Topic: Community Miner Design Discussion - page 25. (Read 34218 times)

I'm all in.

A miner which can be volt-adjusted has a much longer viable lifetime. The BM1384 chips on the S5 could be run down to 600mV and 0.25J/GH versus the 800mV/0.45J/GH setting of the S5. If the machine was undervoltable it could run as efficient as an S7. Sure you'd only see 500GH at that setpoint, but that's not bad for something you paid for a year ago and it's probably already paid off. Look how long the SP20 was around - some are still around, because you could run it hot or run it cold. When did that come out, October 2014?

I don't like the idea of designing a board for a specific controller, because then either I have to make that controller or design around one that already exists (in which case you have to have one or get one). And then you limit the software to using exactly that controller, especially if you're using low-level IO. I'd much rather put the extra $2 in parts directly onto the board to make it more generic, able to connect and run off anything with USB and cgminer. If the software and hardware package is going to be generalized or in any way reusable, that is a mandated requirement. Why build a reusable framework that requires you to hook up to a Pi, when some folks would rather have something that hooks up to a Cubie or an old XP laptop? I can find no convincing reasons to settle on that arbitrary limitation when it's no more difficult to make it a lot more flexible. And then if you wanted, you could hub a dozen boards together and run them off a single controller instead of requiring a Pi for every 2, or 4, or whatever.

Heck, with that flexibility, the same controls framework for the TypeZero boards could be used on a pod or a stickminer as well.

If you have good power prices better efficiency can let a miner earn a long time.

Underclock/Undervolt - I understand you can underclock both the Antminers and Avalon but you cannot do both.  As stated by Padrino, it is for efficiency.  The compac is way more efficient than any S5 as far as I know although they use the same ASIC.

In regards to the life span, I am not referring to how long the miner will actually work but the Lifespan as a Miner - if the efficiency can be increased by making adjustments then the lifespan is increased.

Other uses would be if Bitcoin shoots up to $900 then you can dial it up to a greater Hash and accept the decreased efficiency.

I'll start by saying that this is really exciting stuff.

I get the direction of making boards that you can swap into existing hardware, it will work the way people are used to. It raises some questions though, the higher yield hardware all has built in management boards, why is that the norm? Raspberry Pi and cubie board and probably countless other development boards have i2c interface to manage hardware. Would it cut cost to make unmanaged boards that we could control with a pi and cgminer? I'm picturing a pi hooked up to 5 Frankenstein s7s running at double speed and efficiency and maybe slightly less heat.

In the spirit of efficiency is how I read that.. The value of undervolting depends on the design of the chip and its performance charactistics..

Speaking of which, Bitmain for example made some noise about 9V for a while but it seems to have faded..

Everyone lost out of WASP, everyone.

You can underlock S7s and Avalon6s, but why exactly would you want to undervolt them? Increasing life of hardware which is quickly becoming obsolete doesn't make much sense, especially when they may last 3-5 years anyway.

Well the good news is the only person that would lose money is me.  I am not familiar with the WASP project but I can assume it didn't go well.
I think I need to define the term "Community Project" in the way I am using it.  The objective is to produce a small efficient miner that is affordable.  We aren't saying it will have a nice pretty shipping box and/or a nice pretty enclosure.  There are a lot of ways to keep cost low and that's what we are aiming to do.

Other important design issues I will leave to SideHack et al.  Imagine if you could underclock and undervolt S7's,  Avalon6 etc.  By adding things that the big wigs leave off increases the life span and value to the miner.  

I/We are NOT asking for any money or Bitcoin from anyone.  I think this thread can allow people to have a voice in the miner and make suggestions.

Actually Sidehack already did some usb stick and everything was perfect. He now have a pick and place machine to be able to best faster.

Yeah its more hard to do it when you are not a big corporate but you can do a home miner 

I have faith in kilo17 and Sidehack

I don't think the finances will make this feasible...  It's not possible to "compete" with a mass produced product once they hit the market.  so a small batch produced by a community isn't viable financially, plus someone still has to be the face of the project, legally, to enter into agreements with manufacturers etc...  Unfortunately the days of making your own miner as a project are over.  It's now more mainstream with several manufacturers.

An s1 can be had on ebay for $25 possibly, I'm sure around here easily at that price. Buying an older miner to retrofit new boards will still be more cost effective than new gear. Also look at how much gear failed last few rounds from all the asic companies, this will be overbuilt with quality parts with the end user in mind and will be very flexable with underclock/undervolt and possibly overclock/overvolt. Look at how much an s7 was brand new, now imagine if you could have gotten it 2/3 of the price at the same time. Everything is calculated into a new miners price including having the asic built, sure this project will have some markup (why work for nothing) but nothing like batch 1 of any miner.

Watch the computer hardware thread, I guarantee when these boards hit that everyone who has extras will be selling their frames they have laying around.

I generally only support private initiatives however I have access and agreements in place for various chips... If you and sidehack already have something in play is there anything in particular you are looking for, or just a general conversation?

not sure that he meant those for sale, but if so, then it's good.

well he does say he has  40-45 case  frames .

I believe I can get five more frames as buysolar has at least four or five s-1s'   more like 10 but we would need some for refitting in the array.

seems like times are looking up a bit.

Very exciting project...thanks

Just a side note: Personally, i don't have S1, S3 or S5-had all three types, but sold them off as they became obsolete.
Buying used S3 or S5 again just to fit the new board?
Miner might have higher demand than boards, but I might be mistaken.

yeah between buysolar and I we are really looking forward to seeing this thread take off.

Here is a link to the solar array that we hope to use this threads gear in.


https://bitcointalksearch.org/topic/m.13932062

Yea I was in the middle of typing that, went to eat.... came back and finished it and you had already responded which made my post seem awkward. Either way I posted it. Total speed or number of boards don't matter to me and Im sure others wouldn't either. I have a s1, 4 s3s and 2 s5s around the house, Retro fitting 1 to possibly 2 would be more than enough for me.

It'd be a full-size blade, so two boards would make a full 4TH miner. If you had a C1 you could run it at 8TH, or more if you can keep it cool (potentially pushing 10TH at about 1200W DC).

I'm a big fan of stuff in the 300-500W neighborhood, and also between purchases and donations I have around 45 S1/3/5 laying around the shop to refit.

I'm in for anything that is quieter than a jet engine (s1/s3 range) and is reliable 24/7 without attention. Something within 100-600 watts would be great too, or at least something that can be underclocked/undervolted to that range at less than .1w/Ghs.

I like the idea sidehack was talking about building boards to fit s1/s3/s5/c1 heatsinks/frame when he was messing with the 1384. Either 2 or 4 boards per miner would give the community back choices depending on money they have or watt usage available. No more being railroaded into 1kw/$2000+ miners at launch with no other options rendering us out of the game. I am a home miner with limited funds paying ~$0.13/kwh. A 1/4 miner running 200w at about 2Ths would be great, giving an s3 (they are everywhere) a capable 8Ths/800w. Granted they can be packed more dense for higher speed at lower wattage but this raises the cost per board, I am figuring the worst .1w/ghs so 20 chips per 1/4 board at 100Ghs (maybe less). This would give a c1 a theoretical almost silent 16Ths.......... damn that would be a great dream for some guys out there.

The thing about designing off this chassis is almost everyone has one hashing/collecting dust or can grab a s1/s3 dirt cheap and retrofit it. Why waste time making/designing a case or heatsinks let alone make us pay for them when we have so many available for retrofit within 5 min to 3 days. Pump as many of these bad boys out as you can.

I have an interest in these. I just started a thread on the solar array project. With buysolar and myself.

So a bitfury miner would be very nice.

We hope to have 16kwatts. Since I have five avalon6s that use around 4.8 kwatts and he has an s-7 using 1.1 kwatts we have about 10 kwatts extra power.

I think the timing suits us since the array will finish just about when the bitfury gear drops.

So I am thinking this tech is something buysolar and I would want in on.

I see there are a lot of us in mining section that have a lot in common.

Kind of like a classic car club.

I will look in on this thread often.