Quote from: Bitfort on September 27, 2018, 05:45:16 PM

Quote from: aarons6 on September 27, 2018, 05:24:49 PM

Quote from: Bitfort on September 27, 2018, 05:01:32 PM

Quote from: aarons6 on September 27, 2018, 04:52:51 PM

i have one that is consistently slower than the other ones.

i haven't adjusted anything, should i try for the core or memory? im running them at speed 796.

Definitely should tinker all of them.
You have 35%-40% HW error rate !!! ( HW are hardware errors) ... your miners are actually working only 60 - 65% capacity)

I'm keeping HW under 3%.

EDIT: Read the OP. There is guide how to test them (using --benchmark flag).

the hw rate on minera is calculated by accepted vs error. its not accurate. im running on a solo pool with very high worker diff.. so the accepted will be a low number compare to hwe

in bfgminer log its only 1.5%
[2018-09-27 19:24:18] 20s:34.65 avg:34.79 u:33.88 Mh/s | A:64840 R:230+0(.35%) HW:40689/1.5%

ive tried speed 756 and 768 and it still is the same.

Then it's ok. Suprised you can go that high. Might be some lucky batch.

About the one miner try this:

Quote from: jstefanop on March 01, 2018, 11:01:03 PM

Quote from: s1akov on February 28, 2018, 07:54:45 PM

I was running Moonlander 2 for a couple of weeks, from internal USB of the laptop and had stable 4.7MHz/s speed at 832 freq.
Now, I got a good quality 2A USB hub. I put the core voltage slightly up and frequency 852. For a couple of hours it gave me speed of 4.9MHz/s and then stopped. I put the voltage back to previous position but now, no matter what frequency I put I get a speed of 3.1-2.4 MHz. Anybody knows how to check if I damaged the device? E.g. is there a way to check number of alive cores?

Thanks!

Put "-D 2>log.txt" at the end of the bat command. Start it up and it will spit out the log.txt file in same directory. In there you will see startup sequence and it checks for number of active cores. Should say 64.

Yea I don't know why minera does not pull the HW error rate % directly from the miner instead of doing its own wrong calculation. I should probably update the driver to normalize the error shares based on the difficulty its mining (each HW error is hardwired to the internal ASIC difficulty which is very low).