GekkoScience Compac BM1384 Stickminer Official Support Thread

[drew_ii]

Could be a code error and the PLL runs 205 instead of 410, which would put you at a hair over 11GH. Dunno.

I have some text grabs for you.  Same stick, same duration ~5 min runtime.
Rums approx same under Win 7 64 or RPI
CGM doesn't seem to like COMPAC @ 410


400MHz @ ~5 min = 21.78G at 1m average (allowing previous 4 mins for ramp)
         400 * 0.055 = 22.00G

Code:

 cgminer version 4.9.2 - Started: [2016-01-04 17:10:33.438]
--------------------------------------------------------------------------------
 (5s):18.02G (1m):21.78G (5m):13.76G (15m):6.112G (avg):21.47Gh/s
 A:512  R:0  HW:0  WU:300.0/m
 Connected to stratum.mining.eligius.st diff 512 with stratum as user 1MTeg4gPun
 Block: 508fd330...  Diff:104G  Started: [17:13:33.545]  Best share: 2.33K
--------------------------------------------------------------------------------
 [U ]SB management [P ]ool management [S ]ettings [D ]isplay options [Q ]uit
 0: COMPAC 10000762: 400MHz                  | 20.17G / 21.47Gh/s WU:300.0/m
--------------------------------------------------------------------------------
>time
The current time is: 17:15:48.39

410MHz @ ~5 min = 11.51G at 1m average (allowing previous 4 mins for ramp)
         410 * 0.055 = 22.55G

Code:

 cgminer version 4.9.2 - Started: [2016-01-04 17:16:18.012]
--------------------------------------------------------------------------------
 (5s):11.84G (1m):11.51G (5m):7.083G (15m):3.151G (avg):10.76Gh/s
 A:0  R:0  HW:0  WU:150.8/m
 Connected to stratum.mining.eligius.st diff 512 with stratum as user 1MTeg4gPun
 Block: 508fd330...  Diff:104G  Started: [17:16:18.119]  Best share: 275
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: COMPAC 10000762: 410MHz                  | 13.44G / 10.79Gh/s WU:150.8/m
--------------------------------------------------------------------------------
>time
The current time is: 17:22:06.93

415MHz @ ~5 min = 23.01G at 1m average (allowing previous 4 mins for ramp)
         415 * 0.055 = 22.82G

Code:

 cgminer version 4.9.2 - Started: [2016-01-04 17:28:09.263]
--------------------------------------------------------------------------------
 (5s):22.20G (1m):23.01G (5m):14.90G (15m):6.735G (avg):22.71Gh/s
 A:1536  R:0  HW:0  WU:317.9/m
 Connected to stratum.mining.eligius.st diff 512 with stratum as user 1MTeg4gPun
 Block: 305653a9...  Diff:104G  Started: [17:28:09.370]  Best share: 3.01K
--------------------------------------------------------------------------------
 [U]SB management [P]ool management [S]ettings [D]isplay options [Q]uit
 0: COMPAC 10000762: 415MHz                  | 24.25G / 22.75Gh/s WU:317.9/m
--------------------------------------------------------------------------------
>time
The current time is: 17:29:48.39

[sidehack]

My guess is the 410 is actually setting the chips to 205, which would give you an expected about 11.3GH. Likely we just messed up the hex value - though I think the table is copied straight off the S5.

[Deezil]

Honestly get a RPI and you will be so thankful.  You don't need a ton of power off a machine and could save your laptop some stress of really hot.. that does not sound good.

But you could even use minera on RPI.  For a little investment your power cost vs a laptop will be decent.  And RPI's are pretty cheap.  I highly suggest getting one vs your laptop.

And very cool to see the pod in the works.  Thanks for keeping us in the loop I love watching your dev process.

I have an RPi B+ and I got the minera RPi image from the support page here and had it all setup, but it wouldn't work with my 3.0 hubs.  Before I could get my hands on a 2.0 hub to act as the middle man, my son got a gopro for christmas and I sacrificed my only microsd from the pi so he could use the camera.  I just ordered a new bigger microsd and will set it all up again when it comes in.  I already have a subdomain forwarded to the lightppd server on my laptop so I can just browse to the minera page from anywhere, I'm going to set the RPi up the same way.

I don't think the laptop is really stressed, All it does is run Ubuntu 14.04 LTS and minera. Not nearly as much as when it had windows 7.  It's got an Core i7-2640M / Radeon HD 6630M and 8 GB of Ram.  But all that is crammed into a really flat case.  I don't know how they got a fan in there but it's got a really powerful/loud when it gets real hot.  It sits on its dock all the time and that has some warmth of it's own as well that contributes to the overall temp.  The area near my desk stays relatively warm with the laptop, two desktops, and the 12 miners all going pretty much 24/7 

[aarons6]

ive never tried minera, if i install it on ubuntu 14.04 can i connect to it over web like cryptoglance?

edit never mind.. i installed it..

for anyone else installing to ubuntu 14.04 it tells you to put it in /var/www/minera but you need to put it in /var/www/html/minera


it seems to be working tho, looks good.

[Biodom]

I understand that cgminer works for most, but it is in reverse for me:

bfgminer works beautifully, on all sticks, both on pi zero and pi2 revB, but cgminer-gekko does not work on pi and on linux PC.
maybe once they worked on bfgminer, then sticks have some setting change, but I doubt it. It remains a mystery.

I have cgminer-gekko compiled and built (kudos to aarons6) .
watch for cgminer-gekko-rev2 instead of cgminer-gekko directory that is being built after you untar cgminer-gekko.tar.gz file.
Yes, you can see this (that it is going to cgminer-gekko-rev2) if you keep staring at the screen as it is being done, but I didn't (initially).
once cgminer was done and I can see that it could connect to stratum.mining.eligius.st (and see blocks), the following command was shown:

Gekko detect (1:13) failed to initialize (incorrect device?)

Back to bfgminer and everything works again.
A bummer.

Have to think and try again.

[Deezil]

ive never tried minera, if i install it on ubuntu 14.04 can i connect to it over web like cryptoglance?

Yeah that's what I did.  It installs as it's own user.  I can login from anywhere now.  I just forwarded one of my domain names to my gateway, and from there to the http port on my ubuntu "server".  It has control for network miners as well.  Was a bit tricky to get it all setup adding the custom miner and all, but there's plenty of documentation out there.  Make sure you change the defualt minera admin login and minera site login credentials if you plan to expose it to the web.

[drew_ii]

I understand that cgminer works for most, but it is in reverse for me:

bfgminer works beautifully, on all sticks, both on pi zero and pi2 revB, but cgminer-gekko does not work on pi and on linux PC.
maybe once they worked on bfgminer, then sticks have some setting change, but I doubt it. It remains a mystery.

I have cgminer-gekko compiled and built (kudos to aarons6) .
watch for cgminer-gekko-rev2 instead of cgminer-gekko directory that is being built after you untar cgminer-gekko.tar.gz file.
Yes, you can see this (that it is going to cgminer-gekko-rev2) if you keep staring at the screen as it is being done, but I didn't (initially).
once cgminer was done and I can see that it could connect to stratum.mining.eligius.st (and see blocks), the following command was shown:

Gekko detect (1:13) failed to initialize (incorrect device?)

Back to bfgminer and everything works again.
A bummer.

Have to think and try again.

You have to run in sudo from the root folder containing the build.

Code:

sudo ./cgminer -o stratum+tcp://stratum.mining.eligius.st:3334 -u <username> --compac-freq <frq>

It has something to do with permissions to access the USB devices somehow, I just stumbled across it late last night when i was heading to bed so didn't read too much into it.  I just know that if I launch a terminal on RPI and sudo it, it runs beautifully.  If I don't sudo, or su, it gives me the same error.

It might be that Pi has the generic user that logs into desktop at boot, and pi (which has su access) and root... and that the /etc or /dev compac's aren't set with enough x permissions for non-root groups, but now we're getting into file and group and user permissions in *nix, which is 95% alien to me.

[Biodom]

thanks to everybody who helped this stick newb with cgminer.
finally cgminer-gekko is working and I can manipulate the stick at various settings.

quick questions:

1. do you have to do sudo reboot everytime you change something in hardware setup? it looks like this to me.
2. 800mv/433 mhz somehow resulted in stick not initializing, but it initialized OK with wiper at 3pm (750-760mv?) at 350mhz.

Q: which course of action is generally better: keep mv same (750) and see up to which mhz it will initialize OR increase mv to max (800) and then see which mhz will work?

It looks like even at 750/350 stick fluctuates wildly between 16 and 24Gh/s with average ~19 Gh/s (WU 265).
In my crazy setup watts and amperage from hub are not a problem.

[sidehack]

Bitmain's BM1384 chart notes 325MHz at 750mV, so running 350MHz is pretty decent. And 350MHz should average you 19.25GH. Which by my estimates would be about WU 268. So it sounds like you're pretty much right on target. Most people can't get sticks too far past 400MHz at 800mV. Recall the S5 "defaults" to 800mV (from a 12V source, assuming balanced nodes) and its stock clock is 350MHz.

Whether you set voltage and increase clock or set clock and decrease voltage really depends on what you want to accomplish. If you want the most hashrate for a given power output, or just the absolute most hashrate, will have different strategies.

[Biodom]

Bitmain's BM1384 chart notes 325MHz at 750mV, so running 350MHz is pretty decent. And 350MHz should average you 19.25GH. Which by my estimates would be about WU 268. So it sounds like you're pretty much right on target. Most people can't get sticks too far past 400MHz at 800mV. Recall the S5 "defaults" to 800mV (from a 12V source, assuming balanced nodes) and its stock clock is 350MHz.

Whether you set voltage and increase clock or set clock and decrease voltage really depends on what you want to accomplish. If you want the most hashrate for a given power output, or just the absolute most hashrate, will have different strategies.

I want to beat the Gekkoscience stick hashrate (Gh/s on 24hr average) world record, everything else is secondary...
Max does not count  ...cheers

[Deezil]

Bitmain's BM1384 chart notes 325MHz at 750mV, so running 350MHz is pretty decent. And 350MHz should average you 19.25GH. Which by my estimates would be about WU 268. So it sounds like you're pretty much right on target. Most people can't get sticks too far past 400MHz at 800mV. Recall the S5 "defaults" to 800mV (from a 12V source, assuming balanced nodes) and its stock clock is 350MHz.

Whether you set voltage and increase clock or set clock and decrease voltage really depends on what you want to accomplish. If you want the most hashrate for a given power output, or just the absolute most hashrate, will have different strategies.

I want to beat the Gekkoscience stick hashrate (Gh/s on 24hr average) world record, everything else is secondary...
Max does not count  ...cheers

Do I smell a challenge??     I can break out my setup, add my RPi and just run two sticks on a hub, I bet I can get 437 (or maybe 450MHz muahahaha).  What's the current record?

[Jake36]

Bitmain's BM1384 chart notes 325MHz at 750mV, so running 350MHz is pretty decent. And 350MHz should average you 19.25GH. Which by my estimates would be about WU 268. So it sounds like you're pretty much right on target. Most people can't get sticks too far past 400MHz at 800mV. Recall the S5 "defaults" to 800mV (from a 12V source, assuming balanced nodes) and its stock clock is 350MHz.

Whether you set voltage and increase clock or set clock and decrease voltage really depends on what you want to accomplish. If you want the most hashrate for a given power output, or just the absolute most hashrate, will have different strategies.

I want to beat the Gekkoscience stick hashrate (Gh/s on 24hr average) world record, everything else is secondary...
Max does not count  ...cheers

Do I smell a challenge??     I can break out my setup, add my RPi and just run two sticks on a hub, I bet I can get 437 (or maybe 450MHz muahahaha).  What's the current record?

Don't think it's the record, but I did 435MHz in the compac solo pool club's 24 hour stock stick over-clock contest that won (I actually let it run for 60 hours before I shut it down).

As for a challenge, I see where there may be a another contest coming up...

[Mikestang]

I think the record is held by sidehack with a one-off stick he beefed up, or does that not count?

The up-coming "no limit" oc stick contest should prove interesting.

[RichBC]

Don't think it's the record, but I did 435MHz in the compac solo pool club's 24 hour stock stick over-clock contest that won (I actually let it run for 60 hours before I shut it down).

As for a challenge, I see where there may be a another contest coming up...

I would be up for a contest.... Mine is plugged via a Hub into an Antrouter R1. Best reliable speed I have is 375MHz, at 400MHz ir runs but does not give the correct increase in hash rate so some work needed.


Rich

[Searing]

Well hopefully the below means we can get some 'new' type chips?

BitFury to Mass Produce New 16nm Chip and Sell to Public 1/5/16

https://bitcoinmagazine.com/articles/bitfury-to-mass-produce-new-nm-chip-and-sell-to-public-1452010171

That would 'perk up' the potential of the 'toys' on here NO DOUBT!

Let it be so....anyway Sidehack should have some 'bona fides' with past usb stick product and such to be maybe first on the list?

crosses fingers

[sidehack]

They haven't responded to my inquiry yet.

Also, we lit up Max last night and I think we fried the ASIC. Doesn't initialize even on low voltage and frequency. Novak's prize stick, cherrypicked for top specs, also didn't really like running top speed like it used to. That stick ran 437.5MHz stable, completely unmodified. But cherrypick is a bit cheating so he didn't write it up. I think the stick actually ran 450MHz but errors were high. And apparently we tuckered out the ASIC because we can't talk it into doing that again.

[drew_ii]

thanks to everybody who helped this stick newb with cgminer.
finally cgminer-gekko is working and I can manipulate the stick at various settings.

quick questions:

1. do you have to do sudo reboot everytime you change something in hardware setup? it looks like this to me.
2. 800mv/433 mhz somehow resulted in stick not initializing, but it initialized OK with wiper at 3pm (750-760mv?) at 350mhz.

Q: which course of action is generally better: keep mv same (750) and see up to which mhz it will initialize OR increase mv to max (800) and then see which mhz will work?

It looks like even at 750/350 stick fluctuates wildly between 16 and 24Gh/s with average ~19 Gh/s (WU 265).
In my crazy setup watts and amperage from hub are not a problem.

So, the bit about using SUDO is located here
http://ck.kolivas.org/apps/cgminer/README
Search for 'rules.d' which will bring you to the LINUX section of Setting Up Devices - it speaks to the usergroup modification for access to the USB devices.

In re to the questions above,

1) You can adjust the mv pot on the fly while it is running, just be careful not to short if you're using a metal phillips.  However, if you want to change the mHz (freq), you have to stop and start CGM w/sudo and let it ramp 2-4 mins (cgminer-gekko is designed to ramp the mHz to avoid power surge on the usb port).

2) >400mHz has been very difficult for me to do.  350 is as easy as the ratio of a circle's circumference to its diameter.

If you're looking for as fast as possible, plug the stick into an extension cable into your port, and using a multimeter check the mv and adjust accordingly until you hit 800.  CGM doesn't have to be running.

Once you are at 800, start up CGM at 400 and let her ramp.  If no errors, go to 405 or 415 (410 is broken - see previous) and wait for ramp again.  If errors, adjust pot clockwise until errors go away or the stick locks (this is when you go from 800mv back down to 550 - you will have to shut down CGM and unplug/replug the stick to reset (remember to safe-eject on win/mac)).

If you adjust pot while CGM isn't running after restart down to 550 and nudge it backwards, you should be able to hit the max of the pot.  However, I think the specs say that 800mHz is max, so if you hit 900+ look out - Danger Will Robinson.  I know one of my sticks can get to 900, I just checked it yesterday when I was setting up my three new ones.  I can get the three new up to 800x420 with very low hardware, but I can't get my initial stick over 375 clean and 400 w/errors, and when I checked the mv it was > 900!  Pretty sure it has caused some heat damage.  I *can* run her at 800x400 but she errors like the day is long, but thats keeping my other sticks at less than full potential, so I'm trying to replace her and get my bank running at 800x455 for 100GH - but that looks to be impossible from upthread   I'll settle for 430+ tho   Ahh pipe dreams..

Anyway, I ramble - good luck with the OC!

[alienesb]

The 4 sticks I got from Scott are vexing me. The three original sticks do not want to work with them at all. I can run the O3 (original 3...) at 250 not problem and get solid results with at least 30GH. If I start adding the N4 (new 4...) I start getting hardware resets and the hash drops to like 3 or 4GH each. I'm using CGminer and I got a new Anker USB hub to run these and no joy 

If I try to run the N4 all by themselves they seem a bit weak too not hitting more than 6GH or so each.

The Anker is running the O3 at 300 solid (around 200WU each) too so that's good.

I'd love to be able to run all 7 but for now I'm a sad panda.

[sidehack]

I'm assuming you've tested each stick individually and have tried adjusting voltage and other basic troubleshooting steps outlined in the first post.

[drew_ii]

FWIW, throwing this out there for any other Raspberry Pi Zero nerds - don't use the Raspberry Official WiFi dongle to stratum mine.

cgminer-gekko loses connection with eligius over stratum *very* frequently, as much as 1-2x/min.  I've had 60k accepted and 25k rejected after running overnight (no HW).  I got a Pi 2 B which has on-board NIC, but used the same configuration with the WiFi dongle in my unpowered 2.0 and the loss of connectivity still happened.  Its not signal power or reception as they are literally in the same room < 8' apart.  I plugged in a Cat5 into the Pi 2 B and its been running 3h38min with 0 stale, 0 unable to get work, 0 delay, and 0 rejects.  I cannot say that it didn't lose connection as the "started" time was literally 2 mins prior to my starting this and stopping CGM, but before I loaded this page it still had the same start time 3 hrs ago.  On the WiFi dongle, yeah its not very good.

This might be resolved with a USB Cat5 NIC if its not USB bandwidth saturation and is actually a signal problem, not sure.  I do know that my Cat5 is 1GB (does Pi 2B support that fast?) whereas the WiFi is only ~52MB/s or so, but WiFi has to share with > 10 devices in the house.

Mah 0.000044BT