Quote from: sdjernes on November 12, 2014, 09:44:43 PM

Hi All,

First Thank You Canary for doing the group buy!

Has anyone noticed that running these on the USB Controller is highly unstable?

I have two of my three units so far and both are highly unstable. I can not get them to keep up at anywhere near 1.4Ths. I can get them to run, but they drift all over the place from 600Ghs to an occasional burst at 1.5Ths.

I have tried running them on the RPi and the CubieTruck. I have used the stock 4.7.0 CGminer, Fixed Debian Build of 4.7.0-1 CGminer, and the BlockErupter 4.6.1 build of CGminer from GitHub. None have fixed or even made improvements. I have tried both the standard (no --bet-clk option) and --bet-clk=23 timing.

One thing I have noticed is that there is a lot of extended output in the BlockErupter Controller, that shows the chip status. Can we get that status from CGminer somehow?

Try pointing them at a pool port set for difficulty 2048 (S4).
I seen issues when when it comes to pool port set to vardif - units would hash for 1-1.25 hours then stop hashing - when this happen the difficulty would drop below 500 then bottom out and stop hashing till BE Controller was rebooted; had a similar experience on the RPi except it managed to hash for about 2.5hrs before dropping out. Since I changed to pool port set to 2048 it has been going non-stop for over 21.5 hrs.

I currently have 2 prisma using CG Minera Rpi image - with AM CGminer fork - running great. When I used the BE Controller the I could only have 7 of the 8 hashboards going - if I tried all 8 - one of the boards didn't work 85% of the time and had chips on 1 board go into PwrDn state (around 13 on both physical & logical blade).

Screenshots below are pool report hashrate, minera, & CGMiner output. The drop in hash around 5pm seems to be pool or internet related as my 10 S3s also experienced a drop at the same time.

http://i.imgur.com/yxqjoou.png

http://i.imgur.com/HZNK8CT.png

http://i.imgur.com/9tBewQm.png