BITMAIN AntRouter R1: 1st wireless networking device with bitcoin mining chip

[RichBC]

I have connected up a fan and have been pushing up the frequency Today. 



Have now got up to 225MHz giving 12.375GH/s using a setting of "2:225:0882"

This as high as you can go with the Core voltage at it's standard setting of 0.61V. Actually this is better than I expected as the data sheet calls for about 0.67V. At this point I am seeing a few too many HW errors and at 250MHz there are a lot of errors. I will fit a pot Tomorrow, adjust the voltage up a little and push on up.

This is BTW with an external 5V PSU as I suspect the built in one would have run out of steam by now?

Rich

[valkir]

Wow some nice test done there Rich! Thanks man 

[notlist3d]

I have connected up a fan and have been pushing up the frequency Today. 



Have now got up to 225MHz giving 12.375GH/s using a setting of "2:225:0882"

This as high as you can go with the Core voltage at it's standard setting of 0.61V. Actually this is better than I expected as the data sheet calls for about 0.67V. At this point I am seeing a few too many HW errors and at 250MHz there are a lot of errors. I will fit a pot Tomorrow, adjust the voltage up a little and push on up.

This is BTW with an external 5V PSU as I suspect the built in one would have run out of steam by now?

Rich

Wow that is pretty impressive about 2.5X the normal speed.   What are temps you are getting running this.

I'm guessing it got a little toasty wen running that speed .  Thanks for sharing your push on OC though!

[klondike_bar]

I have connected up a fan and have been pushing up the frequency Today. 



Have now got up to 225MHz giving 12.375GH/s using a setting of "2:225:0882"

This as high as you can go with the Core voltage at it's standard setting of 0.61V. Actually this is better than I expected as the data sheet calls for about 0.67V. At this point I am seeing a few too many HW errors and at 250MHz there are a lot of errors. I will fit a pot Tomorrow, adjust the voltage up a little and push on up.

This is BTW with an external 5V PSU as I suspect the built in one would have run out of steam by now?

Rich

damn, thats impressive. I stopped at ~8.5GH because the unit began making a bit of buzzing noise (AC/DC?) and I did not know the amperage available to the chip. Impressive to see it can actually keep cranking up.

with a few stick-on heatsinks to the R1 case and a nearby fan, 8.5GH ran quite cool

[notlist3d]

You can get zombies.  I had it running about a day and got 1 zombie out of 3 U3's. 

Still a nice cheap controller though.  The thing with is it turns your U3's into a "bitmain" feel on setting vs command line I was using.  I log into IP like a bigger miner instead of using team viewer.

[kodokbuduk]

hi guys, anyone can help me with this..i got two antRouter but one is showing slow speed on antpool..
i checked on the device setting, the speed is 5gh/s but on antpool showing 1gh/s average..

[RichBC]

Antpool is quite slow to update, also depends on which screen you are looking at?. Go to the worker tab and look at the 10min box up the top, if after 15 mins you have refreshed and it's still 1GH and at your end it's showing 5GH then I am not sure what the problem is, particularly if you have another one that is ok?

Rich

[kodokbuduk]

yea another one works fine, i wish i can attach picture here. How to attach picture here?
I've been waiting this a day but still showing 1gh/s average, i think something wrong with the setting?
i tried to delete the worker on antpool, refresh.. and press default button on antRouter but still getting 1Gh/s showing on antpool..

[RichBC]

Wow that is pretty impressive about 2.5X the normal speed.   What are temps you are getting running this.

I'm guessing it got a little toasty wen running that speed .  Thanks for sharing your push on OC though!

It's all a bit of a lash up at the moment, but I have hooked up a fan controller with a thermistor on the heatsink and have set it to keep the temperature at 35 DegC which it is doing quite nicely, with still some fan speed in hand.



I need to tidy things up a bit now, everything is very insecure at the moment and could do with a bigger heatsink. Fit a pot so that I can adjust the core voltage up a bit, then onward & upwards.   Not sure if I will hit the Buck Converter, Cooling or chip limit first?

Thing is I am cheating by feeding it with an external 5V supply, however that is only a 5V 3A Wallwart. At 225MHz, 12.375GH/s it's taking 6.8W at the wall which is 0.549J/GH which is nothing special but do not know what the efficiency of the PSU is, and of course with just a single chip the controller current is a higher proportion than if this was multichip.

I will try it at some point with it's own supply, but don't like working with 240V exposed on the bench.


Rich

[RichBC]

I have tidied things up and mounted the parts on a piece of wood, fitted a pot to allow the Core Voltage to be adjusted and we are ready to go.



I have been taking the Frequency up 25MHz at a time and adjusting the Core Voltage for a low HW error rate. Here are the results so far.

Code:

Setting	       MHz	V Core	    W	  GH    J/GH
														
1:300:0b82 	300	0.670	  10.2	16.50   0.62
2:275:0a82 	275	0.650	   9.0	15.13   0.60
2:250:0982  	250	0.635	   7.9	13.75   0.57
2:225:0882 	225	0.610	   6.8	12.38   0.55
5:150:0b83 	150	0.610	   5.0	8.25    0.61
7:100:0783 	100	0.610	   3.9	5.50    0.71

An interesting observation is that the Core Voltage needed for this single chip to give a low HW error rate is quite a bit lower than I have found with the S5 and the data sheet?

I have however had it just stop hashing for at 300MHz after 30 Minutes for no reason, started again with no problem? SoI am going to run for a while as 300MHz to check that things are stable before going any higher.  

Rich

[notlist3d]

Anyone happen to notice it reset like three hours?  Mine I have tried a few things to but seems like every 3 hours it resets, but starts back in like 30 second.

I'm thinking maybe try less U3's next I have 3 right now.  So thinking about trying 1 today.  Just wondered if anyone else has seen this.

[RichBC]

Anyone happen to notice it reset like three hours?  Mine I have tried a few things to but seems like every 3 hours it resets, but starts back in like 30 second.

I'm thinking maybe try less U3's next I have 3 right now.  So thinking about trying 1 today.  Just wondered if anyone else has seen this.

Have not seen that, but have not run with U3's. Does it do it without a U3 attached?

Mine stopped again after 25 Mins at 300MHz, so have upped the Core Voltage to 0.68V which gives 0 HW errors, has now run for 40 Mins, we will see?


Rich

[notlist3d]

Anyone happen to notice it reset like three hours?  Mine I have tried a few things to but seems like every 3 hours it resets, but starts back in like 30 second.

I'm thinking maybe try less U3's next I have 3 right now.  So thinking about trying 1 today.  Just wondered if anyone else has seen this.

Have not seen that, but have not run with U3's. Does it do it without a U3 attached?

Mine stopped again after 25 Mins at 300MHz, so have upped the Core Voltage to 0.68V which gives 0 HW errors, has now run for 40 Mins, we will see?


Rich

I will test without and with just one.  So far have been trying to use 3 as it does it very well.  It seems to work pretty good.

Just odd on the 3 hour reset.  So going to do some more testing of different variables.

[aarons6]

I have tidied things up and mounted the parts on a piece of wood, fitted a pot to allow the Core Voltage to be adjusted and we are ready to go.



I have been taking the Frequency up 25MHz at a time and adjusting the Core Voltage for a low HW error rate. Here are the results so far.

Code:

Setting	       MHz	V Core	    W	  GH    J/GH
														
1:300:0b82 	300	0.670	  10.2	16.50   0.62
2:275:0a82 	275	0.650	   9.0	15.13   0.60
2:250:0982  	250	0.635	   7.9	13.75   0.57
2:225:0882 	225	0.610	   6.8	12.38   0.55
5:150:0b83 	150	0.610	   5.0	8.25    0.61
7:100:0783 	100	0.610	   3.9	5.50    0.71

An interesting observation is that the Core Voltage needed for this single chip to give a low HW error rate is quite a bit lower than I have found with the S5 and the data sheet?

I have however had it just stop hashing for at 300MHz after 30 Minutes for no reason, started again with no problem? SoI am going to run for a while as 300MHz to check that things are stable before going any higher.  

Rich

rich, do these values look good? im adding a freq dropdown and i need the numbers.

freq.default = "7:100:0783"
freq:value("5:150:0b83", translate("150M * Warning"))
freq:value("6:143.75:1687", translate("143.75M * Warning"))
freq:value("6:137.5:0a83", translate("137.5M * Warning"))
freq:value("6:131.25:0a03", translate("131.25M * Warning"))
freq:value("6:125:0983", translate("125M * Warning"))
freq:value("7:118.75:0903", translate("118M * Warning"))
freq:value("7:112.5:0883", translate("112.5M * Warning"))
freq:value("7:106.25:0803", translate("106.25M * Warning"))
freq:value("7:100:0783", translate("100M * Default"))


this what i have so far, and here is what it looks like, the custom lets you input any value you want, not sure id go above 150 tho? prob wont even go above 125 myself.

[RichBC]

rich, do these values look good? im adding a freq dropdown and i need the numbers.

freq.default = "7:100:0783"
freq:value("5:150:0b83", translate("150M * Warning"))
freq:value("6:143.75:1687", translate("143.75M * Warning"))
freq:value("6:137.5:0a83", translate("137.5M * Warning"))
freq:value("6:131.25:0a03", translate("131.25M * Warning"))
freq:value("6:125:0983", translate("125M * Warning"))
freq:value("7:118.75:0903", translate("118M * Warning"))
freq:value("7:112.5:0883", translate("112.5M * Warning"))
freq:value("7:106.25:0803", translate("106.25M * Warning"))
freq:value("7:100:0783", translate("100M * Default"))


this what i have so far, and here is what it looks like, the custom lets you input any value you want, not sure id go above 150 tho? prob wont even go above 125 myself.

These are not fully tested but are what I have found works. However note that without upping the Core Voltage you cannot go above 225MHz because of HW errors.

pb:value("7:100:0783", translate("100M (default)"))
pb:value("7:206.5:0803", translate("106.25M"))
pb:value("6.5:112.5:0883", translate("112.5M"))
pb:value("6.5:118.75:0903", translate("118.75M"))
pb:value("6:125:0983", translate("125M"))
pb:value("6:131.25:0a03", translate("131.25M"))
pb:value("5:150:0b83", translate("150M"))
pb:value("4:175:0d83", translate("175M"))
pb:value("3:200:0782", translate("200M"))
pb:value("2:225:0882", translate("225M"))
pb:value("2:250:0982", translate("250M"))
pb:value("2:275:0a82", translate("275M"))
pb:value("1:300:0b82", translate("300M"))
pb:value("1:325:0c82", translate("325M"))
pb:value("1:350:0d82", translate("350M"))
pb:value("1:375:0e82", translate("375M"))
pb:value("1:400:0f82", translate("400M"))

Rich

[aarons6]

rich, do these values look good? im adding a freq dropdown and i need the numbers.

freq.default = "7:100:0783"
freq:value("5:150:0b83", translate("150M * Warning"))
freq:value("6:143.75:1687", translate("143.75M * Warning"))
freq:value("6:137.5:0a83", translate("137.5M * Warning"))
freq:value("6:131.25:0a03", translate("131.25M * Warning"))
freq:value("6:125:0983", translate("125M * Warning"))
freq:value("7:118.75:0903", translate("118M * Warning"))
freq:value("7:112.5:0883", translate("112.5M * Warning"))
freq:value("7:106.25:0803", translate("106.25M * Warning"))
freq:value("7:100:0783", translate("100M * Default"))


this what i have so far, and here is what it looks like, the custom lets you input any value you want, not sure id go above 150 tho? prob wont even go above 125 myself.

These are not fully tested but are what I have found works. However note that without upping the Core Voltage you cannot go above 225MHz because of HW errors.

pb:value("7:100:0783", translate("100M (default)"))
pb:value("7:206.5:0803", translate("106.25M"))
pb:value("6.5:112.5:0883", translate("112.5M"))
pb:value("6.5:118.75:0903", translate("118.75M"))
pb:value("6:125:0983", translate("125M"))
pb:value("6:131.25:0a03", translate("131.25M"))
pb:value("5:150:0b83", translate("150M"))
pb:value("4:175:0d83", translate("175M"))
pb:value("3:200:0782", translate("200M"))
pb:value("2:225:0882", translate("225M"))
pb:value("2:250:0982", translate("250M"))
pb:value("2:275:0a82", translate("275M"))
pb:value("1:300:0b82", translate("300M"))
pb:value("1:325:0c82", translate("325M"))
pb:value("1:350:0d82", translate("350M"))
pb:value("1:375:0e82", translate("375M"))
pb:value("1:400:0f82", translate("400M"))

Rich

hmm i think id have to take it apart like you did to OC this thing.. it gets quite warm at 112.5M i wouldnt imagine running it at 225.

you think if i glue on a large heatsink to the top side it will help? or you think its the pad?

guess just going to leave it at default for now.

[RichBC]

I did not try overclocking until I had it open on the bench and could measure what was going on. This post has the best info on how far you can go with the case on.

actually, seemed to be related to other parts of the frequency instruction.
8:125:0983 = ~4.5GH (8min sample)
6:125:0983 = ~6.4GH (30min sample)

connect to antpool: its your login email. mining takes up to an hour to show up (which is a bit annoying)

I look forwards to a modified firmware with actual asic stats/details like on the U3 page, since this minimalistic bullshit isnt particularly useful

6:125:0983    = ~6.35GH (9.5hr sample)
6:150:0B83    = ~6.81GH (2hr sample)
5.5:150:0B83 = ~7.02GH (30min sample)
5:143:1687    = ~7.55GH (1.5hr sample)
5:150:0B83 = ~7.62GH (20min sample)  
4.5:143:1687    = ~8.00GH (30min sample)
4:143:1687    = ~7.92GH ( 2.5hr sample)
4.2:156:0c03 = ~8.54GH (1.25hr sample)  55.3% overclock  

with a heatsink and fan its at a comfortable temperature, but the buzzing noise that starts at >7.5GH is present constantly (sounds slightly like a noise wall wart) so while the chip could go a lot further I don't know if the AC/DC and DC/DC can

anyone able to explain the function of the first value (ie: 6 vs 7 vs ? I assume timing

There are several factors limiting how high you can go. The buzzing is probably the transformer in the Mains to 5V converter . I will make some separate measurements on it to confirm. The case is quite a good heatsink and can be added to with more heatsinks & a fan. The Core voltage can be upped with a simple mod but the PSU runs out of steam first.

In practice not much point in doing any of this, except for the fun of seeing how high it can be pushed. However some overclocking with no mods should be quite safe and a then a bit more with some additional heatsinking or a fan.

Rich

[notlist3d]

Anyone happen to notice it reset like three hours?  Mine I have tried a few things to but seems like every 3 hours it resets, but starts back in like 30 second.

I'm thinking maybe try less U3's next I have 3 right now.  So thinking about trying 1 today.  Just wondered if anyone else has seen this.

Have not seen that, but have not run with U3's. Does it do it without a U3 attached?

Mine stopped again after 25 Mins at 300MHz, so have upped the Core Voltage to 0.68V which gives 0 HW errors, has now run for 40 Mins, we will see?


Rich

I will test without and with just one.  So far have been trying to use 3 as it does it very well.  It seems to work pretty good.

Just odd on the 3 hour reset.  So going to do some more testing of different variables.

I was able to confirm 3 hour reset stop's once down to one U3.  With 3 U3's mine reset's every 3 hour's. Tried a lot of different hubs, and before finding this out.

If someone else could please try 3 if they have it and see what happens.

[CrazyGuy]

Anyone happen to notice it reset like three hours?  Mine I have tried a few things to but seems like every 3 hours it resets, but starts back in like 30 second.

I'm thinking maybe try less U3's next I have 3 right now.  So thinking about trying 1 today.  Just wondered if anyone else has seen this.

Have not seen that, but have not run with U3's. Does it do it without a U3 attached?

Mine stopped again after 25 Mins at 300MHz, so have upped the Core Voltage to 0.68V which gives 0 HW errors, has now run for 40 Mins, we will see?


Rich

I will test without and with just one.  So far have been trying to use 3 as it does it very well.  It seems to work pretty good.

Just odd on the 3 hour reset.  So going to do some more testing of different variables.

I was able to confirm 3 hour reset stop's once down to one U3.  With 3 U3's mine reset's every 3 hour's. Tried a lot of different hubs, and before finding this out.

If someone else could please try 3 if they have it and see what happens.

There's a cron job setup to run a binary called cgminer-monitor every 3 minutes.

Code:

crontab -e

*/3 *  *   *   *     /usr/bin/cgminer-monitor

I'm not certain, but my guess is that the monitor application probably restarts cgminer if it detects a zombie. Perhaps it also reboots the R1 if cgminer restart doesn't work. I'm thinking that as you plug in more U3s you have a better chance of hitting a zombie..

[notlist3d]

Anyone happen to notice it reset like three hours?  Mine I have tried a few things to but seems like every 3 hours it resets, but starts back in like 30 second.

I'm thinking maybe try less U3's next I have 3 right now.  So thinking about trying 1 today.  Just wondered if anyone else has seen this.

Have not seen that, but have not run with U3's. Does it do it without a U3 attached?

Mine stopped again after 25 Mins at 300MHz, so have upped the Core Voltage to 0.68V which gives 0 HW errors, has now run for 40 Mins, we will see?


Rich

I will test without and with just one.  So far have been trying to use 3 as it does it very well.  It seems to work pretty good.

Just odd on the 3 hour reset.  So going to do some more testing of different variables.

I was able to confirm 3 hour reset stop's once down to one U3.  With 3 U3's mine reset's every 3 hour's. Tried a lot of different hubs, and before finding this out.

If someone else could please try 3 if they have it and see what happens.

There's a cron job setup to run a binary called cgminer-monitor every 3 minutes.

Code:

crontab -e

*/3 *  *   *   *     /usr/bin/cgminer-monitor

I'm not certain, but my guess is that the monitor application probably restarts cgminer if it detects a zombie. Perhaps it also reboots the R1 if cgminer restart doesn't work. I'm thinking that as you plug in more U3s you have a better chance of hitting a zombie..

You are correct about better chance but it actually did pretty good I think I had 2 zombies in 24 hours.  (How long zombie not sure as current pool does not have graph).  But with 3 it does it every time around 3 hours.

Thanks for finding that I was wondering what might be monitoring it seems it was behind  the scenes.    I have now went to 1 U3 and no problems with reset.