Topic: GekkoScience Compac BM1384 Stickminer Official Support Thread - page 123. (Read 268015 times)

I'm only running at 150 MHz, but at least one is ringing. I'll try tomorrow on a different hub to see if there's an issue with power.

interesting, I have tinnitus also but anyway the 2 compacs I have dont sing (highest clock is 175 on stock USB 2 ports in my HTPC, Ill go higher on a modded hub later). but my GTX285s I ran for folding@home had bad coil whine. a somewhat common fix was to put nail polish or something similar all around the coils, which I did to two of them, and it did fix the singing. they ran overclocked for about a year 24/7 like that, and are still runnig today in my game rig.

not that I am recommending that with the compac, and Im sure sidehack is cringing reading this (I am prepared to be chewed out BTW). the compaq inductors are not the same as in the GTX285s and Im sure run in a different power profile.

some PSUs have coil whine too but I never had one do it. Seasonic, PC Power and Cooling and Antecs here but Im wanting one of sidehacks PSU/breakout boards eventually.

The only times I've had any sound happen was when clocked high and getting a voltage drop on input power that the regulator on the board wasn't happy with, causing there to be a bit of coil singing in the inductor.  That would only occur during (attempts at) initialization, though, so a relatively short 'beep' and not all that high in pitch (but tone did vary a bit).  Example video: https://www.youtube.com/watch?v=zxcYBbIvZEg

not sure if youre talking about windows, but the only time ive ever had this problem in linux is when the hardware has malfunctioned.

I have a little info for Luke-Jr: My use of -d? -D produced about 20K bytes of output that I haven't tried to post here. I'll find a way to provide the whole file. I am however providing the last few lines which may provide a clue:

 [2015-09-17 18:30:05] lowlevel_scan: Found vcom device at com:12 (path=\\.\COM12, vid=0000, pid=0000, manuf=(null), prod=(null), serial=(null))
 [2015-09-17 18:30:05] lowlevel_scan: Found vcom device at com:11 (path=\\.\COM11, vid=0000, pid=0000, manuf=(null), prod=(null), serial=(null))
 [2015-09-17 18:30:05] lowlevel_scan: Found vcom device at com:10 (path=\\.\COM10, vid=0000, pid=0000, manuf=(null), prod=(null), serial=(null))
 [2015-09-17 18:30:05] Devices detected:

What's noteworthy about these is that COM10-COM12 are the device names that Windows assigned to the sticks. I can see them in a Control Panel display, and they are found as "AMU" if I add them using the \\.\COMxx path name after bfgminer is started.

One thing I just thought of: This system is actually my work laptop, and it's Windows 7 Enterprise. I am however NOT the Administrator of this box. I wonder if perhaps device detection is impacted by that?

Thanks for your interest.

Thanks, after restarting with the new link it seems to have normalized my reported hashing.

http://solo.ckpool.org/workers/14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd.compaq

I have to ask, I have really bad tinnitus (ringing in the ears) from doing time over in the sandpit... is it me or do these compaq's sometimes have a high pitch coming out of them?

Got my 2 sticks. Working great out of the box at stock frequency with cgminer 4.9.2 master with xnsub patches (initializing as AU3-GS10000). With this setup theyre hashing happily alongside 2 old Rockminer R-Boxes. For a whopping total hash rate of 82G or so.

Probably wont get to tweak them until this weekend. Looking forward to it though.

too fast bro  . give me 1 min

change  sticks to this:

bfgminer.exe -o stratum+tcp://solo.ckpool.org:3333 -u 14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd.1 -p x --set compac:clock=x0b83


then  http://solo.ckpool.org/workers/14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd.1

gives you just the sticks

looks like you just started it

then set the u2's to this

http://solo.ckpool.org/workers/14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd.2


and the master address to this for grand total

http://solo.ckpool.org/users/14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd

Correct, I have my U3's also pointed at this address.

at ckpool you are:

 http://solo.ckpool.org/users/14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd


if you change the sticks to

I'm running 3 sticks with BFGMiner, what's weird is that my 5 second average and all time is clocking at about 40GBs, but I only have them running at 150MHz with:

bfgminer.exe -o stratum+tcp://solo.ckpool.org:3333 -u 14LxDWtRRdQAbnJdd4Ew5jwag8Bx9UfNRd -p x --set compac:clock=x0b83


 bfgminer version 5.3.0 - Started: [2015-09-17 17:46:07] - [  0 days 01:10:31]
 [M]anage devices [P]ool management [S ]ettings [D]isplay options  [H]elp [Q]uit
 Pool 0: solo.ckpool.org     Diff:23  +Strtm  LU:[18:56:36]  User:14LxDWtRRdQAbn
 Block #374994: ...1a3a3002  Diff:59.3G(424.7Ph) Started:[18:29:30] I:19.39uBTC/
 ST:5  F:0  NB:8  AS:0  BW:[ 62/ 54 B/s]  E:224.10  BS:110k
 3            | 121.8/121.5/32.89Gh/s | A:1298 R:112+0(.21%) HW:19/.06%
--------------------------------------------------------------------------------
 CBM 0:       | 40.28/40.54/10.91Gh/s | A: 433 R: 47+0(.26%) HW: 5/.05%
 CBM 1:       | 40.35/40.49/11.15Gh/s | A: 444 R: 45+0(.25%) HW: 2/.02%
 CBM 2:       | 41.09/40.70/10.88Gh/s | A: 421 R: 20+0(.11%) HW:12/.11%

Okay after a few days of running and testing 3 windows builds I have concluded I can run 3 sticks at freq 331 with zero issues on win 7 build.

and I can run 2 sticks with no issues on win 10 build.  both with   novack's gekko cgminer build.

I can't do better with more sticks with windows and his build so I will now play with  bfgminer..


So at alh    please post your bfg  bat file  so I can work off it.


now for the good news.  I disassembled my stud hubs to be sure they are working correctly  and I found this psu inside


http://www.meanwell.com/mw_search/NES-100/NES-100-spec.pdf

specs are
5volts
20 amps
100 watts max
4.75 to 5.5 volts

80% efficiency

159mm by 97mm by 38mm

this  is a direct replacement meanwell hrp-150-5

http://www.onlinecomponents.com/datasheet/hrp1505.aspx?p=37810127

5 volts
26amps
130watts
4.3 to 5.8 volts

85% efficiency

159mm by 97mm by 38mm

so I will buy two of them put them in the stud hubs

it will free the 2 psu's in the stud hubs and I will use those psu's to run my lesser usb hubs.

Got my production unit today in the mail.  Thanks sidehack and novak!

Below is setup I will be working on today.   It has the original test black unit, reviewer gold unit, and finnaly the great looking green production unit.

Can you post the output from: bfgminer -d? -D

Depends on how much you overclock it.  Probably around 350 is a practical max for long term, I had four sticks running at 400MHz over a weekend and they had only two hardware errors and averaged 87.90GH/s, where expected is 88GH/s.  Sometimes though, some of them didn't seem to want to properly start up at such a high frequency.  They never had any trouble starting at 350 or so, however.

Also you're probably not going to run very many sticks at 400MHz unless you have a really beefy hub or one with a whole bunch of extra ports not in use.


I've seen odd behavior of cgminer on windows not actually starting to mine at first unless you hotplug as well.  Usually it only does it the first time I plug a stick into my machine.  It seems to be something with the zadig drivers, I'll see if I can figure out why it does that but it's lower priority than some of the other driver improvements.

As far as detecting as an AMU- it gets pretty nasty to differentiate the compac from the U1/2/3 because from a software standpoint the only difference is the extra identifiers we added to our sticks.  If it checks to see if they look like an AMU, they appear identical because bitmain added no such info.  You might ask Luke-Jr if maybe you used a slightly wrong COM identifier or something else that made it come up that way.

--
novak

I have been tinkering with my 3 sticks, and been using bfgminer on a couple of different Windows 7 boxes. I observed a kind of odd behavior that I thought folks might find useful/interesting..

I had assumed it would always be best to have the sticks plugged in prior to starting bfgminer. I have yet to have example where bfgminer discovers anything on it's initial pass at start up. I then have gone through the process of "adding" them after bfgminer was started. I hade to use the \\.\COMxx reference. They would usually add just fine, and they showed up as an AMU.

Without actually realizing what I was trying, I plugged in a 2nd stick while bfgminer was running. I noticed in a couple of seconds, the LED started to flash. I then looked back at the screen, and bfgminer had automagically added a new miner to it's list and started using it. In this case though, it was a "CBM" miner and not an AMU. I then hot-plugged the 3rd miner and it to was detected and added as a CBM.

I then shut it all down, and unplugged the sticks after letting them cool (no fan). I started bfgminer with no sticks and over the course of 10-15 seconds hot-plugged all 3 sticks and I now have 3 CBM miners running at about 10-11GH each.

It appears to me that hot-plugging is actually "better", which I found surprising.

I agree with the overall consensus that these are great little stick miners.

Thanks sidehack and novak!!!

So 3 compacs OC'd equals roughly a U3, that's pretty cool.  Much less power use, much less hardware problems, and made by a member of our community, not some Chinese mining conglomerate.  Can't wait for my 4 to arrive!

yeah that does 7 sticks at about 5.2gh  freq 125.  I am pretty much fine with that setup as it is in a shell.

and I use the mac for a shit ton of other stuff.

but my other win 7 is not in a shell it is a high end pc an i5-4670s 16 gb ram and a 250gb ssd

and it has weird issues basically the same as the w-10 build  but the issues kick in when I get to 6 sticks while the w-10 kicks in at 3 sticks.

I can run 4 sticks on the w-7 pc at freq 325  and get full performance …..

Yet I can't so 5 sticks at freq 200 and get full performance.

at 325 you should get about 18 gh  a stick  I do  with 3 sticks or 2 sticks.  let me screen shot 3 sticks at freq 325

at 200 you should get about 11 gh a stick I get more like 9gh  if I run 5 sticks.

3 sticks with good cooling and tons of power

the fan can do 300cfm  ac driven



numbers are pretty good for 3 sticks at freq 331  should about 51gh and is around 48gh for overnight I am fine with this




a really good i5 win-7 system made for mining only:

 cpu is good
 ram is good
 ssd is good

gpu who cares as it is an  intel built in

I'm seeing something odd regarding hardware errors. I've got two sticks running at stock 150 speed on two separate powered hubs connected to an RPi.

This is in an air conditioned server closet, with Arctic fans blowing on the sticks. They probably don't need the fans yet, but I had them in place already from a previous stick rig. Power comes from a UPS/PDU.

Typically the hardware errors are very low. But sometimes they blossom suddenly. For example, one unit had a one or two hardware errors after running for over a week. I checked this morning and now it has 40. Same thing happened with the other stick earlier in the week. It went for days maybe a hardware error or 2 total, then I noticed it had 70. After that burst, it's only added 9 hardware errors in the last few days.

The thing is that nothing is changing on my side. The power and temp should very consistent. I've been pointing at the same pool. No software updates.

The only thing I can think of is that the sticks are on the top on a gorilla rack, with some Mac Pro build systems a few feet below them. At night at least one of the build systems fires up and runs full blast for a few hours. Maybe this increases ambient heat enough to cause hardware errors? If so, why aren't both sticks affected at the same time? Also, several days go by without a burst of HW errors, but builds happen every day.

The total percentage of hardware errors to accepts is very low at 0.02%. So from a pragmatic aspect, this really isn't a problem. It's just strange how bursty they hare. I haven't seen that with other hardware before.

You're running Windows under a VM on a Mac, right? Have you tried running on the Mac native instead of a VM?

According to the nwools' tutorial, seems it's fairly easy to build bgfminer on Mac (if there isn't a 5.3 Mac binary already). I ended up going with RPi, so I didn't end up building for Mac.

I still think that the easiest way to go is RPi. Although I'm only running 2 sticks at stock speed (no y splitters yet). I haven't pushed things yet.