Yes. Set to any IP within your router's range which isn't being used by another device. .98 and .99 are probably unused.

Yes, no (if you've got up to date firmware).

Absolute safest method is to set them both as DHCP so they'll accept whatever IP the router gives them. It will mean the IP to connect to them may float around a bit, but you can pin the IPs as preferred ones to the S3s so they won't move via your router.

2014 12 16 was bugged and was quickly replaced by 2014 12 19 so don't worry about that. A copy of 2014 10 24 can be found here, but make sure you heed the warning!

20141024 firmware contains broken reset button functionality. If you set network settings wrong or are unable to access to the UI/SSH, the controller is considered unrecoverable and bricked. It will have to be sent back to Bitmain to be reflashed manually. Use at your own risk.

Scamming scammers does not help the people the scammers were scamming and is likely to just make things worse.

Seems a coincidence that multiple PSUs would die simultaneously, especially the hard core PSUs you get in Spondoolies gear. Is there a chance that a breaker has tripped, a cable melted, PDU died etc etc which is causing this? Tried swapping the power cables (and so circuits) from a working unit with one of the dead units?

I don't remember specifically but maybe 2014 10 16. There maybe a frequency / voltage table someone has made in another thread but your best bet is setting the frequency you want, then see how low you can get the voltage but keep the chips stable. That point may be different unit to unit and will only take about an hour to find.

The second one looks like an unhappy board (try on different PSU, then swap its port on a controller to see if we can coax life into it) but the first one looks like an unhappy controller/PSU. That is evidence by BOTH boards being unstable (very unlikely naturally) and that the fan is going at 3k without any temp indication. You can try connecting the hashing boards on another controller if you lay 2x S5s on their sides and get the cables to reach.

Yes its disabled in the software (since around the new year) but depending on your board type it may not be possible on the hardware itself. I have the following if anyone needs them, but not all of them may be good useable firmwares.

2014 07 11
2014 07 21
2014 08 11
2014 08 26
2014 10 13
2014 10 24 <- staple firmware but with broken reset button
2014 12 19 <- reset button fixed
2015 01 09

Maybe because they're running low and S7 is still some way out. Or maybe unicorns, we're just guessing here

No idea, but its only $20. Probably an adjustment for a cost increase somewhere.

Oh sweet, innocent child You have yet to experience the glorious roar of 5,000rpm 120mm fans.

The typical reasons are inconsistent earnings due to low network %, high pings, higher complexity, lower performance / earnings. P2Pool is a hard sell towards new users.

I think its about time to lock this thread now and get on with out lives. Continue your CIYAM work, continue contributing and time may be able to repair some of the trust and trust ratings.

What sad news And I finished mining 180 but with 11 blocks, so my OCD will never resolve that my average won't catch up now.

810 is already too high really, I wouldn't run 24/7 higher than 780. Remember that C1 boards are already clocked fast compared to S3 boards and they're not that different.

Fuck your cake he's having mine

Kano, again you're looking for a fight rather than a discussion and I won't indulge that. You can believe whatever you want.

I bring you cake and backup cake if you want more cake. Not too much backup cake as I ate most of the cake. There is only this cake left. Omg it smells so good.

I agree, the ideal is to set something like an i7 4790K as 1.00 performance and everything else is a proportion of it. Performance from that point won't increase more than 10% YoY and will remain at the same price. If anyone other than AMD / Intel / ARM + other can design a chip that outperforms an i7 4790K then they've got a bigger market than just Bitcoin to play with and several $B in the bank already.


Having 2 companies almost too large to care about Bitcoin as the ONLY competitor is a great thing and a significant improvement on now. At the moment its too easy to design an ASIC and there is too much performance improvement YoY. We went from 10W/GH to 1W/GH now to 0.2W/GH. And that best chip is controlled by ONE company who's sole purpose is to mine. If they choose not to sell they simply win the network and no one can do anything. Comparing that to the Intel / AMD self mining scenario, we all have access to the same weaponry with the same performance. That is a much healthier scenario than 5 companies manufacturing the ONLY weapons, most of which who refuse to sell to us peasants.


Why don't you email him to check? I'd reupload but I have less than 1mb of upload bandwidth.

That actually sounds promising; rather than just varying the service string every block, you also vary the algorithm. I'm looking at this through absolutely naive eyes so there's a good chance that I've completely misunderstood how things works but here goes; simplified:

1) Take previous block header -> Put through a known, standard algorithm to generate a string.
2) Break that string up into areas, ie the first 4 characters = 1 new string, second 4 characters = another new string
3) Map those new strings with range values into configurations of the algorithm either by function or feed. [I just made that up].
     a) By function means you either loop the entire thing X times (not that useful) or run multiple loops on individual some operations in sha2. Ie run S1 multiple times using the outputs as new inputs then continue as normal. Or even more complex, vary how many times the individual operations rotate and by how much.
     b) By feed means you run multiple loops or partial loops, and then repeat it but with a translated character order. So ABCDEFG = FAEBGDC and repeat. The translation string from the block header could include one set of translations or multiple sequential ones to increase the number of combinations and prevent ASICs calculating them all.
4) PoW that, somehow .

As long as the resultant algorithm is sufficiently different or potentially different from say sha-256, it couldn't be overcome by a standard SHA256 ASIC. Even if you made areas on an ASIC to do the component functions of a generic sha2, you'd be limited in speed by your controller which surprise surprise isn't that different than a CPU. And as soon as you stop running a standardised sha2 function then ASIC usage should just be non existent.

Nothing to stop you loading some of the older firmwares if you need to undervolt now. New firmware could be a week out, or it could be many many weeks out. Just don't wait up for it.