Quote from: dmcdad on February 09, 2014, 01:31:27 AM

Quote from: Mabsark on February 09, 2014, 12:56:08 AM

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.

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.

Quote from: dmcdad on February 09, 2014, 01:31:27 AM

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.

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.