Topic: ANTMINER S2 upgrade kit? EDIT: BITMAIN WHERE ARE YOU?!?! - page 5. (Read 35658 times)

You've gotta admit the S2 being a larger 4u case allows larger/quieter fans and with the backplane the simplest upgradeable design.

Novak and I talked for an hour or two yesterday, ironing out some dimensioning and such for what we want to do for a big miner. We're probably going to end up publishing the mechanical and electrical standards we'll be working within in case someone else wants to make boards for the same platform. But that discussion has digressed farther from the point of Bitmain putting out an S2 Upgrade, so it should probably be taken up somewhere else.

That sounds really interesting.  I've always thought that modular miners with standardized and easily replaceable hashing boards was the way to go.

Try $200 plus still on Ebay.  Posting another one today but hate dealing with the shipping - they don't travel well unlike the S4.  Just impatient in the long wait that may never come.  It would still be wise for Bitmain to release these kits as their are far more S2's in the world than S4's or S4+'s.  The hashing power of the kit was suppose to be around 2+ th when they first announced it.  And like beating a dead horse, would have been great back in the first quarter of this year.

In my area an Antminer S2 is highly sough after because we have low electricity rates and can make more money with an S2 then with one S3.

And I am sure in your area there are people who live in apartment buildings or get free electricity at work and would take the S2. You're best bet is to just sell them. You might be able to get $100-$200 for them. They sell like CRAZY on ebay and amazon.

If you want to thank me, buy some stickminers - of course made with Bitmain chips because those guys are alright.

It appears that I currently own two S2, one stock unit and one "kit" which I built a case for, so I've been looking forward to the prospect of an S2 Upgrade as well. The numbers are really promising with BM1384 - just like with the BM1380, high density low clock efficiency is about twice that of the high-clock low density of the S5. Maybe when they start moving an S7 with new chips around 0.35W/GH (I hope) they'll be more inclined to release a 0.3W/GH S2 upgrade. They've already demonstrated with the S4+ that they're willing to run out new designs with previous-gen chips; unfortunately they also have been demonstrating a willingness to sell things for much more than they're realistically worth (which is funny because they also sold out of everything even at those prices...) so I'd expect the upgrade kit to cost about as much as 2x S7 if the hashrate and efficiency are maybe 10% better than a pair of S7. Remember the S3 upgrade kit only cost about $40 less than just buying an S3. All the expensive-to-manufacture parts are being replaced but all the expensive-to-ship parts aren't, so they'll use that logic to keep customer savings marginal.

Even so, I'd buy one if they made one. It'd be pretty sexy. A GekkoScience-built upgrade kit is not impossible, but also not what I'm looking to work on until a few other projects are finished. Although... I just had an interesting idea so I'm gonna have to spend some time probing the backplane and see if it's even possible. Each board has specific data lines run to its socket and they're all multiplexed together on the control board. If I could use a pair per socket for USB signal (will require testing, noise and such being a big concern) I could cook up a new control board with a basic SBC attached via USB to some hub chips and still use a generic interface, just with the specific form-factor and backplane connection. If I'm thinking right, the WASP guys were looking into a USB-pairs backplane standard a while ago, and not because it was a bad idea. New boards could be built in the same shape with whoever's chips, with a standard USB converter chip to whatever base protocol the chips use and a few additional tie-ins like temp monitoring and voltage control. Updating cgminer on the control SBC would get new drivers for new boards, but the controller hardware itself wouldn't need to change at all. Hm... a lot of that would inherit from TypeZero design work so it might not be too bad to append something like that to the list of potential projects - unless anyone else wanted to work on it.

Once again, a very smart approach Sidehack.  Thank you for being willing to at least think about making an S2 upgrade option in the future.  Thank you for your support of the community, by actually coming up new miners.  Congratulations GekkoScience!

We're working on boards with a USB interface per board. This works well for using a board as standalone, but works just as well for using multiple boards in a larger unit. Avalon uses SPI, where Bitmain uses UART, but both have $1 USB converter chips readily available. If a standardized form-factor of board existed with such a ridiculously standard interconnect, piecewise upgrades and competition from multiple companies would be trivial. We're looking at S1/3/5 formfactor as a standard small machine to work with, and we're looking at boards the size of Prisma as a larger set which could easily be bladed into something S4-sized - and all of them would have a standard USB interconnect. Internal hub, simple controller, power cabling direct to the blade, done.

If we end up being able to do what we're wanting to do, and it turns into something we can continue with future generations, we'll continue building boards to the same formfactors. If other folks want to get in on that we can release the standards when it's ready.

Working with the S2 platform would be nice, but since it was sorta purpose-built for a specific power range and specific comm protocol, it's kinda limiting. I'd rather start from scratch to make a framework that can cover a more flexible range using fairly standard tools. I'm very much in favor of reusing the heavy parts of things, and I hope Bitmain shifts more toward that with the S[odd] series using a bit more generic/standardized controller connection and making upgrade hashboards available - in addition to the S2 upgrade path.

We need standardized designs for miners IMO, with a few different form factors.  1x USB stick, 1x Grid-seed sized-styled, 1x Asicminer Tube or S1/3/5 sized-styled, and 1x SP30/S2/S4 sized-styled.  I don't understand why the cooling, framework, and board communication can't all be re-used in further generations at this point, if the design parameters are taken into account.  That way once the miners are worthless, you can just scrap the boards and be back in business with the latest tech, without having to worry about re-selling redundant hardware, and trying to ship it.  That would be a huge benefit to the consumer. 

Also as a benefit to the manufacturers, when they are pricing new hardware they have to take potential ROI/income into account, and sell units for as much as possible so that the customer MAY have a chance to break even and make money on them.  If they did not have to fabricate, assemble and ship heat sinks, fans, controller boards or PSU's with every new miner, the savings could be partially added to the selling price to increase their mark-up while maintaining the same level of profitability to the customer.

Creating and distributing small form factor miners for every generation does not seem like an efficient process.  I can only imagine that Bitfury's hardware, since it is designed strictly to be used in mines, does not resemble typical miners like we have seen from SP-T or Bitmain whatsoever. I also feel like home mining would have a greater chance at survival if design elements are made more efficiently.

Because I may not like how they do comms. Why limit my own design to the framework of someone else's? I'd rather make something a bit more universal.

I'll try, it's better than nothing. 

Why not make a big miner with hashing boards completely compatible with S2's backplate?

You could talk a company into selling him lots of chips at a good price.   I think Bitmain was not really responding to chip orders unless it changed.

I hope when he does this big of gear he does ground up, like he did with usb drive.  I really would love to see a GekkoScience big miner that is not an upgrade.

Can anything help you to start sooner? May be any donations campaign? 

200MHz would be 3168GH proper. It's probably closer to 1100W wall once you do the math a bit better and factor in fans. But that's still pretty decent.

I did not realize it was 3T on 1 KW.   That is more impressive then I thought it was.   I could have seen them wanting to sell out of S4+ before releasing them if those specs are close.

But with S4+ sold out.  I'm not sure what they are really waiting for.   I think the silence still means a good chance they will not end up doing this.

It probably wouldn't, since I don't really have the resources to prototype something that large anytime soon. I also currently know almost nothing about the socket pinout or the protocol or the driver code. I'll worry about the four miner boards and USB hub and PSU boards and hosting I'm already doing, but if all that goes well I might look into it later this year?

sidehack, may be you can produce these boards? It will be faster anyway. 

No, it was prototyped over four months ago. If I'm thinking right, 8 of these boards clocked at 200MHz would get you over 3TH off 1KW from the wall.

It was possibly a design, but I don't think it made it farther then that.  I'm guessing it never made it to prototype phase.

I still think they are working on other things.  If they ever come back and do a S2 upgrade... doubt it.... but who knows.

They were shooting for 6 or 7 blades.