1. We did not lie on Jan 3. We were planning to tape out before the Chinese New Year. On Jan 3 we were informed that the back end team cannot reach their target and need to try another layout. This meant to do some changes in the RTL code and start back end process all over again. We have immediately announced the worst date we could think of for batch 2 and not announced details about batch one because we did not had them. We just got the final data from the back end team 5 days ago. They have now reached their target and are now 99% sure that they'll tape out 1 week after Chinese new year (15-20 Jan).
2. We have looked into our competitors claims together with Global Foundries. GF is obviously not at liberty to discuss the specific projects with third parties but they assured us that it is impossible to finish 28nm in 49days and to cover every step. 28nm is very complicated and DRC/LSV and JDV checking takes at least 14 days. The wafer manufacturing takes 0.8days / layer at their highest speed. Our conclusion is that our competitors have deliberately omitted several steps in their advertising and were only referring to wafer manufacturing.
3. The tapeout costs millions of dollars. We do not have a customer that paid that much to make a tapeout just for them. We had to think further than 1 month. 1.5W/Ghash is not only electricity problem. The whole power translates into heat, therefore more things are required: power chips, heatsinks, AC, new case design, etc.
4. We cannot produce the PCBs without the exact pin layout of the chip. This has not been yet received. We could not produce the case without the exact heat estimation.
5. Black Arrow is the only Bitcoin ASIC manufacturer that has not betrayed customer's trust. While others let the deadline pass by and did not announce anything we have clearly informed our customers of the situation.
The tapeout is almost complete and the path from here to ASIC is clearly drawn in the industry.
6. Yes, we should be able to run at 120Ghash/sec according to our tools and Verisilicon. Verisilicon gives us the SS corner for speed and TT corner for power. FF corner can be even faster.
For those who do not know:
SS = slow slow - absolutely lower limit (speed & voltage) where it can run
TT = typical typical - normal usage; note that 28nm is rated at 25C but leakage increases dramatically with the temperature. The real power consumption at 60C is only a guess at this stage.
FF = fast fast - higest speed, highest voltage, also the highest power consumption. - we have no data on this, it's only discovered while testing real chips.
7. We only have few Bitfuries left. It would mean to give everybody less than 1% of what they have ordered. This i more trouble than it worths.
8. The chip is designed in China, manufactured in Germany, packaged in Taiwan. Finally the Prospero will be assembled in China
http://ecointalk.net/topic/463-we-want-answers/page-3[/quote]
Why was there no notification on jan 3 that tapeout targets could not be met? Even if jan 3 target was met, based on the current timeline, the feb 24 target would have still been missed. This was not disclosed to us. Why?
Do you know that the BA chip will be two months later than the AM 40nm chip and have worse performance? Why are we continuing with this?
PCB layouts will soon be out or are already out for AM chips.
BA kept from us knowledge that the jan 3 target was missed. That affects trust. We could've at that point in time mitigated our losses if given full disclosure and options.
You still haven't answered questions re: who runs BA at the top, who pays your wages and the Shenzhen address?