Wrong. These manufacturers are already building 8x and 16x boards at _less_ cost per FPGA. Manufacturing largely has a problem where the cost of manufacturing a populated board effectively costs the same no matter how big the board is. In our case, a 16 FPGA board is not going to cost significantly more to manufacture than a 4 FPGA board.


Who said 64x board? I said 64 FPGAs in 4u. Thats four 16x boards in 4U.


Its an unusual design and connectors love to break off of boards. Its not a good idea and it costs money. There is the exact same risk in the even of failure: some of your FPGAs stop working.


Why are you so intent on essentially hacking this up like an idiot? Thats expensive to mass produce, stop that.

Lets try to figure out the total cost of making these the napkin way, and a quad costs $640 to buy.

My way:
per 16 FPGA/1U: $2310 for a 16 FPGA board, $90 for a Norco case that fits EEB and has 5 40mm fans, $50 for a Athena Power 1U 300w PSU
total cost per 4U: $9800 for 64 FPGAs, or $153 per FPGA
if something goes wrong: I lose 16 FPGAs minimum and maximum.


Your way:
Impossible to calculate because you're talking nonsense. What rack mountable case is going to fit 24 inches of cards? And don't say "just make a 8 FPGA 24 inch card" or something. The reason I specifically picked PCI-sized boards is because it is cheaper to produce PCI-sized boards due to the entire industry based around making them.

So, sure, lets use rainbows and unicorns and say you can fit 24 inch cards in cases that are somewhere around 23 to 31 inches deep inside without modifying the case (removing unused drive bays COSTS MONEY, changing the case design at all COSTS MONEY). And each one of those cards costs about, oh, $1220 to produce, and you can fit 7 cards in there, so thats $8540 for 56 FPGAs, plus who knows what for some controller board (lets say $100) because you just really really want one instead of just using SATA plugs like BFL did for the minirig.

per 56 FPGA/4U: $8540 for 7 56 FPGA boards, $100 for a board that does nothing but route serial connections, $70 for a Norco 4U case that fits EEB and has 2 80mm fans (and we need more than that so theres even more money wasted), and $160 for a NZXT Hale 90 750.
total cost per 4U:$8870 or $158 per FPGA.
if something goes wrong: You lose 8 FPGAs minimum, 58 maximum.

So not only do I get more density per 4U, my solution comes in cheaper.

You're trying to go for a modminer "blade server" kind of design, which isn't entirely appropriate either. Lets try this instead.

Lets say you use a 4U case (3U won't fit full height cards) and used full sized PCI-shaped cards and no backplane at all (power and USB/GPIO (say, using a SATA plug instead like BFL is doing) is on the end of the card like GPUs), and you put 4 FPGAs on the card. The card will have to be 107mm by 312mm (or less: many cases, even rackmount, won't fit a full length card; 6990s and 7970s are full length), so you're fitting the FPGAs in a straight line instead of a quad configuration.

Now, because of the heatsink requirements, you're going to have to fit these in double thick configurations (using, I assume, pure copper 2U northbridge heatsinks and forcing airflow through them from the front of the case), so you're putting around 16 FPGAs in a case (7 case slots, 1 overlaps with the case so it needs a 1 slot bracket)

Now, if you pack these in even harder using 1U northbridge heatsinks, that'd be 4*7, powered using 7 PCI-E 6/8 plugs, or 28 FPGAs in 4U.

Now, my 16 FPGA in a 1U or 2U case? In 2U, thats 32 FPGA per 4U, in 1U, thats 64 FPGA per 4U. I think my idea wins.

Quite a few rackmount cases will fit EATX (330x305), thats 4.18 times more area, so that could theoretically fit 16 FPGAs (assuming ATX fit 12), and it'll draw about 200 watts (assuming 4x Spartan 6 uses 50 watts with the new ~250 mhash bitstreams). You can't just fit 4x existing boards in manually because you can't fit 376x252 inside of 330x305 (and EATX is pretty cramped inside cases as it is), so they'll have to place stuff differently on a new board.

Huh, thats actually bigger than miniITX which is 170x170. ATX is 305x244, which could probably fit 12 FPGAs with room to spare.

Maybe one of the manufs will start making 12 FPGA boards with the four outer mount holes fit for ATX cases and power itself off a P4 (16a 12v) or EPS12v (28a 12v) plug?

There isn't enough bonds, but I think the larger bonds are growing to accommodate my plans. YABMC just launched a huge expansion project, for example.

I really doubt I can exchange all 200k shares this way. However, if someone wants to try, they are more than welcome.

I've recently answered this for someone else, so I might as well repeat it here.

Some people are viewing this premium as strictly "buy DMC for 2/3rds the cost", which is not entirely true. Up front, yes, you buy the DMC shares for that, but you give up two short term abilities: dividends (they are cut in half as per the 50% dividend/50% growth agreement in the contract), and the ability to sell (there is not a huge demand for DMC shares yet as the company is still new). If you factor in all the "lost" dividends until DMC dividends increase past what you would have originally gotten from the original bonds, it evens out in the short term.

Also, there is this other issue where if they keep buying shares to sell to me, the effective premium I paid/pay on past/future shares goes down because value increase. Not that I'm complaining, mind you, it'd help everybody in the Bitcoin world if bond prices started rising.

Still, theres a fair bit to go on that. If people want to go buy 1000 shares of something right now and trade me, it might not effect some of the bonds too much and still be worth it for them and still worth it for me.

Sounds like the version of Bitcoin ecoinpool is running is dangerously out of date.

Wow, DMC is really taking off here.

We've passed the 1000 share mark, and we're not slowing down.

Also, I want to mention this so it doesn't get lost in the shuffle: someone traded exactly 666 shares of BMMO in. I love it when people totally get me.

https://bitcointalk.org/index.php?topic=77469.msg901464#msg901464 is acting as a live tally on the asset trade. I'm not going to leave this window open forever. Once Bitcoinica's destruction stops effectively halting GLBSE and everything is flowing the way it should again, the window gets closed and everyone has to buy shares the more expensive way.

Don't wait, act now, orders in the next 20 minutes get free S&H, etc etc. You know the drill.

Hey Yohan,

I know this may sound like a strange question, but is there any mass produced case that would fit the quad board? It looks to be roughly the same size as a PC104 board, but I doubt any of the mounting holes match.

So THATS what ZipConf is. Okay.

For example, mtgox makes several thousand transactions a day. I wouldn't want to manually handle that.

By popular request, I've added TYGRR and TYGRRMBA

Well, with any site that needs to send bitcoins back out you need whats known as a hot wallet, ie, a wallet that ONLY has enough to do day to day business... you setup your software to send excess coins to a cold wallet (offline or otherwise hidden on another machine), and message you if you need to manually transfer from cold to hot.

No one has $90k worth of coins in their hot wallet.

Its not a Delta AFB1212. There is no reason to EVER use non-enterprise/industrial fans in a computer.

http://www.delta.com.tw/product/cp/dcfans/download/pdf/AFB/AFB120x120x38mm.pdf

They make ones that are slower/quieter than what most people are used to from Delta (although, if you're worried about noise, taking a faster than and undervolting to 5v is often superior, as is ramping voltage on chip temp; it also reduces wear greatly).

No, virtually every secure server uses those concepts.

This is what I do: I disable root login locally AND through ssh (local root disable is borderline security through obscurity though, you have to know the l/p of people who can su before you can cause any real damage), and I disable ssh login with passwords (key only), and I also use keys with passphrases only (so even if they steal the key from me physically, they can't use it).

Installing a firewall (ie, a handful of iptables rules) is somewhat overkill, you should only be running software that can be told to listen on specific IPs (ie, listen locally only). iptables kills network performance, so only use it if you must.

Running the actual httpd/db/maild/etc on a vm that itself is stored on an encrypted image is rather popular. You can boot the machine and do basic maintenance on the VM (ie, back up everything), but you can't start it or read the image unless you have the passphrase. Its basically a ghetto way of doing LOM cheaply, especially if your VM has its own IP and the host isn't masquerading for the VM.

Bare metal over VPS only stops one single attack: host reading VM memory. Which, for financial things, is worth it for some people.

Having your own hardware coloed instead of leasing a dedi is not going to increase security: others still have causal physical access to your rack, so you'll need to rent your own locked cage if you want to take paranoia the whole 9 yards.

Ive added ABM to the swap list.

You keep using that word. I do not think it means what you think it means.

Due to popular request I've added FPGA.Contract and YABMC for the trade offer.

A lot of people are requesting this feature, and I hope you think about this for future products: use programmable VRMs. A lot of people want to undervolt their FPGAs to increase efficiency at the end of their effective mining lives and it would increase sales over the BFL juggernaut.