I have this weird problem that started after I received my 2nd batch of 5 MLD2s last friday. I think the same problem has been reported earlier by others but I never found a good resolution.
First I had 6 MLD2s, no problems, I was running them all the way up to 954Mhz but now I have scaled back again to 876Mhz.
Last friday I received a 2nd batch of 5 (total 11), and since then 1 or 2 of the MLDs hash at about 1/2 the normal rate (normal is around 5Mh/s with my clock setting, these sit around 2-3Mh/s). Several people have reported a problem like this before.
What happens is that at least for one of the MLDs I first get the error "MLD : unrecognized response", then usually a "MLD: ASIC has stopped hashing, attempting to restart" and from there it goes downhill. And the MLD usually settles around 2-3Mh/s.
The red lights keep flashing and I really cannot tell which one is hashing at halve speed, I can't tell a difference. Which miner gets assigned to which MLD seems also random and bfgminer gives a ASIC serial number of sorts, but one has plug in one MLD after the other, restart bfgminer to see which ASIC serial number corresponds with which MLD serial number to see which physical unit is playing up. So it is close to impossible for me to tell which 1 or 2 MLDs or USB ports they are connected to cause problems.
What I understand is that too low mem voltage may result in more frequent restarts, but that seems not the problem
What I understand is that too low core voltage may result in more HW errors, but that is neither the problem
I am still not clear about the symptoms in a situation when the MLD draws too much current, but as I am using the eyeboot industrual 19 port 40A hub with only 11 units, power should not be the issue. I have also observed the same going back to 5 units, so it seems something with 1-2 units. But I have no clue what!!!
I have turned of the USB hub, disconnected everything, reconnected, restarted, still same problem, not sure if same MLD?
Other details
Using win10, latest bfgminer + drivers, clock at 900MHz, problem also seen at lower clock (but I have not tried every clock setting).
It drives me nuts. I could live with one unit that's having issues, but 2 is too much.
I tried to google the unrecognized response, but did not get any further, so I might now check the source code and see what kind of conditions may trigger this.
Any input much appreciated.
try to double the worker difficulty pool side..
what i found is if its too low, the usb gets saturated somehow and it cant keep up..
