PuertoLibre:  You seem to be confused on a great many details.  I'd be happy to refute your incorrect points if you want, so which is it:  Do you want me to leave this thread alone or continue to point out the ludicrous nature of both Tom's waffling and your incorrect assessment of details (that can be easily verified if you spent 10 seconds actually, you know, reading my posts)?

Tom admitted that he lied about BFL's power, good enough for me.  It's a shame that he continues to lie to his customers, but that's not really any of my business if he continues to hide the important details about his product.  He's fleecing people for their pre-orders, and that's sad, but there's not much I can do about that.

Here's something I want you to ask yourself:  Tom has announced that he's using Cell ASIC design.  He's not announced the process node, but lets be generous and say it's 45nm, since he has already stated that it was BGA based.  He's stated that he has licensed the SHA256 IP core from somewhere, but where?  If it's licensed from somewhere, has it been optimized for bitcoin's requirements? Perhaps it's a hardcopy from his "prototype" he claimed to have back at the beginning of September... but regardless of the source of his the IP core, it's going to be power hungry by the very nature of it being a CBIC, and again, being generous, lets say it uses only 30% of the power of the Modminer Quad.

Modminer quad: 40w @ 800 MH/s = 20 MH/w

30% of 40w = 12w per 800 MH/s or 810w for 54 Gh/s ... and I'm being pretty darned generous with the figures.  But heck, I'm a generous guy, lets cut that in half just for giggles!  405w for 54 GH/s.  Welcome back GPU's!  At least we've found a use for all the extra PSU's out there from the GPU mining era.  Of course, I'm just speculating, since Tom won't release any details, so it could turn out that my speculation is totally off base... but if not and I were Tom, I'd be keeping my power estimates secret, too.

How's that transparency working out for you, BTW?