Bump for added another unit. Got one more testing, hopefully will be added tomorrow. If I can find an agreeable source I'd like to keep these in regular stock.
|
|
|
You could probably get 625MHz at around 660mV, which would be a decent reduction from stock power use. The two I worked up today look to be pretty good; one of them is hitting 550MHz at 630mV with almost zero errors. I couldn't quite keep boards from dropping out at 575.
The MCU is programmed only from the 3.3V line from the controller. No power needs be connected to the hashboards or anything else for programming to work right, but disconnecting the 18-pin will prevent voltage from flowing back from the hashboard into the controller which overcurrents the PICKit.
|
|
|
No, it's really not possible to get the S7 reliably below 0.2J/GH which is still twice the power of the S9. Check the chart on page 4.
|
|
|
That's pretty much what I came to as well. Additionally, notes 4 and 5 (assuming "heat" is derived more from adjusting core voltage than clock) are mandatory requirements.
|
|
|
These do have a higher heat density than the S7LN so you can't run the fans as low. That said, there's less heat density than a full S7 so you can run the fans lower. And since the boards are pretty much at baseline stability, you do want to keep them warm to make sure the current is flowing. One of them I have the fans set at 30% in about 90F ambient and doing fine. The others have default fan speed and board temps in the 50s so they can be turned down some.
Short answer, quieter but still loud.
|
|
|
Don't forget to factor in the reduced hashrate. It's 33% less power cost but about 15% less return. Though the improved efficienty disproportionately decreasing power cost does extend viable life, which is the real benefit.
|
|
|
Looks like the S7LN has sold. I've got the numbers for the undervolted S7s so I'll put them in the first post.
|
|
|
You can probably get better answers if the question is posted in the altcoins forum.
|
|
|
Another difference between what I have listed and what Bitmain is selling, mine runs a slightly lower hashrate at substantially reduced power and fan noise. Phil can attest to that; he's got one of 'em.
DaveF, PM and we'll work out the details.
Later this afternoon I'll have some full S7 that have been undervolted as well. Overnight stability test looked good.
|
|
|
I have hosting space available if anyone's needing a place to put some miners.
|
|
|
If you need replacement PSUs I can hook you up. The best thing I have right now would require 208/240V. I don't have any S7 boards up for grabs at present though, sorry.
|
|
|
Make sure in your piecewise testing that you isolate whether it's the IO board or the BBB itself, since the two are separate boards that socket together. If the BBB crapped out, I can hook you up with a replacement for pretty cheap. I don't have any spare IO boards though.
|
|
|
I have some full S7s available, all of which have been undervolted and tested for stability (<0.01% HW errors) at a peak clock setpoint. Each has been tested on a DPS1200 PSU with breakout board and 10 cables. The numbers are as follows:
#60: 600MHz (4TH) 1050W 0.26J/GH (sold) #61: 550MHz (3.7TH) 870W 0.234J/GH (sold) #62: 600MHz (4TH) 950W 0.235J/GH (sold) #63: 600MHz (4TH) 925W 0.23J/GH (sold) #64: 600MHz (4TH) 970W 0.240J/GH (sold)
I'm looking for $450 from each 4TH unit and $425 from a 550MHz unit. This includes shipping (within the US) and PSU hardware - a DPS1200 PSU, interface board and 10 cables. If you don't want the PSU, subtract $75 from the asking price.
|
|
|
Step one is talk to holybitcoin
|
|
|
Right, but blocks of 12.95 since kano.is is paying about that each block
which is clearly visible in the screenshot. Anyone actually paying attention to the values they're putting into the calculator should notice the block payout set it to accurate. - though I wonder if bitcoinwisdom is going to update the default settings anytime soon; perhaps they should be reminded?
|
|
|
If you have asked me a question and I didn't answer it, and then you asked four more times and I still didn't answer it, that's either because I can't answer it or I won't answer it. I figured you'd realize that at some point.
So, my advice for helping solve the problem - if you're going to use long wires from a lab supply, drop a large capacitor (probably at least 220uF, maybe 470-1000) across the 5V and GND pins on the USB plug at the end of the wires you're hooking the stick into. Keep the input voltage below 5.5V; adjust the potentiometer for about 660mV and try a 200MHz clock. If that works, try 700mV and 250MHz, then 740mV and 300MHz. If nothing works, the stick is probably bad (but I won't point fingers) so contact the manufacturer for assistance.
|
|
|
I've got a few I'll be doing similar work to, with intent to resell. I should probably be doing that today but it's Saturday and I want to sit on my couch instead. Probably start on it tomorrow.
At 825W you can get away with 2 cables per board (assuming they're not garbage cables), so a less-well-outfitted PSU can still do the job.
|
|
|
Who would buy an S7 for their prices? The return math at this point is even worse than an S9.
|
|
|
I think a USB jack on a board with nice beefy power rails and a big capacitor close by, such as you would find on a hub, is better than 1 meter of pretty much any wire. Especially when you're asking it to feed current to a switching regulator. Long wires carry inductance, and inductance opposes current changes; switching regulators have really rapid input current changes. Long wires carry resistance, and resistance means voltage drops at high currets, which mean your regulator could be bottoming out and turning off.
I'm pretty sure this has been covered in the thread already. I know you're not a native English speaker, so I'll try and be nice, but I also get kinda annoyed easily.
|
|
|
|