Topic: Mini Rig announcement by Butterfly Labs - 25gh/s - page 10. (Read 34251 times)

I believe I could. Much larger heatsinks, (perhaps the height of where the individual fan would be if removed) and toss the individual fan on top of each "super single" (I like that term ). Same cross-ways fan arrangement, but less turbulence and localized heating. The case fans may need to be a little higher velocity.

I went and looked at the dimensions in the first post, and it looks like in its current form it will take up 7U of rack height, but it is 10 inches wide which would make it a tight squeeze to get 2 of them beside each other in a standard 19" rack. Perhaps it would be best to ignore potential improvements here though, because I don't want to delay anything by causing a redesign halfway through. However, for the future SuperComputer, if it isn't watercooled, the best thing to do would be to make it fit in a rack as best as possible. That could mean specially ducted airflow, offset processors (to prevent waste heat from cooking another processor), standard width, (no shelf needed), and no more than 4U height.

Take for example a 2U server with quad 90 watt processors: They have fairly large heatsinks, and none of them are in line with each other. Although these might have to be stacked several deep, an offset arrangement would prevent the heat from travelling linearly as it does now, cooking the rear devices.

Hi D&T,

Can you arrange the boards within the enclosure to make the airflow more efficient and share it with BFL?

Thanks,
gigavps

The good news is that everything indicates 1 RigBox = 18 independent "super singles".  So custom cases and cooling would be possible.  With mounting holes on the boards finally (yes this applies to all FPGA designers releasing a board w/ no mounting holes is asinine) even watercooling would be possible. 

Still I agree the airflow seems .... inefficient (I am trying to be more nice).

I wouldn't see why immersion cooling wouldn't also be an option.

Classic rule of three, any scheme that requires more than 3 different things to happen will never work in real life.

That said, kudos for considering it anyway. I'd probably take a shot if I had that sort of cash lying around.

I gotta say, looking at that picture, I cant believe how poorly you thought out airflow.  There is not nearly enough clearance for the fans and no clear in/out stream of air, its just going in all directions, and the fans are mostly blocked... by other fans.  This is really just a case of "lets slap more fans on it and it will work". Its almost as bad as the single where you added fan blowing to the back of the PCB.

How hard could it have been to arrange those boards facing each other, rather than just stacked,  and create  wind tunnels front to back between them? Im pretty sure a decent design would allow lower temps with less than half that amount of fans.

- Need excitement, holding bars of Gold is boring.
- Bitcoin profit calculator shows very promising results
- I trust BFL more than Largecoin as BFL already has working products, and LargeCoin's 25Ghash@100watt is hard to believe.
- Once the box breaks even, at worst case I can ship it to a friend in Ukraine with electricity cost of $0.05, with current level it would cost $0.01 to produce a Bitcoin, so even at $0.05/BTC it should be profitable.
- Import tax is payable and unavoidable anyway, if you buy from abroad you pay it, if you buy from local shops then it is already paid and added to the price.

My other option is to spend $15k and build a mining rig with HD7970s or HD7990s, but with electricity prices I would have to leave UK and do it in Ukraine, which I don't really want.

Your power and USB cables are going to run into the far side fans.

Maybe offer to pay a non-refundable 7500$ per minirig to secure parts and the rest when product is near ready ?

Leaving 15k$ sleeping for months is hard to live with for us opportunist Bitcoiners.

That, right there, is the only thing that makes sense to me. I'd love to have a shiny $15k box that pushes 25GH!!

But I am not going to send BFL $15k for 12-15 weeks up front based on a 3D image and a couple of numbers. I've already sent them $5k down payment once for a product that will never see the light of day (the big brother of the Mini Rig). And they still haven't shipped their January orders of the single as far as I can tell.

Who knows if they won't pull the same stunt with the mini rig and cancel it once they realize it's not possible just like with the Rig Box? Or who knows if this whole thing is not just here to mess with LargeCoin or others? Who knows Sonny's real name and place of business? And their competition is certainly not sleeping.

Spending $599 via PayPal alright, but now we are at $15 bank payment upfront or Bitcoin only and their track record isn't really the best when it comes to keeping promises and releasing new products.

Again, the only thing that speaks for BFL is that they have a product shipping. But if you take the single release as reference you just have to expect trouble down the road going from a 3D image to a real product.

Sorry, but that's the way I see it.


My guess, it's based on computations based on 5 fingers and a tip calculator. Correct me if I am wrong 

Is this based on a simulation of the new chip, or on an actual real board?

Your probably right in the risk of all the ifs, but where is the fun without the risk and the shiny 15k box to play with.

Since everybody is speculating, let me jump on the bandwagon.

If I am not wrong then ice_chill is in the UK and his electricity cost is probably 3 or 4 times more than the quoted $0.07 (and indefinitely more than $0.00).  

He'd have to pay a few hundred US$ shipping cost to the UK for his Mini Rig.

He'd have to smuggle the Mini Rig to the UK in order to avoid a $3202 (21% VAT) charge of import duties.

A $3202 crime (smuggling/undervaluing shipment etc) is not just a slap on the finger if caught, so he is probably going to do the right thing and support the queen.

He is selling his 24K gold in order to buy a Mini Rig and pay the shipping/taxes.

His strategy" depends on many IFs

- if the price of BTC goes up
- if BFL delivers before August
- if the Mini Rig does what was promised and just works
- if the gold wouldn't have done better
- if the BTC goes up after block reward is halved

He legitimates this purchase with the fact that he lost $10k on the Forex market and hence it's no big deal. I am wondering if he didn't lose his $10k because it his "business strategy" looked somewhat like his new master plan to increase his wealth by using a BFL Mini Rig.

If this is true, then wouldn't it make more sense for ice_chill to just buy BTC for his gold and lean back and enjoy the show if he is already betting on so many IFs?

Watching this thread.

I'd like to know more about the production ramp up of the Singles if possible.

You would be, but thats the gamble that they would be going for. Or betting on the value of the bitcoin to increase once the block reward is halved.

Hmm, difficult to decide now.

Mini Rig US$ 15k 1250Watt 25GH 12-15 weeks (real weeks or BFL weeks?)
LargeCoin US$ 15k 100Watt 20GH 12-15 weeks (July)

What speaks for BFL is that they have a product shipping, although it looks like they'd need a lot of luck to actually deliver what they promise and in the promised time frame.

What speaks for LargeCoin is the persons behind it are known with job history and they don't seem to be total nut jobs (although I have claimed at one point in time that they are clowns, when they dropped the price from $30k to $15k within a few days after announcing the product, lol).

Don't understand this wrong BFL. I love your single and it's doing a great job for all the people that have one.




At $1/BTC it would take 7.2 years to break even with electricity of $0.07/kWh. At $0 electricity it would take 5.3 years. At $5/BTC it would still take a bit more than a year at $0 electricity cost.

(assuming difficulty at 1600000 and 25 BTC/block)

Am I missing something here or do I need a coffee first (I just woke up)?

If you calculate with 4-5 months (Aug, Sep, Oct, Nov) at 50 BTC/block you are really gambling on BFL being able to pull this off in the next 3 months and ship them out the door, aren't you?

In business it's called having an edge if everyone knew their secret then they would not be as competitive anymore.

If they get the chips produced,  how are they getting such a good deal on them?  I would think if they were not readily stocked items they would come at a premium. 

Fixed.

18, not 16.

I don't anticipate them selling us at least 18 new 28 nm FPGAs that perform better than the single.

Current speculation is that the single uses two $7000 chips or something like that, the reason they are cheap is because of age. If they were to use new chips, we'd be looking at a large fortune for the box.


Edit: The weird thing is that said the lead time is longer because the chips are made to order.

I'm not really sure how BFL works its magic, but I really like the speculation around it.