Video of (I assume) our very own Bitcoinorama 

http://www.engadget.com/2013/12/06/up-close-with-a-bitcoin-mining-powerhouse-video/

My thumb arrived too :-)
[New Zealand, so it wasn't too far]

It's currently hashing away on a raspberry pi (re-compiled cgminer as per instructions on the drillbit forum)

Many thanks to Barntech for all his hard work

Yes - the ASIC arena seems to have gone crazy.

As far as I can tell, you will never get ROI on any of the first generation ASICs (110nm avalon, etc).
You might get a return on the Bitfury (55nm), although it all depends on when (if) the next generation of 28nm chips turn up, and what their actual performance is.

And yes, the guys auctioning off gear will probably have a nice return in hand if they manage to sell it.

Thanks for the reply Puppet

I've kept digging - the penalty for a single mask asic isn't as big as I originally thought; it apparently ends up about 2x the size of a comparable standard cell asic.

So it's an interesting design approach, although I'm not sure that the window is large enough to get a return on investment - as it's still in pre-order, and there are more customised ASICs claimed to be arriving next month (KnC) and December (Cointerra). Plus Bitfury shipping 55nm product that seems to be performing well.

Overall, the whole asic space is crazy right now, but better than a soap opera.... grin

I've been having a look at VMC, and I believe they are real (mostly due to the posted document from eASIC).

eASIC are definitely real, and seem to be a leader in single mask ASICs. This is basically an ASIC where an initial design is done, producing a base layout, and the masks prepared for that. When you go to them, they will design you a single mask, which is just used for metal interconnects.
You end up somewhere in between an FPGA and a full custom ASIC - It seems to come out about 2x larger than a standard cell design, with consequent penalties on clock speed & power consumption.
On the plus side, it's got much lower engineering costs, and is fast to market.

Overall, it feels like a good short term strategy, but not sure it's going to still be competitive in the medium term (against things like the Cointerra chip).
(Credit to Puppet for some of the ideas here, in a post in the newbies forum)

The bit I don't understand with VMC is that their press releases suggest that their mining rigs are based on a single mask asic.

As I understand it, a single mask asic uses a standard base cell design and the mask sets up a metal layer which hooks the cells together so that they will perform the required function. The advantages should be much lower mask costs, and faster time to market - but the disadvantages seem to be that if you're competing with a full custom asic, it must feel a bit like the classic "turning up with a knife to a gunfight" situation. 

Does anyone know any more about this? Best case, won't the VMC chips have a much lower density, and so they will require more of them to equal the hashing power of Cointerra/KNC/etc?

Hi snot,

While the IBM Dataplex is a very good system, even with a room of them it is not worth using them for bitcoin mining.

The dataplex is fitted with Nvidia GPUs, which were never very good for mining (ATI was much better). And the CPU is no use for mining now, no matter how good.

Mining bitcoins is a competition, with a high reward (25 bitcoins = about USD$3325) for whoever finds a block which meets the requirement. So there are a lot of people now with very large systems specially built for mining. These use ASICs (custom silicon designed to mine bitcoins) and even the lowest specification asic (usb block erupter) can do 333MH/s which will be more than one Dataplex machine will be able to do.

In summary, you will be outgunned, and it is not worth trying to mine with a dataplex room.

Hi Byteme,

Most USB hubs aren't designed for a lot of maximum load devices (which block erupters are). If you think about average computer users, they are more likely to have lower load devices plugged in - mice, keyboards, flash drives and so on, which aren't drawing anywhere near the 2.5W limit. So hub manufacturers design for a lot less than 2.5W per port.

Also, in general, yes it's probably fine to supply more power to a hub, and then run more devices from it. That said, with the current bitcoin mining difficulty, a block erruptor is not earning a lot, and you are almost certain to spend more on a power supply than you will ever earn back...