Jutarul did say "Depending on fab capacity", so I am sure scheduling enough fab time might be an issue.
That said, worldwide 40nm fab capacity 2 years ago was 4 million wafers per month. So on the high end we are talking about AM using a fraction of a tiny fraction of worldwide capacity. As an example, Global Foundries alone has capacity to do 50K to 100K+ (depending on wafer size) / month at 40 nm.
Topic: ASICMINER: Entering the Future of ASIC Mining by Inventing It - page 498. (Read 3917531 times)
February 08, 2014, 11:47:09 PM
February 08, 2014, 10:41:39 PM
Who cares about nm size? Do you buy the camera with the highest MB pixel count, or do you get the one with a good aperture and optics that produces the best photo quality? Based on the specs AM is about to release chips that are over 2x efficient as their competitors 20nm chips. If competitors want to spend a boat load of money to do 14nm nodes, then I say they can knock themselves out.
How would a 40nm AM chip compare to a 2nd or 3rd gen 14nm BitFury chip? Who cares about nm size? Everybody with a scrap a sense because it allows you to increase the number of transistors in the same size package or produce a smaller, more power efficient chip. Is process size everything? No but to pretend it's meaningless is just silly. A highly efficient design on a 14nm process would completely destroy AM's 40nm chip. Is such a chip available now? No but 2.5 years from now though, there will be far better designs than AM's 40nm chip. That's how technological progress works.
How would a 40nm AM chip compare to a 2nd or 3rd gen 14nm BitFury chip? Who cares about nm size? Everybody with a scrap a sense because it allows you to increase the number of transistors in the same size package or produce a smaller, more power efficient chip. Is process size everything? No but to pretend it's meaningless is just silly. A highly efficient design on a 14nm process would completely destroy AM's 40nm chip. Is such a chip available now? No but 2.5 years from now though, there will be far better designs than AM's 40nm chip. That's how technological progress works.
Quote from: Mabsark
If you think that's clear, then tell me when gen 4 chips will be released, based on that information. All that that paragraph say is that AM will try to double the number of transistors every 2 years while making the chip more efficient.
Well I think it is about as clear as it can be. It means they aren't hanging up the towel after gen3. Gen3 isn't even taped out yet but you want a specific date when gen4 is going to be done? Please...I don't want a specific date, I want a rough estimate in order for those numbers to be meaningful. It doesn't matter though because those numbers posted are pure bullshit given that it would take decades to produce that many chips. Then again, I could be completely wrong and it may be perfectly possible to produce and package 14-55 wafers per day.
So, how many wafers can be produced and packaged per day?
February 08, 2014, 09:52:42 PM
This is to inform that friedcat met with the board yesterday and provided some updates.
General Update
==========
General Update
==========
Thanks for the update Jutarul
February 08, 2014, 09:44:57 PM
In order to be viable 2.5 years from now, AM would basically have to give the chip away. 14nm will be readily available by then and most manufacturers will have produced far more efficient chips. 2.5 years is far too long a life for gen 3.
No one knows what BTC/USD exchange rate or difficulty is going to do over the next 2 years. If there is a crash or stall, then yes, 2.5 years could very well be the life expectancy of gen3. Do I think that is likely? No. But it is an upper bound. 9 months is my lower bound, which I think is also unlikely. But it is still reasonable to think in min/max caps when doing projections.Who cares about nm size? Do you buy the camera with the highest MB pixel count, or do you get the one with a good aperture and optics that produces the best photo quality? Based on the specs AM is about to release chips that are over 2x efficient as their competitors 20nm chips. If competitors want to spend a boat load of money to do 14nm nodes, then I say they can knock themselves out.
Quote from: Mabsark
If you think that's clear, then tell me when gen 4 chips will be released, based on that information. All that that paragraph say is that AM will try to double the number of transistors every 2 years while making the chip more efficient.
Well I think it is about as clear as it can be. It means they aren't hanging up the towel after gen3. Gen3 isn't even taped out yet but you want a specific date when gen4 is going to be done? Please... February 08, 2014, 09:20:11 PM
If that's the overall expected production for gen3, then an obvious question arises. When is gen4 expected? This year? Next year? Never?
If it's this year, then those numbers seem highly suspect. If the answer is never, then those numbers are meaningless. I'd say gen4 won't be out for a year or two going by those numbers, which would see the chips become pretty weak again compared to the competition. Aren't AM supposed to be working on gen4 already though? In which case, surely it won't take a year or two to be released and therefore AM won't be putting out anywhere near 1600 Ph/s.
Basically, those numbers are completely meaningless without knowing the lifetime of gen3.
No one knows how long gen3 will last, which is one of the reasons for the wide spread of quotes Ph/s. My guess is it will still be a viable (sellable) chip for 9 months to 2.5 years.If it's this year, then those numbers seem highly suspect. If the answer is never, then those numbers are meaningless. I'd say gen4 won't be out for a year or two going by those numbers, which would see the chips become pretty weak again compared to the competition. Aren't AM supposed to be working on gen4 already though? In which case, surely it won't take a year or two to be released and therefore AM won't be putting out anywhere near 1600 Ph/s.
Basically, those numbers are completely meaningless without knowing the lifetime of gen3.
In order to be viable 2.5 years from now, AM would basically have to give the chip away. 14nm will be readily available by then and most manufacturers will have produced far more efficient chips. 2.5 years is far too long a life for gen 3.
As to gen4, that was pretty clearly answered in Jutarul's update:
Quote from: Jutarul
4) What is the currently anticipated long-term strategy for ASICMiner after the 3rd generation?
re 4) Long-term strategy is to follow the step of Moore's Law, while squeeze all possible efficiency out of each stage of chip design and production.
re 4) Long-term strategy is to follow the step of Moore's Law, while squeeze all possible efficiency out of each stage of chip design and production.
If you think that's clear, then tell me when gen 4 chips will be released, based on that information. All that that paragraph say is that AM will try to double the number of transistors every 2 years while making the chip more efficient.
Let's say that gen 4 will be out in 2 years from now and 1600 Ph/s of gen 3 chips were produced. That would mean AM would have to produce about 2.2 Ph/s every day. With wafers at 40 Th/s, AM would need to package 55 wafers per day, producing 171,875 chips per day.
That sound pretty damn implausible to me.
If 400 Ph/s of gen 3 chips were produced in that 2 year period, 14 wafers per day would need to be packaged producing 43,750 chips.
These numbers are clearly ridiculous and they're over a two year time period. As far as I'm aware, it's not even possible to produce and package 1 wafer per day. If that's correct, then it would take AM decades to produce that many chips.
February 08, 2014, 09:01:06 PM
Let's hope the value of BTC doesn't completely tank in the next 6 months and ruin the Gen3 party. If Gox implodes, which it appears to be doing as we speak, there will be a lot of negative sentiment in bitcoinland for a while to come. Of course, Gox is pants and deserves what it gets, but lots of people have their money / BTC tied up in Gox and no way of getting it out.
February 08, 2014, 08:51:21 PM
February 08, 2014, 08:31:27 PM
If that's the overall expected production for gen3, then an obvious question arises. When is gen4 expected? This year? Next year? Never?
If it's this year, then those numbers seem highly suspect. If the answer is never, then those numbers are meaningless. I'd say gen4 won't be out for a year or two going by those numbers, which would see the chips become pretty weak again compared to the competition. Aren't AM supposed to be working on gen4 already though? In which case, surely it won't take a year or two to be released and therefore AM won't be putting out anywhere near 1600 Ph/s.
Basically, those numbers are completely meaningless without knowing the lifetime of gen3.
No one knows how long gen3 will last, which is one of the reasons for the wide spread of quotes Ph/s. My guess is it will still be a viable (sellable) chip for 9 months to 2.5 years.If it's this year, then those numbers seem highly suspect. If the answer is never, then those numbers are meaningless. I'd say gen4 won't be out for a year or two going by those numbers, which would see the chips become pretty weak again compared to the competition. Aren't AM supposed to be working on gen4 already though? In which case, surely it won't take a year or two to be released and therefore AM won't be putting out anywhere near 1600 Ph/s.
Basically, those numbers are completely meaningless without knowing the lifetime of gen3.
As to gen4, that was pretty clearly answered in Jutarul's update:
Quote from: Jutarul
4) What is the currently anticipated long-term strategy for ASICMiner after the 3rd generation?
re 4) Long-term strategy is to follow the step of Moore's Law, while squeeze all possible efficiency out of each stage of chip design and production.
re 4) Long-term strategy is to follow the step of Moore's Law, while squeeze all possible efficiency out of each stage of chip design and production.
February 08, 2014, 08:13:33 PM
What's the price per share?
Thanks
Thanks
February 08, 2014, 08:12:06 PM
Still a pretty nice outlook AM does remain competitive if not all the year its does improve itself during cycles
Thanks Jutarul mid February isn't that far away gives it I guess 2 weeks in real AM time not BFL time
Thanks Jutarul mid February isn't that far away gives it I guess 2 weeks in real AM time not BFL time
February 08, 2014, 08:03:33 PM
thats not really amazing news at all.... 1600PH?!?!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
No offense, but it's been obvious for quite some time that mining was going to become an industrial-scale domain.
Don't forget that many of these chips will be sold to miners.
I'm hoping they release a next gen cube and usb miners.
February 08, 2014, 08:01:06 PM
thats not really amazing news at all.... 1600PH?!?!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
No offense, but it's been obvious for quite some time that mining was going to become an industrial-scale domain.
February 08, 2014, 07:56:08 PM
b ) Submitted Questions:
1) What is the order of magnitude of 3rd generation chips ASICMiner anticipates to sell (total / first batch / first 3 months)?
re 1) Order of magnitude: Depending on fab capacity. Total target is 10k to 40k wafers, relying on competitors, Bitcoin price, and other factors. A wafer is rated at 40 TH/s.
Did I read that right? 10,000 - 40,000 wafers, at 40TH/s each, = 400 PH/s - 1600 PH/s total target gen3 capacity?
That number can't be right. Last estimate was 2-20ph.
If that's the overall expected production for gen3, then an obvious question arises. When is gen4 expected? This year? Next year? Never?
If it's this year, then those numbers seem highly suspect. If the answer is never, then those numbers are meaningless. I'd say gen4 won't be out for a year or two going by those numbers, which would see the chips become pretty weak again compared to the competition. Aren't AM supposed to be working on gen4 already though? In which case, surely it won't take a year or two to be released and therefore AM won't be putting out anywhere near 1600 Ph/s.
Basically, those numbers are completely meaningless without knowing the lifetime of gen3.
February 08, 2014, 07:51:53 PM
thats not really amazing news at all.... 1600PH?!?!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
thats why i sold all my mining gear and bought AMshares
1600ph would equate to 4th/share. Not bad
February 08, 2014, 07:39:59 PM
thats not really amazing news at all.... 1600PH?!?!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
thats why i sold all my mining gear and bought AMshares
February 08, 2014, 07:36:55 PM
thats not really amazing news at all.... 1600PH?!?!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
You realize, he just put the rest of all us miners out of fucking business. Yall should be outraged...not jumping up n down with joy!
February 08, 2014, 06:59:44 PM
If tape out is expected mid-February, does that push production at all?
Thank you for the update.
Thank you for the update.
February 08, 2014, 06:53:19 PM
b ) Submitted Questions:
1) What is the order of magnitude of 3rd generation chips ASICMiner anticipates to sell (total / first batch / first 3 months)?
re 1) Order of magnitude: Depending on fab capacity. Total target is 10k to 40k wafers, relying on competitors, Bitcoin price, and other factors. A wafer is rated at 40 TH/s.
Did I read that right? 10,000 - 40,000 wafers, at 40TH/s each, = 400 PH/s - 1600 PH/s total target gen3 capacity?
That number can't be right. Last estimate was 2-20ph.
400-1600PH/s overall target... A M A Z I N G
edit: @jutarul thx for keeping us in the loop
February 08, 2014, 06:46:43 PM
b ) Submitted Questions:
1) What is the order of magnitude of 3rd generation chips ASICMiner anticipates to sell (total / first batch / first 3 months)?
re 1) Order of magnitude: Depending on fab capacity. Total target is 10k to 40k wafers, relying on competitors, Bitcoin price, and other factors. A wafer is rated at 40 TH/s.
Did I read that right? 10,000 - 40,000 wafers, at 40TH/s each, = 400 PH/s - 1600 PH/s total target gen3 capacity?
That number can't be right. Last estimate was 2-20ph.
February 08, 2014, 06:25:04 PM
b ) Submitted Questions:
1) What is the order of magnitude of 3rd generation chips ASICMiner anticipates to sell (total / first batch / first 3 months)?
re 1) Order of magnitude: Depending on fab capacity. Total target is 10k to 40k wafers, relying on competitors, Bitcoin price, and other factors. A wafer is rated at 40 TH/s.
Did I read that right? 10,000 - 40,000 wafers, at 40TH/s each, = 400 PH/s - 1600 PH/s total target gen3 capacity?
That number can't be right. Last estimate was 2-20ph.
Jump to: