ASICMINER: Entering the Future of ASIC Mining by Inventing It

[ning]

avalon is launching their gen 3 chip in about 1.5 days.

40nm, 7GH/s per chip, 5Watts, 0.775Watts per Gh/s
http://avalon-asics.com/

How can we compare these? How to compare J/Gh with W/Gh, anyone? thanks

ASICminer : 40nm,   12.8GH/s per chip,   0.72V,   0.35J/GHash

AVALON :    40nm,   7GH/s per chip,        5Watts,  0.775Watts per Gh/s

from google >
Joules to watts calculation formula
The power P in watts (W) is equal to the energy E in joules (J), divided by the time period t in seconds (s):
P(W) = E(J) / t(s) >>> watt = joule / second

DC volts to watts calculation formula
The power P in watts (W) is equal to the voltage V in volts (V), times the current I in amps (A):
P(W) = V(V) × I(A)

They are not the same (1J = 1W*s), but here they mean the same thing.

Loosely (not in the context of physics exams), J/GHash and W/GHash can be used interchangeably as energy consumption measurements.

[zefir]

They are not the same (1J = 1W*s), but here they mean the same thing.

Loosely (not in the context of physics exams), J/GHash and W/GHash can be used interchangeably as energy consumption measurements.

In fact it is also valid in the context of physics exams: since 1J=1Ws, W/GH/s is equivalent to J/GH

There is a very detailed post from DaT on the forums going into details (can't find it just now), but yes: AM's rev3 is at least twice as energy efficient as Avalon's rev3 (if expected values can be reached).

[Mabsark]

In fact it is also valid in the context of physics exams: since 1J=1Ws, W/GH/s is equivalent to J/GH

There is a very detailed post from DaT on the forums going into details (can't find it just now), but yes: AM's rev3 is at least twice as energy efficient as Avalon's rev3 (if expected values can be reached).

The confusion arises from people using the wrong units - "Gh" instead of "Gh/s"

[RoadStress]

avalon is launching their gen 3 chip in about 1.5 days.

40nm, 7GH/s per chip, 5Watts, 0.775Watts per Gh/s
http://avalon-asics.com/

How can we compare these? How to compare J/Gh with W/Gh, anyone? thanks

ASICminer : 40nm,   12.8GH/s per chip,   0.72V,   0.35J/GHash

AVALON :    40nm,   7GH/s per chip,        5Watts,  0.775Watts per Gh/s

from google >
Joules to watts calculation formula
The power P in watts (W) is equal to the energy E in joules (J), divided by the time period t in seconds (s):
P(W) = E(J) / t(s) >>> watt = joule / second

DC volts to watts calculation formula
The power P in watts (W) is equal to the voltage V in volts (V), times the current I in amps (A):
P(W) = V(V) × I(A)

EDIT : ning, thank you ! so ASICminer *theoretical* specs are at least 2x better than AVALON offering. Lets see if Friedcat is able to deliver.

Bitmine.ch also announced 0.3J/GH, but failed to deliver. I don't want to start any speculation, but it seems very very low and i don't think it's achievable right now. Time will tell.

[kaerf]

avalon is launching their gen 3 chip in about 1.5 days.

40nm, 7GH/s per chip, 5Watts, 0.775Watts per Gh/s
http://avalon-asics.com/

How can we compare these? How to compare J/Gh with W/Gh, anyone? thanks

ASICminer : 40nm,   12.8GH/s per chip,   0.72V,   0.35J/GHash

AVALON :    40nm,   7GH/s per chip,        5Watts,  0.775Watts per Gh/s

from google >
Joules to watts calculation formula
The power P in watts (W) is equal to the energy E in joules (J), divided by the time period t in seconds (s):
P(W) = E(J) / t(s) >>> watt = joule / second

DC volts to watts calculation formula
The power P in watts (W) is equal to the voltage V in volts (V), times the current I in amps (A):
P(W) = V(V) × I(A)

EDIT : ning, thank you ! so ASICminer *theoretical* specs are at least 2x better than AVALON offering. Lets see if Friedcat is able to deliver.

Because AM does not have final specs yet. That's why I asked the question. There have been a lot of theoretical performance numbers spouted by chip makers, but what counts is what the real world numbers are. Also, chips that are available sooner are "better" than undelivered chips...again that's the reason for the question.

[minerpumpkin]

Do we actually know the (probable) foundry AM is using for gen 3 or has been using for gen 1?

[bitcoin.newsfeed]

Do we actually know the (probable) foundry AM is using for gen 3 or has been using for gen 1?

why're you asking? this should be internal company info and know&how

[minerpumpkin]

Do we actually know the (probable) foundry AM is using for gen 3 or has been using for gen 1?

why're you asking? this should be internal company info and know&how

I'm simply interested, nothing more. Really.
Getting to know more about our favourite mining company while we're in the waiting room. There are companies where the respective foundry is known, IIRC. For example KNC or Avalon, isn't it?

[Ditto]

The ASICMINER winter has been a long one... really looking forward to spring...

Aren't we all? Here's hoping our favorite grilled feline will finally deliver.

[ning]

Do we actually know the (probable) foundry AM is using for gen 3 or has been using for gen 1?

According to this [1] document, gen1 was produced by a fab in Shanghai, China.

And gen3 is produced by TSMC [2].

[1] http://cseweb.ucsd.edu/~mbtaylor/papers/bitcoin_taylor_cases_2013.pdf {page 8}
[2] https://docs.google.com/document/d/1_dmXzyMqtr7tVZukk2DAJih2sRq5jDy5DQPkwCce570/edit?usp=sharing

[ning]

http://avalon-asics.com/

A timer doesn't necessarily mean anything, this one above had been reset a few hours ago to allow one more day.

[He1l_Q]

latest update from friedcat

Update Mar. 18
bonding list: https://docs.google.com/spreadsheet/ccc?key=0Al1fvFT7Sd5bdFFZSUllNW5seVRfM01kcDVqZkJfdFE&usp=sharing
data sheet: https://docs.google.com/document/d/1_dmXzyMqtr7tVZukk2DAJih2sRq5jDy5DQPkwCce570/edit?usp=sharing
chip top view: https://drive.google.com/file/d/0B11fvFT7Sd5bWjh6MmJsMFpCYkU/edit?usp=sharing
chip bottom view: https://drive.google.com/file/d/0B11fvFT7Sd5banZkMDZqSTBGTEE/edit?usp=sharing
spec of size: https://drive.google.com/file/d/0B11fvFT7Sd5bUV90cFN4cEhTNEk/edit?usp=sharing

The package is based on the engineering batch of 9mmx9mm. Part of our chips might be changed to 8mmx8mm or 7mmx7mm in mass production batches.

source: https://bitcointalk.org/index.php?topic=438359.160

[Zubilica]

you mean :

https://bitcointalk.org/index.php?topic=438359.msg5759952#msg5759952

[He1l_Q]

you mean :

https://bitcointalk.org/index.php?topic=438359.msg5759952#msg5759952

yes, you're right, and thanks

[dexX7]

For some reason the post disappeared. Maybe it was an intentional leak?

Here is a mirror:

AM BE200 datasheet:
https://www.dropbox.com/s/o87zbble4z5fkhq/AM%20BE200%20Datasheet%20-%20Google%20Drive.pdf

Bonding:
https://www.dropbox.com/s/57mnfscjfsti29a/bonding.pdf

pin_bot.jpg, pin_size.jpg, pin_top.jpg:
http://imgur.com/gjaOoF4,dB6sXBR,iHIqMSJ

[Evolyn]

Another mirror:
https://bitcointa.lk/threads/ann-asicminer-publicly-looking-for-potential-customers-partners-for-new-chips.248792/page-10

Yes, looks like a phishing link

[szmarco]

Maybe he is preparing a more detailed release.

[bitcoin.newsfeed]

Code:

The typical hash power is ~12GH/s in rated mode.

looks ok to me !

but other specs are spanish village for me, maybe someone with required knowledge in chip design can confirm if 0.72V, 0.35J/GHash are met?

[Zubilica]

Havelock will be out for a wile , untill DDoS'ers will get their cheap shares. Popcorn and waiting for the next update form Grilled Cat.

[minerpumpkin]

This is, again, surreal. There's the anticipated (again removed) update and everyone's like "Yeah... Looks good. This is it." Even Havelock is rather quiet.