GekkoScience NewPac / Terminus R606 (BM1387) Official Support Thread

[os2sam]

I'd cut it back to 750MHz. Should get the same hashrate but about 6% less heat. Sounds like your utilization is only about 93%.

It shows WU:94%  I set to 750 for the night.  The light were solid so no overtemp.  The case didn't feel hot at all and the air exhaust didn't seem too hot either.  But 800Mhz just made me a little nervous, so I took your advice.
Thanks,
Sam

[sidehack]

Batch 2 and onward I'm putting in some holes for a pin header on the I2C bus so I can use 'em for testing. An industrious end user might put a sniffer on there and watch the I2C traffic to get temperature readings in realtime. Polls are every five seconds.

[gcanelson]

I have a small situation... I'm running two Win10 machines (YES, that's most likely the problem...). One has cgminer  from last December. It has 10 instances of that cgminer running - only NewPacs thru GS Hubs on that computer and no issues. The other Win10 is running the cgminer from May 13 this year. One R606 and several NewPacs all thru GS Hubs. In general, it too is working as it should but 2 of the cgminers tend to stop working after one or two days with the following error message:

Code:

Assertion failed!

Program: C:\~CC-Ore\cgminer.exe
File: os/poll_windows.c, Line 145

Expression: fd != NULL

C:\~CC-Ore>pause
Press any key to continue . . .

Anyone else get this? Any suggestions?

Edit: The R606 instance is running steady. The two cgminers that drop out are NewPacs set @ 425MHz.

I'll likely do a complete rebuild using the May 26 version cgminer for both computers come the weekend... I don't think I've seen this problem posted here so I will post it now... Oh, and restarting the stalled instance(s) of cgminer does work, without any changes to the batch file(s)... So it's only annoying and inconvenient... the problem's on 2 of 25 instances - so mostly things are working well.

[vh]

hmm.   that is definitely a new / unreported one.   looks like part of libusb complaining.

use this to forcefully bypass the manual intervention part:

Code:

:loop_forever
cgminer.exe -o ...
goto loop_forever

small side effects:
   - you lose visibility to other errors.
   - you won't be able to press q to quit.

[philipma1957]

The Average is 1.012 Ths.  I'm using a PC Power supply w/the 6 pin.  It's still holding at 800Mhz.  It seems stable so I guess I'll let it run overnight.

at my 775 which tunes to 767.5 I am doing 1025-1035gh.

mining to solo.ckpool

so I would suggest dropping to  775 or 750 since you are getting less hash then I do

I add this fan

https://www.cdw.com/product/Thermaltake-Mobile-Fan-12-USB-fan/2258006?

plugged it into the usb port. It is a pull and aids sidehack's fan.

I do not know how long I can do 767.5  but I am going for it.

My only complaint is I have to enter freq 775 and it tunes to 767.5
I rather enter  freq 767.5, but I get  not  a real number.

Here I am 1 r606 and 3 newpacs

http://solo.ckpool.org/users/1JdC6Xg3ajT3rge3FgPNSYYFpmf53Vbtje

does about  3 x 62 = 186 for the newpacs at freq 275
does about 1 x 1035 = 1035 for the r606

total of 1221 for all gear.  they run at 98-100%  best share is 2.9b  needs to be 7.5t

[os2sam]

so I would suggest dropping to  775 or 750 since you are getting less hash then I do  

I add this fan

I reset it to 750 before going to be last night.  The hashrate averaged at 1014 and power consumption went down to 106 watts.  I'm going to test on the Win7 side of the same laptop over the weekend and see if hardware differences have anything to do with my low performance on other hardware.

For now it's back to power level 4 on the Pi doing around 740Ghs at 69 watts.  I find the low power consumption for that hash rate pretty cool.

[gcanelson]

hmm.   that is definitely a new / unreported one.   looks like part of libusb complaining.

use this to forcefully bypass the manual intervention part:

Code:

:loop_forever
cgminer.exe -o ...
goto loop_forever

I took the `pause' out of the cgminer statement(s) so they'd loop if necessary, and I have restarted the two instances that have been dropping out. Everything looks normal so far, I'll report back on this, say Monday or Tuesday.

small side effects:
   - you lose visibility to other errors.

I haven't really had other errors, they just stop - usually when I'm away from home for 14 hours - because Murphy says so!

  - you won't be able to press q to quit.

There's always Alt+F->X or the big 'X'.

Thanks again, vh. You rock.

[mstrozier]

Been looking at different fans to assit with the cooling as well.   I currently have a usb desk fan running behind the units since I have the two stacked and pulling the air assisting the fan Sidehack has installed.   It's been helpful but not something I want long term.   I believe the current fans are 80mm fans?  I have a set of usb powered 80mm fans coming off amazon.   https://www.amazon.com/AC-Infinity-MULTIFAN-Receiver-Playstation/dp/B00IJ2J2K0   is what I'm looking at.  And should be able to bolt up at the rear of the unit and setup a push/pull and plug into a usb port.   

Currently I have one running at 725Mhz pulling 960.6 Gh/s and I had to lower the 2nd one to 700Mhz as it would get missed nounces from chip 0 or 11.  So I dropped from 725 to 700 running in a 2nd instance and so far haven't seen any more errors.  And it's hashing at 924.8 Gh/s.   Both running on power setting 6.

[419mining]

Are there that many overheating units that everyone is adding extra fans or is this just a feel good add on?

[mstrozier]

Are there that many overheating units that everyone is adding extra fans or is this just a feel good add on?

For me it's just an extra security blanket so to speak.  Since they are rated at 500 or 550MHz out the door.  And I'm running one at 700 and the other at 725, never hurts to have additional cooling.

[philipma1957]

Are there that many overheating units that everyone is adding extra fans or is this just a feel good add on?

I have volts maxed out and I am doing a steady freq of 767.5 98-100% to get 1035 gh

I have a usb fan on hand cost = 0

The r606 has a usb port to run it so easy peasy why not cool it some more then it needs.

[ZimbaCardi]

So if they keep not working, and if you want (hopefully you're in the US), and if the seller is okay with it, you can send them back to me (the factory) and I will ship you replacements.

That is rocking customer support, thanks a million Sidehack.
I will keep trying to get them running over the weekend and if all else fails I'll send them over.

Meantime, any way that I can see the voltage so as to adjust the pot more accurately?

[sidehack]

Yeah, if you got a volt meter.

Regarding USB fans on R606 - that's pretty slick. I had only ever thought of using the USB power port for a controller but adding a second fan without needing second power is quite clever. I know I grumbled about a two-fan setup but it is easier to strap a second fan to the outside than it is to open the case and replace the internal fan. Reckon if the result is still the same I have no room to complain.

Unless someone wants a second push-pull fan for stock. Because that's dumb.

But increasing cooling capacity for overclocks, by whatever method, is prudent for long-term health of the machine.

[evade_57]

Added 6 more NewPacs and GS hub today, thanks MFB!

18 NewPacs and 3 GS hubs running with 6 Arctic USB 92mm fans.

Raspberry Pi 3+ with Cicero USB2 powered hub added.

Running at 250Mhz and heatsink temp is average 79F.

Asic Boost on using Ck's pool, hashing is within spec for all sticks.

1.02 T/hs for 112W at the wall plug.

[shrimplomein]

Total NOOB question here:

What are the factors that cause frequency to be adjusted down?

Setup: R606 running through Win10, powered by GekkoScience DPS1200 PSU.  My PC is quite powerful with an i7 3.4GHz quad-core CPU and 32 Gigs of RAM.
Settings: Voltage setting 5, and I set --gekko-r606-freq 643
Results: I'm getting about 840-850 Gh/s on average.

If I try to increase frequency to 750 or higher, it seems to auto-adjust down to something below 700 most of the time. I've tried various voltage combinations.

I'm trying to see if I can get it to 1 Th/s average like some others have. Would using a Linux controller make a difference? Is the USB interface on my PC the weak link?

I'm new to mining, and still trying to learn the basics of the electronics.

[sidehack]

Take a look at the charts in the first post of the product announcement thread, specifically the one labeled "estimated peak hashrate".

[shrimplomein]

Thanks. I guess mine is pretty much within the expected range. I'll keep experimenting to see how far I can push it.

[sidehack]

If it's hitting an upperbound regardless of voltage, it's probably a USB throughput issue. User comments have indicated that's common on Windows 10. Try the alternate ASICBoost option described somewhere in the last few pages to see if that fixes it.

[shrimplomein]

Thanks for the tip. I'm running with the gekko-lowboost switch now and I do believe it's running more efficiently, and it has not plateaued after an hour of running, whereas previously it would auto-retarget a lower frequency by this point.  I'll re-run several tests with this setting and see where it goes.

[mstrozier]

Well today my fans arrived and now have them installed on the back of the R606  Extremely quiet and has about the same air flow as the Artic fans that come with the kit.    These are the fans if anyone is interested.  https://www.amazon.com/AC-Infinity-MULTIFAN-Receiver-Playstation/dp/B00IJ2J2K0

And below are the pictures after I've installed them.

Pic 1: https://ibb.co/QnXfmdN

Pic 2: https://ibb.co/ZJnVZm5

Pic 3: https://ibb.co/rswG4yf

They are USB powered.  The come prewired with 2 USB cables, one to connect to a power source which I have connected to one of the R606's and the 2nd USB connector is for daisy chaining additional fans (provided power needs are met).  They have a speed controller with typical Off, Low, Med and High.   I left them running on High settings and extremely quiet. Can't even hear them.

All in all, nice little setup and should keep them pretty cool while overclocked.