I'll 2nd the notion of the SP20 "underclock" (i.e. reduced voltage/power) and it's flexibility. It will hang right in there with an S5 in terms of efficiency at 1.15 GH/s. Personally I am setting for "power usage" and right at about 1170 GH/s and 600 watts. As the weather warms up, I'll dial that down for lower power, lower speed, lower fan/noise, and lower heat. I am hoping to be able ride it through the summer and maybe speed back up in the fall here in Minnesota. We'll see if difficulty and Bitcoin price will dash my hopes or not.

My simplistic understanding of the SP20 is that the internals of the machine will try and adjust the frequency upwards, until it works at the "Max Voltage" you can set. In other words you don't set the frequency directly, but rather dial the the voltage up and down (or the power), and let the hardware manage the frequency within the limits you have set.

There is a "Basic Settings" button on the "Settings" page that can get you started. You can then dial the voltage up/down, or the power (down from 288) to try and get what you are looking for. When you make a change, you'll need to let it settle in for at least an hour to make sure that it's really hit it's stride, and the parts have gotten fully warmed up.

The general consensus is that despite the advertised 1700 GH/s rate, a more realistic top end is probably closer to 1500 GH/s. For maximum speed, you'll also need a cool environment to make sure the ASIC's don't "thermally throttled". You'll also need a solid power supply with quality PCIe cables for top speed.

Reaching for top speed will also result in poorer efficiency (i.e. more Watts/gh). There are a couple of SP20 threads with lot's of good experience and suggestions.

Sorry for the delay, I wasn't paying attention on Sunday. Took the family to a movie and dinner out. Thanks for the payout, suchmoon:

1DtWBQyiaWyRYRofyPjmRNCpBqWUScqHRc

I agree that while an increase, 5% doesn't hurt anything like 25% last year......

So, what are temps in what appears to be your "mining cave"? With all that gear, I'll bet it gets plenty warm, no?

I buy into the maxim:

Mining follows price. It's a slow response, but when price goes up, mining WILL follow eventually, and vice-versa.

I am 99% sure that Price does NOT follow mining. If it was I'd be way better off than when I started in July of 2013. As I recall BTC was about $100, and difficulty was about 26 million. I paid way too much for my 333Mhz Block Erupter stick miner.

I just wanted to give you an option to consider. If it works best for you to adjust only "Max Volts" until you keep all the loops "cool enough" , that's great. If you want to try and get it "just right" (for you), then you might want to the the "Loop Power Control" method and see how it works for you. I am not trying to convince you of anything, just providing an option to consider.

I can't get at my machine at the present time (I am at work). The values I put up weren't mean to be a canned recipe for maximum performance. Each machine is a bit different because of the variability of the individual ASIC's that are inside. The point was to illustrate that you may be able to cool off one specific loop by adjusting it's power usage, and favor a cooler loop with more power. The external environment will also impact how hot your loops run, though Loops 0 & 2 will always be cooler than Loops 1 & 3, with the normal air flow through the machine.

I run my SP20 with a target power draw in mind, and don't go for maximum speed. As the weather warms up here in Minnesota, I'll dial it down in terms of power and speed to fit better my condition in the lower level of my home. I also have to worry a bit about the noise, though my wife hasn't complained yet.

I'll also add that using the "power Setting" for each loop is a good way to get you some finer control. While you can set a "Starting Voltage" for each loop, there is only one "Max Voltage" that applies to all the loops. I've had a fair amount of luck with setting each loop based what makes it work the best without getting to warm. Consider the following example:

Loop 0: 210
Loop 1: 190
Loop 2: 210
Loop 3: 180

The values here reflect the fact that Loop 0 runs the coolest, and Loop 3 runs the hottest. If you try and control the whole machine with just "Max Volts", then you probably have to dial it down low enough to handle the hottest loop, while the cooler loop isn't running at quite it's maximum speed. You can see by the "lim=288" on each of the PSU lines, that all the loops are not power constrained at all. Loops 0 & 2 are closest the cool air intake end, and Loops 1 & 3 are essentially "downstream" and get air that's already been warmed up by Loops 0 & 2. In my experience you need to give it at least an hour or more on a setting to see how the temperatures shake out.

I really think that it's more effective to try and control "loops" rather than by controlling "Max Volts".

I wouldn't hold my breath for the S6. If the past is any indication of the future, the S6 will be a more "data center friendly" S5. By that I mean it will likely have the same BM138x chips that are in the S5, but more of them in a package that fits in a 19-inch rack cabinet. The S6 will likely have an integrated power supply of unknown quality, and it's acoustic profile is completely unknown at this time.

I wouldn't expect the S6 to be a "quantum leap" over the S5, except in terms of packaging.

While I haven't followed the cloud mining market closely, is there a cloud mining service that has lasted one year on which to test this theory? My general take on cloud mining is that there are two risks: One if the usual risk associated with mining Bitcoins no matter what. The 2nd risk is that the cloud mining service just goes "POOF" and you are left with nothing besides the Bitcoin you have presumably shipped off to your wallet every month or so.

I just don't see the long term (multiple years) business model for mining. After the halving that's expected sometime in the 1st half of 2016, there will be another one. Eventually a block will deliver 1BTC or less. How will "transaction fees" support hundreds of Petahash requiring 100 Megawatts of power? It just doesn't add up to me that a supposedly small sliver (i.e. transaction confirmation) of Bitcoin activity can support even 1/10 of what we have today.

If you have 35lb of plutonium, you should bail on the whole Bitcoin thing and look for buyers of your plutonium. Way big profit there.

The "pipe dream" (ASIC, not Kate Upton), was actually more like 100 Watts, rather than 1000W. The Intel NUC is a smallish box that takes a 19V brick, not a 1000W AT power supply.

Someone else might be able to correct me, but a car battery while billed out as 12V is usually more like 13v+

Seems high risk to connect to your board. Just my $.02

Are you sure that's Bitmain? Maybe you meant Butterfly Labs?

Just skip the Monarch refund and apply it to the next "Learn O Tron" miner......

Maybe you just want to call it early at 5.97%.....

Just a thought!

Special delivery to Iran.......via F-16 maybe? Don't mind the 500lb package!

Looks like I won't need that 240V circuit for your next generation of miner after all......

So much for the 4U, 16KW plan, huh?

Do you only get 50 TH/s with 8GB or RAM?   

Assuming this is accurate, the J/GH must be minuscule (Say, less than .01 Watts/Ghash). Furthermore, assuming you deliver 1000 units/MONTH, that's roughly 100 Petahash/MONTH?