Next consumer model miner

[Mikestang]

I certainly hope technology and efficiency converge at some point, where something that small, cheap, and power saving could actually mine something of value.  1 TH/s USB ASIC sticks, put me down for a couple hundred!   

Or we can hope the other way, that large scale farm mining will become cost-prohibitive such that network difficulty will drop and existing "consumer" miners like usb sticks become viable again.

[dogie]

Avalon4 chips are and have been open for 3rd parties since release. Spondoolies often opens their chips up but availability is a bit behind actual release. I'd assume if you can put in a big enough order that Bitfury might consider taking your money as well, even Bitmain.

You'd know better about Bitmain, could you quantify 'big enough order' in their case?

What would Bitmain take to hand off a couple dozen sample chips?

I didn't mean to get your hopes up, I have zero information on if they'd actually sell chips. What I meant about money is that the more you have, the more likely they're going to be to listen. There's a good chance they'll never release them though.

[I don't usually hang around here, just passing through].

[sidehack]

That's kinda what I figured. I've never seen them release chips before, and I certainly won't have enough money to give them suitable encouragement to do so, but I'd sure like to play with some of them.

[sidehack]

So it's 5AM and I can't sleep. Second night in the last three nights that I couldn't sleep for all the thoughts happening. Who would get behind a BE300 board which could take in 7-20VDC, connect to a host via USB, and was fully software-adjustable between 350GH/140W and about 140GH/35W? If they were built so four of them could mount on an S1 chassis? These are just theoretical numbers based on BE300 performance test data from FriedCat's post, and what is hopefully a low estimate for regulator efficiency with the topology I have in mind. Hopefully the chips are affordable because I like this idea. We could sell boards direct which would save a lot on not shipping heatsinks, put a driver into cgminer and anyone with a retired S1 (or S3) and a USB hub would be sitting pretty. S1 heatsinks and AM Tube heatsinks are a similar size, so building them to fit either would be pretty sexy (though I'm not sure it's possible).

[dogie]

So it's 5AM and I can't sleep. Second night in the last three nights that I couldn't sleep for all the thoughts happening. Who would get behind a BE300 board which could take in 7-20VDC, connect to a host via USB, and was fully software-adjustable between 350GH/140W and about 140GH/35W? If they were built so four of them could mount on an S1 chassis? These are just theoretical numbers based on BE300 performance test data from FriedCat's post, and what is hopefully a low estimate for regulator efficiency with the topology I have in mind. Hopefully the chips are affordable because I like this idea. We could sell boards direct which would save a lot on not shipping heatsinks, put a driver into cgminer and anyone with a retired S1 (or S3) and a USB hub would be sitting pretty. S1 heatsinks and AM Tube heatsinks are a similar size, so building them to fit either would be pretty sexy (though I'm not sure it's possible).

I'm unsure on the need to retain the need for a large voltage input range as the only real sources of affordable power are still 12V. A bit of a band to keep 12V 12V would be nice if you're running DC-DC else it'd suffer some of the variation that S5 has due to different PSUs having a different interpretation of what 12V is.

[sidehack]

The model I'm currently looking at would probably have a 15V input limit. I was hoping for 20V input because a simple one-board quiet desk miner would be great for home users to play with and would run off a decent laptop brick. The requirement to keep 12V at 12V would be removed entirely from this board and input voltage variations between the listed upperbound and lowerbound would have very little effect on operating efficiency - maybe a 3% W/GH variation across the band.

[sidehack]

Also, just throwing this out there - Dogie got me an email address for someone at Bitmain and it looks like they're gonna be handing off some sample BM1384 chips so we can start prototyping that. I actually was working on the prototype power circuits earlier today and hope to have it fairly ironed out by tomorrow afternoon. Pretty cool.

[Xian01]

Also, just throwing this out there - Dogie got me an email address for someone at Bitmain and it looks like they're gonna be handing off some sample BM1384 chips so we can start prototyping that. I actually was working on the prototype power circuits earlier today and hope to have it fairly ironed out by tomorrow afternoon. Pretty cool.

Very interesting. Looking forward to following your efforts.

[sidehack]

I'm hoping to make something which can run quiet and fairly low power, something Jalapeno-style which is more powerful than a USB stick but not S1/S3 caliber. I got ahold of some New R-Box last week for repairs and that type thing is pretty good, but I'd like if it could run off a decent brick as an option instead of requiring a 6-pin PSU; that'd make it a bit more desk-friendly. I know bricks tend to have terrible efficiency, but with a 20V input rating on the regulators you could use a decent laptop brick. I'm pushing for 95% on the regulators instead of a current typical efficiency of 85%.
I want to make the clock and core voltage dynamically adjustable from cgminer command-line flags. Run the thing as a native USB-tethered device, and put as wide a clock/voltage range as possible on it so you could crank it up to full speed/power or turn it way down to silent and super efficient mode.
If the board is modular and practical enough, several could be put in a larger case for scaling to larger setups. I'm not sure how practical it'll be but I'm hoping we can work out building something that'd mount to an S1 chassis, basically I'd just have to ship you four bare boards and you could make your own approximately-S5 out of it (requiring a small USB hub and controller of course, like a Pi with cgminer). That's the dream, anyway.

I haven't put up a lot of details but Bitmain's rep says they're also looking forward to the project and "would like to support the good thing which is good for the bitcoin community" so here's hoping it doesn't crash and burn. Gekkoscience has always fought for the small miner, and if we can make this project happen affordably it'll be a good solid victory I think.

[Shazam!!!]

@sidehack...It will be nice to follow this. Are you gonna start a thread so we
can follow your progress??? Goodluck!!!

[sidehack]

I probably will at some point. I'll have to talk with Novak tomorrow about size/power targets, firmware code, and a few other things. We've had two big jobs fall on us in the last week or so that are going to keep us busy so hopefully we'll be able to budget enough time on this to get something going soon. I'm afraid this is going to be another one of those nights where I only sleep about four hours because I can't stop rolling the numbers around in my head. It's productive, but annoying.
Depending on what we decide tomorrow could make things run quicker, or make them wait for another project or two to come to fruition. I'm a bit torn over which is the better option right now so conferring is good.

[Searing]

People's idea of a 'home user' vary, but I don't think any of the major manufacturers have announced any concrete plans. Market realities and all that.
There's a page of two in the Spondoolies thread where they express the desire to continue to be able to cater to that market, though.

I'm on that thread last I heard they'd like to ...but have no way at this point how they would get it to fly...just may not be feasable...at least as a home miner and
any hosting option would likely be way to expensive for ave buyer from the past

not a slam ..correct me if I'm wrong (so want to be wrong and see another possible "reasonable chance of roi" home miner either bitcoin or scrypt)

problem is if btc or ltc price pops up and it looks releasable for a home miner (5-7 months down the road to develop and on sale) then this also applies to
data halls looking at the same price pop upward and same timeline (5-7 months) so by the time both the home miners and data hall units hit the market
the home miner buyer would again be buried by 'difficulty skyrocketing' etc

again correct me if i'm wrong but so far can't see this idea of a home miner 'in the air' so to speak
 

[TheRealSteve]

Home miners depend entirely on either the main producers catering to that market, or at least sending out chips to third parties.
Those third parties might fail miserably, or the price might come out too high, etc.  But at least they'll try.

For that reason, I commend Bitmain for sending off some samples of the BM1384 to sidehack.

In other news, SFARDS has finally gone public on their next BTC/LTC chip, claiming tapeout.  Remains to be seen what they'll do with that.

[Mikestang]

I probably will at some point. I'll have to talk with Novak tomorrow about size/power targets, firmware code, and a few other things. We've had two big jobs fall on us in the last week or so that are going to keep us busy so hopefully we'll be able to budget enough time on this to get something going soon. I'm afraid this is going to be another one of those nights where I only sleep about four hours because I can't stop rolling the numbers around in my head. It's productive, but annoying.
Depending on what we decide tomorrow could make things run quicker, or make them wait for another project or two to come to fruition. I'm a bit torn over which is the better option right now so conferring is good.

Looking forward to it.  I have an old S1 that I'm still running, as well as a couple U3s, so I would be someone in your target audience.  Good luck with this project, hope to have one of your boxes in my home mine soon.

[sidehack]

BM1384 sample chips have arrived. Snazzy.

[philipma1957]

BM1384 sample chips have arrived. Snazzy.

pictures pleaseeeee!

[sidehack]

Why? It's just a little plastic bag with a cut-tape strip of chips in it and some labels written in Chinese. Maybe once we have a sample board with one or two chips on it working I'll have something worth photographing.

[ezeminer]

Why? It's just a little plastic bag with a cut-tape strip of chips in it and some labels written in Chinese. Maybe once we have a sample board with one or two chips on it working I'll have something worth photographing.

The community is just really excited to see some new mining hardware

[sidehack]

Sorry, today's primarily a "manufacturing D750 boards" day so we can ship orders out. I probably won't even be able to play with power stuff for the board until Wednesday at the earliest, and I'm not sure how long it'll be for Novak to have even a one-chip test board to play with code.

[philipma1957]

Its cool I can wait.