Avalon Batch#4 Supporting Evidence

[greaterninja]

I stand corrected, it is 110nm packaging structure...coulda swore it use to be 135nm...ah well..you are right

[1715633389]

1715633389

[1715633389]

1715633389

[dogie]

Agian, more FUD from greaterninja. This is the guy who's worked in IT for 500000 years btw, thinks cpus are 32nm xD

[titomane]

Agian, more FUD from greaterninja. This is the guy who's worked in IT for 500000 years btw, thinks cpus are 32nm xD

there is nothing more stupid than other error laugh. When this has already recognized.
If my nephew does. I dont like, I could understand. But you have 30 years minimum.

[Vycid]

Only thing that really makes sense is for bitsyncom to license the Bitfury multi core chip and reference design at this point. 0.39W/gh is hard to beat (if tests are conclusive) and since they provenly can build and deliver PCB etc. and have the (still somewhat) reputable Avalon name, they could make a killing selling basically Bitfury clones.

Metabank say 0.7w per Ghs
https://metabank.ru/asic

Where has seen 0.39w/ghs?

Thanks

https://bitcointalk.org/index.php?topic=228677.msg2515472#msg2515472

This is absurd.

Produces a chip that can operate at 5Ghs, manufacturing a device with 24chips and 120GHs.
But for the device consumes less you get a device with 24Ghs.
It's like buy a Ferrari and driving without gaining 3000rpm, consume less. But it is a waste.
Do you hashrate highly stable? or lower power consumption?

You should know that to have a 120Ghs device. Have 0.39w/ghs performance requires a device with 120 chips. This product more expensive above $ 18 / Ghs to over $ 60 / Ghs that if you consume only 50W.

actually if they reduce the package size, the asic chips can hash more and use less power.   Avalon has a 135nm package size and it does 280-400mh/s    .  I believe BFL has a 65nm package size and it does 5.9 GH/s per chip (I could be wrong).     KNCminer is working on a 28nm ASIC which may very well do ~60 - 100 gigahash/second per chip.

135nm is the feature size, not the package size, doofus. You'd need a scanning electron microscope to find a 135nm chip.

(actually, their feature size is 110nm, but nevermind...)

110nm isn't the size of the chip is the size of transistor.

Erm, this isn't right either. The process node refers to the half pitch of a memory cell.