Topic: ASICMINER: Entering the Future of ASIC Mining by Inventing It - page 1298. (Read 3917029 times)

I'd like to see it listed on Bitfunder instead.

Back on topic for a moment -

Have we made progress on re-listing ASICMINER on another exchange (now that the list has been received from GLBSE)

I'd like to reiterate my vote for    BTCT.CO   https://btct.co/

Another asset of mine (BAKEWELL) has just been reborn there, and I'm pleased with the results.

I believe we'll actually see the share price of ASICMINER jump once trading has resumed. I know I'll be bidding...

Both at the same facility in Austin, correct?

Samsung has 2 large memory fabs in the US.

It's been more than a few years. Hell, for DRAM this is from 2001

Micron in the US is 17%, Elpida in Japan is 6%, Samsung and Hynix in Korea total 43%. I'm not really sure where all Infineon's fabs were in 2001 but I don't think they had any in Taiwan or China, I know they had a joint fab with Nanya later than that in Taiwan.
For flash, the market share has even less produced in China/Taiwan
http://www.statista.com/statistics/201746/market-share-of-leading-nand-flash-memory-manufacturers-worldwide/
Intel's production is in the US, Samsung in Korea and Toshiba in Japan. Micron is in the US and Singapore.

You can order GloFo runs through MOSIS both for MPW and COT runs.
http://www.mosis.com/vendors/view/global-foundries
CMP offers access to ST's processes down to 28nm on MPW and small volumes, or you could go direct to ST for larger runs.

We should/could get another production update from Friedcat tomorrow....

No, the 98% figure was true up to about a few years ago. Now its Taiwan and South Korea, both countries being in the shadow of China (and Taiwan even still calls itself the Republic of China, rejecting the Communist revolution). I simply forgot about the newer Korean fabs.

GloFo AFAIK doesn't take jobs this small, but I could be wrong. And a lot of their fab power is in Singapore, which is still within the sphere of the Chinese tech sector's influence.

Now, don't get me wrong, I'd love these chips to be fabbed in the US. Like, NatSemi still has a fab in Portland Maine, which is like 5 hours away from me, but I don't they think they take contract jobs (and I don't think they offer a node size small enough at that location, but they do at ). But look at the rest of friedcat's competitiors: its all TSMC (China, Taiwan), UMC (Taiwan, Singapore), or SMIC (China).

Friedcat needs to produce these at a fab that can do this with sufficient quality and still be cheap, as does everyone else. If doing it in the US or EU increases the cost even 10-15%, then this makes it that much harder to profit.

Nice... another month. I think many people now are angry that they put their money into bfl for months and the outcome of this is very open. The better for asicminer.


Sounds correct taking friedcats last update into account. At the moment there should be around 8 layers left if the speed remains the same. I guess the layers are done at decembre 26th. Could be a hard race but friedcat said asicminer has the opportunity that shipping is done in the same city. So shipping will be faster probably. At the end... i wonder how much worth one day mining without competition will be.

Only important and relevant parts are:

1) When those chips are ready, is it guaranteed they actually work or will there be a random testing?

2) Will this happen before or after the shipping to ASICMINER team?

3) How many chips (%) to you think will be lemons?

Do not get too exited about the rest

You didn't say "the kind of fab you need for this project is in Taiwan", you said that 90% of chips are fabricated in China and Taiwan, and the ones that aren't are high end CPUs or rad-hard chips, as well as 98% of memory being made in Taiwan. Both of those are very, very incorrect.

ST in Europe does a lot of contract fab work, and is another very viable option for a project like this. GloFo is also accessible @ 65nm.

I forgot about the Korean memory fabs. Its still the same thing in the end.

As for your list, yes, I'm aware of it. I have it in my bookmarks just for chip origin research purposes. Problem is, the list does not include how many chips the fabs produce every year. For example, TI's fab do not produce a lot of non-TI chips, and TI's most popular chips (which are not complex things like ARM SoCs) are often not fabbed at TI.

And yes, Freescale is still largely US, which is great. Samsung also does a lot more than just memory, also.

Now, given the kinds of fabs we need for this kind of project.... TI and Freescale and Samsung can't help us. Virtually all of those really are in China/Taiwan/Korea, although I think I remember ArtForz saying there was one in Mexico thats popular but the quality is worse than Chinese fabs.

http://en.wikipedia.org/wiki/List_of_Semiconductor_Fabrication_Plants

Pretty much all of TI's stuff is made in the States, as is NationalSemi and OnSemi (formerly Motorola's power devices). Freescale (Motorola's IC spinoff) is mostly US. ST is all in Europe. Samsung the #1 memory maker in the world and their fabs are in Korea and the US. Micron is the number 2 memory maker worldwide and they run fabs in the US and Elpida's fab in Japan. Hynix is mostly in South Korea and the US, with one Chinese fab.

Well, stuff being made in China really is an issue. Its not about racism or nationalism. I, and most of the rest of the people living in my country, don't care where its made. Most of us don't even care that its a modern form of slave labor (I don't include myself in that group, however).

I was just backing up the Toyota will survive nuclear war sentiment.

That said, over 90% of chips fabricated are done so in China or Taiwan. 98% of all memory is done in Taiwan. Pretty much the only chips that aren't fabbed in China or Taiwan are either high performance CPUs (Intel, AMD, IBM, etc) and/or rad-hard milspec "its a matter of natsec" chips.

It doesn't matter how racist or nationalistic you are, you can't change the fact that China is the technology industry. It is not worth bringing up the fact friedcat has fabbed his chips in China because there really is nowhere else to do so.

This thread seems to periodically derail into a nationalistic/xenophobic pissing contest. Not productive, nor interesting... *yawn*

Erm, except Toyota is the only major car manufacturer in the US. Ford, Chrysler, and GM do manufacture a few cars in the US but ship them over to Europe; the ones sold in the US are all Chinese made and Mexican assembled.

Toyotas will not fucking die. Ask Top Gear.

When those chips are ready, is it guaranteed they actually work or will there be a random testing? Will this happen before or after the shipping to ASICMINER team?
How many chips (%) will be lemons? 10%? 20%? 30%? Lets not forget that this stuff is coming from China, the biggest producer of crap in the world.

BFL nor ASICMINER are big clients. They can scream and threaten, but this will not guarantee quality nor delivery dates. Yes, Chinese manufacturers are well known for "me price so low" but we all know that "Made in China" is a synonym for "shit"  
Friend of mine, who is in business of building non BTC related hardware, told that expecting those greedy slimy chinese dogs to deliver anything on time and up to the quality requirements is flat out stupid. The fun starts, when your shipment arrives and you realize, that up N% of the stuff is actually junk.
Talked to another friend in optics business. Same story. Price per unit is low but the local quality control is probably done by blind kids.
BTW, if you ever make it to China, go visit some of the factory dump sites. You will see mountains! yes, mountains! of crap that has never made it out the door.
Make sure you do not get caught and ware protective clothing. China is not inly the biggest crap producer but they are also the biggest polluters. Humans cost nothing over there.   
 
I did not want to piss on your parade. I honestly hope you guys get a better results. Be optimist and x 2 everything they tell you (and you still get screwed by half) 

So let's do some back-of-the-envelope calculations here: on Dec 5, friedcat said there were 12 layers remaining and the rate is 1.1-1.5 days per layer, which brings us to chip completion date of between Dec 18 and Dec 23. Add another two weeks for packaging and shipping (estimate pulled out of thin air!), then one more week of unwrapping/assembly, which brings us to an estimated hashing starting date of between Jan 8 and Jan 13. Sound reasonable?

And for BFL: chips delivered to their HQ Jan 11, unwrapping/assembly: 1 week, shipping to customers: 4 days, which gives us an estimated hashing starting date of Jan 22.

Pretty close race, but it looks like ASICMINER definitely will win it. I even think it is likely that ASICMINER will move a little faster than estimated above, and BFL will move slower than estimated above. But these back-of-the-envelope calculations give us at least a week as the only ASIC gig in town!

Edit: Anyone familiar with the bASIC and Avalon willing to make a guess at the date they start hashing? Both seem to have been giving mid-January delivery dates, but I don't know if that's the latest available information, or if it is at all accurate.

The standard contracts you get from the foundry put customers at the mercy of the foundry. Even if they are past the deadline, they have the right to continue the order without penalty. All they usually require is some good faith estimates. BFL seems to experience that first hand. And if you decide to scrap the order due to the delays, they'll write a bill nevertheless.

If you're a bigger fish in the sea you can play smarter and pressure a foundry for more aggressive contracts, where they loose money on pushing back dates. As we already know, the chip production of ASICMINER is also already past due date. It's a problem of trying to be economical. Let's just hope that friedcats choice for the foundry was a good one (mainly determined by luck).

It seems our ASICMINER shares just got worth a little bit more.

BFL announced another much anticipated new delay. Will be about 30 days from now until they get their chips delivered from the foundry. Then they need to turn those into mining rigs of course, which means BFL customers probably won't be mining before the latter part of January 2013.

So if everything with ASICMINER stays on schedule that means we will be mining on a lower difficulty level for 3 to 4 weeks. With about 1/5th of the total network network capacity.

Sounds good to me  

edit: BFL update link: https://forums.butterflylabs.com/bfl-forum-miscellaneous/519-10-dec-2012-bfl-asic-update.html

There has been no communication yet on where the shares will be relisted