Quote from: erk on September 23, 2013, 08:07:39 PM

Quote from: -Redacted- on September 23, 2013, 07:58:42 PM

Quote from: markm on September 23, 2013, 07:55:23 PM

Quote from: erk on September 23, 2013, 07:50:53 PM

Quote from: rizzman on September 23, 2013, 06:34:16 PM

Dude, your question shows exactly how far back you read... The whole reason this shit about bitfury cropped up is because people were claiming ridiculous numbers on the efficiency of a 65nm chip which for all intensive purposes should not have been possible and then transformed into an argument over whether or not the BF chip was more efficient than KNC's. That was the point he was getting at I believe, people where claiming some crazy shit, with out any proof to back it up...
...

The KNCminer specs from there own data is Energy consumption: 1.6 W/Gh/s. the Bitfury is 1.0W/GH/s what's so hard about that to understand? The Bitfury is more energy efficient, though they both are pretty good.

The KnC 1.6 is at the wall though, right? What is bitfury's at the wall?

-MarkM-

No one will actually know until KNC does at-the-wall tests on a real unit hashing.

My two bitfury cards pull ~ 46 watts at 66 Gh/s

Is that each of combined?

It wouldn't surprise me if a 55nm full custom design turned out to be more power efficient than a 28nm standard cell design. I would assume that it's possible to under clock the KNCminer chip to get the Watts/GH/s down. Looking at the cooling hardware they have used, you can see the KNCminer chip is extremely overclocked, they are trying to squeeze the most out of it at the expense of efficiency. It would have made more sense to use twice as many chips at half the clock rate to get the power usage down, but that means they could only build half the units per wafer. In the end the machines left mining will be the ones that use the least Watts per GH/s. There is no room in mining for overclocking.

I suspect it is both. I have two bitfuries running 65-66GH/s at 88W at the wall, that includes 40W monster AC fan.