[ActiveMining] The Official Active Mining Discussion Thread - page 70.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: kleeck on September 04, 2013, 07:44:17 PM

"Survived" is putting it lightly...

http://www.easic.com/easic-rated-third-fastest-growing-semiconductor-company-in-north-america-on-deloittes-2012-technology-fast-500/

http://www.easic.com/easic-wins-alwayson-ao-100-top-private-company-award-for-its-innovative-structured-asic-technology/

"Notability: - Over 100 Design Wins for Nextreme means that a large Number of Engineers are working with the devices - EDN Innovation Award Finalist - the award is very significant and should count towards meeting the notability criteria even if the company was 'only' finalist and did not win the award - Offers a new approach to ASICs and is therefore of general interest to engineers"
(Source: http://en.wikipedia.org/wiki/User:Bonf/EASIC)

just to display a bit of the eASIC prowess.

Yeah, there's no doubt eAsic is in a very different league than anyone else doing bitcoin chips.

kleeck

hero member

Activity: 532

Merit: 500

https://karatcoin.co

Quote from: Vbs on September 04, 2013, 07:41:35 PM

Quote from: Ytterbium on September 04, 2013, 07:38:02 PM

Quote from: Vbs on September 04, 2013, 07:36:30 PM

Quote from: Ytterbium on September 04, 2013, 07:31:01 PM

Not exactly, You won't be able to sell, say, 900k easicopy chips the day after your nextreme chips come out of the fab. there will be a whole other development process, possibly taking months, before you can sell orders. Companies like Cointerra/HashFast/ and KnC will be able to sell chips/modules with a much shorter turnaround time, which is pretty critical in the bitcoin space.

eAsic takes a different and very efficient approach to minimize design time on Nextreme to standard-cell conversion.

That's nice. Unfortunately "minimized" is not a very concrete amount of time.

True, but you are also paying a premium for their expertise. These are not guys that have the luxury of simulating one stuff and delivering another very different. They have survived on the market since 1999 for a reason.

"Survived" is putting it lightly...

http://www.easic.com/easic-rated-third-fastest-growing-semiconductor-company-in-north-america-on-deloittes-2012-technology-fast-500/

http://www.easic.com/easic-wins-alwayson-ao-100-top-private-company-award-for-its-innovative-structured-asic-technology/

"Notability: - Over 100 Design Wins for Nextreme means that a large Number of Engineers are working with the devices - EDN Innovation Award Finalist - the award is very significant and should count towards meeting the notability criteria even if the company was 'only' finalist and did not win the award - Offers a new approach to ASICs and is therefore of general interest to engineers"
(Source: http://en.wikipedia.org/wiki/User:Bonf/EASIC)

just to display a bit of the eASIC prowess.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: Ytterbium on September 04, 2013, 07:38:02 PM

Quote from: Vbs on September 04, 2013, 07:36:30 PM

Quote from: Ytterbium on September 04, 2013, 07:31:01 PM

Not exactly, You won't be able to sell, say, 900k easicopy chips the day after your nextreme chips come out of the fab. there will be a whole other development process, possibly taking months, before you can sell orders. Companies like Cointerra/HashFast/ and KnC will be able to sell chips/modules with a much shorter turnaround time, which is pretty critical in the bitcoin space.

eAsic takes a different and very efficient approach to minimize design time on Nextreme to standard-cell conversion.

That's nice. Unfortunately "minimized" is not a very concrete amount of time.

True, but you are also paying a premium for their expertise. These are not guys that have the luxury of simulating one stuff and delivering another very different. They have survived on the market since 1999 for a reason.

Ytterbium

full member

Activity: 238

Merit: 100

Quote from: Vbs on September 04, 2013, 07:36:30 PM

Quote from: Ytterbium on September 04, 2013, 07:31:01 PM

Not exactly, You won't be able to sell, say, 900k easicopy chips the day after your nextreme chips come out of the fab. there will be a whole other development process, possibly taking months, before you can sell orders. Companies like Cointerra/HashFast/ and KnC will be able to sell chips/modules with a much shorter turnaround time, which is pretty critical in the bitcoin space.

eAsic takes a different and very efficient approach to minimize design time on Nextreme to standard-cell conversion.

That's nice. Unfortunately "minimized" is not a very concrete amount of time.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: Ytterbium on September 04, 2013, 07:31:01 PM

Not exactly, You won't be able to sell, say, 900k easicopy chips the day after your nextreme chips come out of the fab. there will be a whole other development process, possibly taking months, before you can sell orders. Companies like Cointerra/HashFast/ and KnC will be able to sell chips/modules with a much shorter turnaround time, which is pretty critical in the bitcoin space.

eAsic takes a different and very efficient approach to minimize design time on Nextreme to standard-cell conversion.

Quote

http://www.easic.com/migration-to-cell-based-asic/migration-to-cell-based-asic-simple-design-flow/
The starting point for most easicopy designs is a eASIC Nextreme or Nextreme-2 design. This enables eASIC engineers to leverage much of the work that has already been done from a design that is already successfully in production. In addition, this helps to reduce the overall time to production for the easicopy design.

The easicopy design flow is shown below. At the front end it requires a eASIC Nextreme or Nextreme-2 synthesized netlist and an SDC timing constraints file. After initial synthesis, the design is taken through a traditional cell-based ASIC flow by eASIC engineers. This includes Design For Test (DFT) insertion and synthesis, and then back-end physical implemention which includes floorplanning, I/O ring design, power mesh design, timing driven place and route, timing closure, parasitic extraction, final STA, and tapeout readiness.

eASIC engineers have extensive experience in converting FPGA designs to via-programmable eASIC Nextreme or Nextreme-2 NEW ASIC, and then eASIC Nextreme or Nextreme-2 NEW ASIC designs to cell-based, easicopy ASIC.

Quote

http://www.easic.com/migration-to-cell-based-asic/migration-to-cell-based-asic-risk-mitigation/

Design Risk Mitigation - having successfully converted an FPGA prototype design to eASIC Nextreme or Nextreme-2, eASIC engineers have already made a number of changes that will make the conversion to an easicopy ASIC simpler. In addition your designers will work with the same eASIC engineers that have already helped you to successfully take your eASIC Nextreme or Nextreme-2 design to production.

Ytterbium

full member

Activity: 238

Merit: 100

Quote from: Vbs on September 04, 2013, 07:25:15 PM

Quote from: Ytterbium on September 04, 2013, 07:18:57 PM

Quote from: Vbs on September 04, 2013, 07:11:13 PM

Least of all, let's not forget VMC also sells chips in bulk, so there's no money to spend on "blank PCBs" there.

Someone has to spend the money on those PCBs. And the point of the chart is that, overall, it will be cheaper to stick with the regular ASICs until you get up to 300k units (per whatever time period they're estimating), it would just be a waste of money otherwise. Obviously, if the power/wattage is better then that will be an additional consideration.

The point of the chart is to compare chip unit price with sales volume, meaning that VMC can/will have access to chip prices that can be competitive with anyone else. This was the original issue you raised, unit price.

Not exactly, You won't be able to sell, say, 900k easicopy chips the day after your nextreme chips come out of the fab. there will be a whole other development process, possibly taking months, before you can sell orders. Companies like Cointerra/HashFast/ and KnC will be able to sell chips/modules with a much shorter turnaround time, which is pretty critical in the bitcoin space.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: Ytterbium on September 04, 2013, 07:18:57 PM

Quote from: Vbs on September 04, 2013, 07:11:13 PM

Least of all, let's not forget VMC also sells chips in bulk, so there's no money to spend on "blank PCBs" there.

Someone has to spend the money on those PCBs. And the point of the chart is that, overall, it will be cheaper to stick with the regular ASICs until you get up to 300k units (per whatever time period they're estimating), it would just be a waste of money otherwise. Obviously, if the power/wattage is better then that will be an additional consideration.

The point of the chart is to compare chip unit price with sales volume, meaning that VMC can/will have access to chip prices that can be competitive with anyone else. This was the original issue you raised, unit price.

Ytterbium

full member

Activity: 238

Merit: 100

Quote from: Vbs on September 04, 2013, 07:11:13 PM

Least of all, let's not forget VMC also sells chips in bulk, so there's no money to spend on "blank PCBs" there.

Someone has to spend the money on those PCBs. And the point of the chart is that, overall, it will be cheaper to stick with the regular ASICs until you get up to 300k units (per whatever time period they're estimating), it would just be a waste of money otherwise. Obviously, if the power/wattage is better then that will be an additional consideration.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: Ytterbium on September 04, 2013, 06:58:25 PM

Quote from: Vbs on September 04, 2013, 06:45:21 PM

Quote from: Ytterbium on September 04, 2013, 06:37:44 PM

What we really need to know is the die size.

It's possible Labcoin's 65nm chip may take up less space, making their chips potentially more profitable.

If we're talking about futurology, we can also talk about VMC going for standard-cell 28nm, using eAsic's easicopy process and getting chips at pitiful costs (price is y-axis below). Then it's game over.

As you can see on the chart, the cutoff for 'profitability' of using the easicopy instead of nextreme is 300-900 thousand units. At 25Gh/s, that comes out to:

7.5-22.5 petahash.

That's the threshold for for it to be profitable for ActM to switch to easicopy.

Assuming he can get 10 chips per board, that's 30-90k boards. At $25/board, that comes out to $750k-$2.2 million for enough boards to put 300-900k chips on.

That means in order for it to be profitable for ActiveMining to go to the easicopy rout to get standard cell asics, he would need to be making an so large he'd have to spend $750-2 million just on the blank PCBs to put them on.

Not impossible. But it's a long way off from what I can tell.

Your overall analysis is wrong. You are analyzing it as if the transition should only be based on volume to lower unit price. It won't be. The point where it's profitable to go for easicpy is not just based on unit cost, but on whatever possible gains in either hashrate or power consumption from using a standard-cell ASIC. The bigger the gains, the quicker the transition will be done.

Least of all, let's not forget VMC also sells chips in bulk, so there's no money to spend on "blank PCBs" there.

crumbs

full member

Activity: 210

Merit: 100

Quote from: N_S on September 04, 2013, 07:00:34 PM

Quote from: crumbs on September 04, 2013, 06:58:41 PM

To the MoooN Cheesy

Oh come off it. It went to the moon....but just popped in for a bite and then decided it had enough.

And it will again, but this was so sad... An ideal opportunity to prove the doubters wrong. All people had to do was *hold*. There was no market on burnside's gameserver. The price would have went through the roof -- but no.
People started tripping all over each other for a chance to get out Sad

Confidence inspiring.

N_S

full member

Activity: 238

Merit: 100

Quote from: SoylentCreek on September 04, 2013, 07:03:09 PM

See that little "Ignore" button under his username? Click it.

Oh you mean that box that looks like the color of my piss after heavy kidney damage?

SoylentCreek

sr. member

Activity: 266

Merit: 250

Quote from: N_S on September 04, 2013, 07:00:34 PM

Quote from: crumbs on September 04, 2013, 06:58:41 PM

To the MoooN Cheesy

Oh come off it. It went to the moon....but just popped in for a bite, ya know?

See that little "Ignore" button under his username? Click it.

N_S

full member

Activity: 238

Merit: 100

Quote from: crumbs on September 04, 2013, 06:58:41 PM

To the MoooN Cheesy

Oh come off it. It went to the moon....but just popped in for a bite and then decided it had enough.

SoylentCreek

sr. member

Activity: 266

Merit: 250

Quote from: Vbs on September 04, 2013, 06:45:21 PM

Quote from: Ytterbium on September 04, 2013, 06:37:44 PM

What we really need to know is the die size.

It's possible Labcoin's 65nm chip may take up less space, making their chips potentially more profitable.

If we're talking about futurology, we can also talk about VMC going for standard-cell 28nm, using eAsic's easicopy process and getting chips at pitiful costs (price is y-axis below). Then it's game over.

Would the ELI5 on this chart basically be that at some point the chips are going to cost next to nothing to produce, therefore any difficulty increases will easily be met via the production of more chips?

crumbs

full member

Activity: 210

Merit: 100

To the MoooN Cheesy

Ytterbium

full member

Activity: 238

Merit: 100

Quote from: Vbs on September 04, 2013, 06:45:21 PM

Quote from: Ytterbium on September 04, 2013, 06:37:44 PM

What we really need to know is the die size.

It's possible Labcoin's 65nm chip may take up less space, making their chips potentially more profitable.

If we're talking about futurology, we can also talk about VMC going for standard-cell 28nm, using eAsic's easicopy process and getting chips at pitiful costs (price is y-axis below). Then it's game over.

As you can see on the chart, the cutoff for 'profitability' of using the easicopy instead of nextreme is 300-900 thousand units. At 25Gh/s, that comes out to:

7.5-22.5 petahash.

That's the threshold for for it to be profitable for ActM to switch to easicopy.

Assuming he can get 10 chips per board, that's 30-90k boards. At $25/board, that comes out to $750k-$2.2 million for enough boards to put 300-900k chips on.

That means in order for it to be profitable for ActiveMining to go to the easicopy rout to get standard cell asics, he would need to be making an so large he'd have to spend $750-2 million just on the blank PCBs to put them on.

Not impossible. But it's a long way off from what I can tell.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: Ytterbium on September 04, 2013, 06:37:44 PM

What we really need to know is the die size.

It's possible Labcoin's 65nm chip may take up less space, making their chips potentially more profitable.

If we're talking about futurology, we can also talk about VMC going for standard-cell 28nm, using eAsic's easicopy process and getting chips at pitiful costs (price is y-axis below). Then it's game over.

Pale Phoenix

sr. member

Activity: 420

Merit: 250

ffssixtynine, thank you for posting this right away, and for all of the pressure you and the rest applied to get this deal confirmed.

I'm very happy that ActM is no longer the only company with a semiconductor vendor that won't take her to the prom.

cw92

member

Activity: 66

Merit: 10

Quote from: Ytterbium on September 04, 2013, 06:37:44 PM

What we really need to know is the die size.

It's possible Labcoin's 65nm chip may take up less space, making their chips potentially more profitable.

Labcoin has yet to prove that they developed the 130nm chip (or give any indication that they aren't running a total scam for that matter). Why don't you worry about that when their "65nm chip" is not a figment of your imagination.

Ytterbium

full member

Activity: 238

Merit: 100

What we really need to know is the die size.

It's possible Labcoin's 65nm chip may take up less space, making their chips potentially more profitable.

Topic: [ActiveMining] The Official Active Mining Discussion Thread - page 70. (Read 479317 times)