrival of faster wireless, you're probably right - people probably moved from their wired 10/100 to the wireless G instead of moving to GbE. It's probably one reason why there aren't more complaints about this. I didn't think I was a particularly early adopter of GbE at home, since switches got cheap fast, and not long after we saw the benefits at work, I wanted the same at home.
The twenty-five year figure probably depends on the technology. But even if your figure is true at face value, that doesn't mean the specifications aren't nailed down sooner. In this case, I would blame the chip vendor long before I blamed Cisco, D-Link, or Netgear, who made the switches I tried and who should well have known what the requirements were to allow traffic to flow reliably in a stable GbE environment. Nearly two decades in, there should be no "learning" necessary to provide GbE speeds and to negotiate that setting.
Respectfully, this isn't a situation of "well this is all still very new." It's an issue of some chip vendor not manufacturing to spec, and that seems to be the line that the PINE folks are pursuing at this point.
We're lucky there may be a workaround for this, but the finger pointing still goes to hardware I think. Whether that's the PINE design or chips not to spec, I dunno. But since other board makers are running into GbE issues with this chip, I think we know where the issue lies. Again, thanks to those here who have made efforts toward workarounds.
(08-15-2016, 03:01 PM)amc2012 Wrote: [ -> ] (08-15-2016, 02:44 PM)yusijs Wrote: [ -> ]Yes. It seems very much like an issue with auto neg. Basically I set it to force 1000m Full Duplex on that port. If I didnt set anything, I got 0.2-1 mbps at best. This was on ubuntu, pretty sure it's the same for android though.
This is the most helpful/hopeful thing anyone has suggested so far. I now have a 5-port managed switch coming Saturday. Guess it makes sense it would be an auto-negotiate issue. If so, I still have hopes there is some kind of firmware fix that could happen for our boards. But this is a reasonable workaround if it works. Will report back this weekend.
Cheers.
Would be very interesting to get those results. As you maybe know, i sent my board to longsleep and it worked fine in his environment, but i could not get it to stable GbE with all 4 of my routers and switches and some of them are pretty new (Fritzbox 7490 e.g.). So if you get it running with a new switch also, would be very interested in the type of switch also....
Good luck