[RESOLVED] - USB Block Eruptors & cgminer 3.3.1 - low hash rate????

[mccminer]

Okay.........I've searched this forum and the internet for hours, with no luck finding my issue.  I'm about to pull out what little hair I have!!!  

Ubuntu Linux 12.10
cgminer 3.3.1
Four 7970 GPU's
Five USB Block Eruptors


This miner has been running flawlessly with the four GPU's in it, so I decided to try to add the USB Block Eruptors to it.  Cgminer recognizes all of the devices, but for some reason, the USB miners mine at a really low hash rate.  Most run between 75 - 95 MH/s.  I also get a lot of "AMUx:  TIMEOUT GetResults took xxxxxx ms but was 100ms" logs with the USB miners.  

I haven't edited the cgminer.config file to add or change anything.  It was my understanding that version 3.3.1 didn't need any options to run the USB miners.  Also, I've tried the USB miners on the mobo, on a hub, one at a time, with the GPU's disabled and I always get the same results.  ARGGHH!


Any thought??

Code:

cgminer version 3.3.1 - Started: [2013-07-24 12:51:10]
--------------------------------------------------------------------------------
 (5s):3.378G (avg):3.424Gh/s | A:449  R:0  HW:2  WU:48.6/m
 ST: 2  SS: 0  NB: 3  LW: 1086  GF: 0  RF: 0
 Connected to xxxxxxxxx diff 1 with stratum as user xxxxxxxxxxxxx
 Block: 00376f8a1122b65f...  Diff:31.3M  Started: [12:59:44]  Best share: 127
--------------------------------------------------------------------------------
 [P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
 GPU  0:                | 676.8M/677.9Mh/s | A:81 R:0 HW:0 WU:  8.7/m I: 7
 GPU  1:                | 676.8M/677.9Mh/s | A:81 R:0 HW:0 WU:  8.7/m I: 7
 GPU  2:                | 661.2M/662.4Mh/s | A:87 R:0 HW:0 WU:  9.4/m I: 7
 GPU  3:                | 625.4M/626.6Mh/s | A:81 R:0 HW:0 WU:  8.7/m I: 7
 AMU  0:                | 174.0M/96.29Mh/s | A:19 R:0 HW:1 WU:  2.2/m
 AMU  1:                | 186.0M/70.64Mh/s | A:10 R:0 HW:0 WU:  1.1/m
 AMU  2:                | 160.3M/63.89Mh/s | A: 4 R:0 HW:0 WU:  0.4/m
 AMU  3:                | 201.4M/73.39Mh/s | A: 9 R:0 HW:0 WU:  1.0/m
 AMU  4:                | 287.6M/90.37Mh/s | A:14 R:0 HW:0 WU:  1.5/m
 AMU  5:                | 290.4M/103.2Mh/s | A:16 R:0 HW:0 WU:  1.7/m
 AMU  6:                | 229.5M/104.9Mh/s | A:20 R:0 HW:1 WU:  2.3/m
 AMU  7:                | 177.5M/78.69Mh/s | A:11 R:0 HW:0 WU:  1.2/m
 AMU  8:                | 175.3M/115.7Mh/s | A:20 R:0 HW:0 WU:  2.2/m
--------------------------------------------------------------------------------

 [2013-07-24 13:00:05] Accepted 74a5fd62 Diff 2/1 GPU 1
 [2013-07-24 13:00:05] Accepted b0cad131 Diff 1/1 GPU 0
 [2013-07-24 13:00:06] AMU5: TIMEOUT GetResults took 7963ms but was 100ms
 [2013-07-24 13:00:06] Accepted 0c0e1af4 Diff 21/1 AMU 5
 [2013-07-24 13:00:08] AMU4: TIMEOUT GetResults took 9544ms but was 100ms
 [2013-07-24 13:00:08] Accepted 16816ead Diff 11/1 AMU 4
 [2013-07-24 13:00:09] Accepted 444a33b5 Diff 3/1 GPU 0
 [2013-07-24 13:00:13] AMU4: TIMEOUT GetResults took 5133ms but was 100ms
 [2013-07-24 13:00:13] Accepted b62ef9f7 Diff 1/1 AMU 4
 [2013-07-24 13:00:16] Accepted 314eb0d8 Diff 5/1 GPU 2
 [2013-07-24 13:00:16] AMU5: TIMEOUT GetResults took 9936ms but was 100ms
 [2013-07-24 13:00:17] AMU4: TIMEOUT GetResults took 4058ms but was 100ms
 [2013-07-24 13:00:17] Accepted 7d64f8c5 Diff 2/1 AMU 5
 [2013-07-24 13:00:17] Accepted b08fea24 Diff 1/1 AMU 4
 [2013-07-24 13:00:17] Accepted eceaaf54 Diff 1/1 GPU 2
 [2013-07-24 13:00:19] Accepted 644c648c Diff 2/1 GPU 3
 [2013-07-24 13:00:20] Accepted d9260317 Diff 1/1 GPU 1
 [2013-07-24 13:00:20] Accepted 8b3d717f Diff 1/1 GPU 0
 [2013-07-24 13:00:22] AMU4: TIMEOUT GetResults took 5283ms but was 100ms
 [2013-07-24 13:00:22] Accepted 8555840c Diff 1/1 AMU 4
 [2013-07-24 13:00:26] Accepted 710d07e9 Diff 2/1 GPU 0
 [2013-07-24 13:00:26] Accepted f1860fe9 Diff 1/1 GPU 3
 [2013-07-24 13:00:28] Accepted cf1c6aec Diff 1/1 GPU 2
 [2013-07-24 13:00:28] Accepted 5dfe03b6 Diff 2/1 GPU 0
 [2013-07-24 13:00:28] AMU4: TIMEOUT GetResults took 5952ms but was 100ms
 [2013-07-24 13:00:28] Accepted 2d9c438f Diff 5/1 AMU 4
 [2013-07-24 13:00:28] AMU5: TIMEOUT GetResults took 12247ms but was 100ms
 [2013-07-24 13:00:28] Accepted 85282ba1 Diff 1/1 AMU 5

[Chronikka]

Supposedly cgminer 3.3.1 is designed to work well with these USB miners. But I had issues getting them to run correctly and I actually reverted to cgminer 3.2.1 and have been happily mining with them ever since. Might be worth a try for you.

[os2sam]

CGMiner 3.2.1 worked great with one instance but crashed with 3 different instances.  Updating to 3.3.1 has resolved that issue and has been running great for me.

Did you run the Zadig utility with no erupters plugged in to install the WinUSB driver and the add the erupters and set them to use the WinUSB driver per the ASIC Readme?

I have not seen that behavior before.
Sam

[mccminer]

I'm not running them on Windows.  They are on Ubuntu Linux 12.10.

[os2sam]

I'm not running them on Windows.  They are on Ubuntu Linux 12.10.

Doh! I did see that too.

Doesn't the readme have something to say about USB drivers?  If not please ignore me
Sam


Edit: and the crash I was talking about with 3.2.1 was a Windoze only problem too.

[Chronikka]

One more thought, you are providing them enough power correct? .5 amps per stick. If your hub is struggling to provide power it could result in hash rates being lower.

[mccminer]

The ASIC-README says the following:

LINUX:

On linux, the direct USB support requires no drivers at all.

As far as the current to each device; yes, I believe each device has the appropriate amount current delivered to it.  I've tried them in all of the mobo USB slots, as well as a single stick in a powered USB hub.  The psu with the hub is rated 5v @ 3.5 - 3.6A.

[mccminer]

Ok....here's an update:

• If I only plug one USB miner into the mobo, without the hub connected to it, it mines at @333MH/s.
• If I only plug one USB miner into the hub, without a miner in the mobo, it mines at @333MH/s.
• If I try to combine 2 USB miners in any fashion, the hash rate drops significantly.

[mccminer]

This is a cgminer issue.  I found this article, so I decided to try BFGMiner, and all 10 of my USB Block Eruptors are working just fine now.

In case anyone is interested, I'm running BFGMiner 3.1.3

[kano]

...
Any thought??
...

Just an update about libusb in case anyone was wondering ...
https://bitcointalk.org/index.php?topic=28402.msg2817682#msg2817682

Basically I'm just verifying (trying to get others to check also) that it is indeed the version of libusb that's the problem.
Upgrading libusb may resolve it ... but once I have verified if it is the cause I'll go with including the latest libusb in the cgminer directory to ensure people are using a working version

[mruiter]

Have the same problem. This morning i updated te to usb 1.0.16 newest and build it.
Also used the latest git pull from cgminer.

Still the same AMU8: TIMEOUT GetResults took 1449ms but was 100ms errors

Using BFGminer at the moment, but like to have both running on my Rasp Pi miner so i can have an alternative.
And cgminer is/was my favorite.

Please let us know if something else has been found to resolve this.

[kano]

Try the libusb code I listed, since that will be what I'll put into cgminer if needed.
That test program I have will tell you if everything will be OK, no need to recompile or relink cgminer for testing
The cause of the cgminer problem is that, I call libusb in the Icarus driver and it either doesn't return for 60 to 90s on any work without a nonce, or it never returns. The timeout value passed is 'ignored'.
The 60-90s version gives you around 100MH/s on that device, the never return version gives you SICK on that device.
On all but my RPi I don't get this problem.
On my RPi I get this problem with Raspbian, but don't get it with Arch.
I've since linked cgminer on the same failing Raspbian with the version of libusb I listed and the problem is gone.
So I'm pretty much certain it's a libusb problem.
The version numbers shown on different Linux distributions don't mean it actually has exactly the same code - since they often put patches into their versions.
I guess for you there could be some other hardware issue (USB3 hub) but for me this works even on a USB3 hub.

[greenbird77]

Thanks for this helpful thread. I hit this problem when upgrading from Debian squeeze to wheezy with libusb 1.0.11  and the bug went away when upgrading to libusb 1.0.17 from wheezy-backports.