Topic: HashFast announces specs for new ASIC: 400GH/s - page 567. (Read 880461 times)

The people involved, the ones with "combined total of 100 years of experience"?  Not sure if the 3 folks listed on the site have a hundred years between them since birth, but since none of them list any HW experience, i assume that's exactly what they mean -- 100 years of life experience, from the cradle 'till today

No evidence?  Except that the people involved have created things in the past successfully, and you have given no reason to think that they won't be able to continue this.  No I don't mean Simon and his paper.  I mention that to show that this isn't some new person coming out of the woodwork because they smell money.

edit- Thanks John, and gateway for the responses.  I'll try that email.

trigeek, I must have ignored your assumption about network hashrate, I was assuming 20% rise until then.

Wait, they're past tapeout?  The masks are at the foundry, waiting to become some serious chippage?  Which fab plant?  
And if you use "there's no evidence to the contrary,"  You should believe in Easter Bunny, Tinker Bell & the Fabulous Land of Equestria.
   Hint: Other than a snapshot of a dev board next to a cheap scope, there's no evidence that they can, or even are planning to, "execute on a plan."  Dude.

Great overall post, but this is the point I always come back to. The professional business world does not conduct themselves in the same manner that Butterfly Labs or Avalon does. There are companies who execute on every step where they failed perfectly because they know their business and have for years.

Anyway, there is a real dichotomy between people who at least have some understanding of chip design and implementation, and those who don't in evaluating these companies.  

For people who understand how all this works, or at least have some idea (I've learned a ton from reading these threads) the idea of making a chip and then using it seems like a pretty straightforward proposition.  People with skill and experience should be able to turn these chips out pretty easily.

On the other hand people who's experience is just BFL and Avalon think that failure and huge delays are totally standard and that making an actual chip is some kind of immense technical challenge that's impossible to pull off.

The big difference is the size cost of bitcoin and thus the size of the bitcoin mining 'industry'. At ~3600 BTC per day, with bitcoin at $100 bitcoin mining is a $130 million dollar industry, annually.

If you invest a million dollars and end up with just 2% of the network, you can still double your money in a year.  That's a good return for any investment.

On the other hand, when this started bitcoin was around $5, although block sizes were 50btc, they would probably be down to 25 before you got online. So realistically you'd need to be like ASICMiner - and capture 20% of the network, or you'd need to be like BFL and just take as many orders for hardware as possible without worrying at all about profitability, or even your own ability to deliver.

So it's not surprising at all that more companies would get involved.  ORSoC and Uniquify have plenty of experience.  Is there a risk?  Sure.  There is a risk that something might not be right with the first IC run.

But, while the costs of a bad batch of ICs are going to deadly when it comes to bitcoin mining, it's still a HUGE cost in traditional IC design if there's anything wrong with your IC. They'd have to re-do the mask, and at 28nm that would be crazy expensive anyway.

So it stands to reason that these companies actually do know how to get chips to work on the first try.

Also keep in mind that Avalon's chips did work on the first run (as far as we know). They were also at a larger feature size.  The problems they ran into weren't related to their chips at all, but rather the logistics of building a bunch of boxes and getting them mailed out.  This is the area that KnC does not have any experience in.  They've outsourced it (which is probably what Avalon should have done), but there's still a possibility they might fuck it up.

There is also the possibility of delays due to the fab, and other people in the supply chain.

I don't think it's realistic to say there's no risk.  If you're not happy with the possibility of a risk, you can ask KnC for a refund.  So far people have not been having any problems getting one.

With HashFast, well, they haven't even taken anyone's money yet (Unless someone took up gridfinity's $500k pre-order deal, lol).  They don't owe anyone any information at all at this point.  Presumably that information will be available before they take orders, and people can make up their own minds.   Unlike other companies, they (supposedly) don't need any additional capital to go to production. They may have their chips done and hashing before they take any orders at all.  Or not.  We'll have to see what they decide to do and people can make a decision based on what they thing the outcome will be.


Again dude, so what?  The news here is that a high quality ASIC design company (Uniquify) is working on a bitcoin ASIC, and that company (Uniquify), not just HashFast, has said they've taped out and expect 400Ghash/sec.

Obviously people should be cautious if they're investing money. But simply pointing out there's no chip and no mask isn't really an argument.  There's no real evidence to suggest that these guys can't execute on a plan.  Not every company in the world is BFL.  

Purely as a thought experiment.  It's a common size for orders, and those are also the quantities you can order 2.7GH/s chips from Bitfury when buying in bulk.


Yes, I took this into account when calculating my numbers. 3000*400 = 1200TH, and if the network was 1000TH, 1200/2200 = 54%.

The point is so blunt that it's embarrassing for me to be the one making it.  Yet this thread has grown to 20 pages of unchecked ASIC mania, so someone had to do it.  
I see you're already orchestrating a group buy?  May i ask what it is you're buying?  

You gave a link to a paper coauthored by Simon Barber.  Bitcoin-related, for sure, but technical?  It's as relevant to ASIC design as The US Constitution.  Have you read it?  In case my point is again obscure, i'll spell it out:  The paper has nothing to do with any sort of silicon design.  Nothing.  Not relevant.

TL;DR:  What ASIC?  What specs?  Where's the cat?  Where's the cradle?

You can send inquiries to [email protected]

John

You know you guys don't have to include the entire quote chain when you reply to eachother right? You can cut it down, like so:


Makes the thread a lot easier to read.

Send him a PM he's usually good to reply here, but I also expect this week has been quite busy for them so give it some time..  I could give you his cell phone # but then I would have to kill u j/k..

Did you get that 3000 chip per reel from somewhere, or as a thought experiment?

Also, if you add 50% of hashing power to the network you'll also increase network size, ending up with only about 33% of total network.  To make a proper 51% attack starting from 0%, you would need to more than double the network speed.  Moot point in this scenario I know

Is your main method of contact a personal message to HashFast bitcointalk.org user?  I'm not finding any email addresses posted.  I can understand not giving dev crew's out, but shouldn't John give his?  I've searched this forum, google, their homepage, and there is no contact info.  At only a couple months out from shipping, I guess I would expect to be able to contact someone directly.  How can people come up with 500K without asking a few basic questions?

Until a little bit more information is released this is just speculation on speculation, but I'm setting up a Group Buy to start organizing interested parties, as people learn of this.
https://bitcointalksearch.org/topic/m.2854667



What is your point Crumbs?  This thread is about a product that will exist in the future.  Whether the cat will be alive or dead is yet to be determined.  You seem to be saying that people are jumping to conclusions too fast, maybe you are doing the same.

There is a large group of knowledgeable people working toward a goal.  Is that nothing?  This is how all chips have been created.  If your point is that people should not buy a product that doesn't exist, you've done that thoroughly.  Unfortunately people who wait until there is a finished product are too late in the ASIC game.

Or if the kitten is alive or dead . I'm sure Schroedinger had just this sort of situation in mind when he came up with his cruel (thought) experiment!

Edit: should have read the post above. Sorry for repetition!

Except in this case we're told by Schrodinger himself that there *is no cat in the box.*  
The cat's aliveness or and dead-as-a-doornailness are both predicated on the cat's beingness in the box in the first place
Sorry for all that technical jargon  

Edit:  replaced inclusive "or" with "and" for clarity -- quirky language, English is.

I guess the point is, that until the wavefunction is collapsed and it's observed as a working hashing chip, it's in a Schrodingers cat state of both being and not being until that observation is made.... and not only that we don't even know whether there's a kitten or a tiger in the box yet.

Boldface.

No, I was being facetious. Clearly.

I don't think all of those things were a setup for a scam, do you?  If not, what's your point?  You feel that good education & scamming are mutually exclusive?

Yes and Simon's been planning this elaborate ruse for years, throughout his education at a prestigious UK university, he styled an entire fantasy career out of it culminating in a seminal piece published in a pier reviewed cryptocurrency journal about the future of Bitcoin, all to disguise his true intentions to scam a forum.

I'm not saying they might not suffer delays, but the guy behind it is real.

With the gold fever that everybody has at the moment I think it's unlikely that an auction (if that does turn out to be the plan) would result in lower prices than expected . Remember the ASICminer blade auctions and how people lost their heads over those? I'd be happy to be proved wrong on this though.