Topic: Looking for system integrators for new asic - page 4. (Read 21950 times)

And then there is the scale thing. If that company is a large one, they couldn't care less about some BFL let alone Avalon. They could throw a few $100K around just for laughs, and see if their chip actually worked / had some market.

Of course he is talking about the recent scamsics. He seems to make a point of not naming competitors ever, but clearly this is not about BFL or avalon, but about primeasic and two others who's name I dont recall.

As for him being online for 59 minutes; maybe if Josh hadnt spent half his time on this and his own forum the past 6 months, BFL would be shipping by now? 

Are you sure that he isn't talking about the recent announcements like PrimeASIC as opposed to the products that were announced and documented a half year ago like Avalon and BFL?

This surprises the shit out of me.


Somebody, tell me how an entity can invest so much time and money into developing a product, but doesn't know a damn thing about their direct competition. Seriously, you couldn't (you forgot that accent thingy) tell if Avalon and BFL are real or not? How about spending a little more time researching, read reading this forum.

Granted, I applaud your efforts, but something doesn't smelling right here. The other two companies are not using your product because it's still in development, hence you coming to board to see if there's a market for a product you're considering developing.

~Bruno K~

Of course. They have chips since January maybe. Still, I wonder how many :-)
Of course the GH/s value per chip is also not so important. More important is GH/s per wafer or GH/s per die area AND GH/joule [if possible at different voltage levels].

Just because they taped out in October does not mean they had the chips in hand in October.

"we (ab)used spare estate on our 28nm SOI MLM test wafers we were running at ST Microelectronics"
interpretations:
1. This is not just few chips if the chip is on the MLM mask. MLM (mask) means there are probably few thousands of chips from few wafers.
The same MLM mask will be used for the production run. Assuming 25 production run wafers produce 1PH/s in chips, the engineering run with 6 wafers generated > 200TH/s. The only reason the chips are not online yet would be that there are no boards for it.
2. (purely hypothetical) when producing chips with MLM masks there is some small area on the wafers left that can be used for other designs (something like MPW on an MLM wafer ). Maybe this would generate only few chips.

tapeout happened in october, this is a technical term that refers to finishing the design. It doesnt mean functional chips, its the last step before you send your stuff to the fab for production. It sounds like they have produced a few test chips earlier this year,  mass production only slated for june. So they may have a few chips mining for a few months now, assuming they have working boards and software, but thats not likely to draw a lot of attention in the stats anywhere.

they may have made their ROI by mining with their test chips, and now that difficulty is going up they will sell chips to us....

I don't really understand this paragraph. What kind of functionality can those chips provide?

If it is a bitcoin mining chip as you said,  you should have started some mining operation on the network since last October?

Thanks for the feedback. Also respect that you don't wish to make statements or promises you then don't deliver on.

I don't like the constant auction method. Will mean they make a stack load of money as customer clammer to get their hands on them. Should see how people are jumping at the idea of 80ghs at 800w. I would rather order at a known price that I can weigh up the pros and cons of features and pricing and be told the lead time for orders.

It would be good to know what your chips can do. Sure you can't make a choice without a prce and timing, but from a tech hungry point of view it would be interesting. A statement like "we predict each chip to operate at xx-yy ghs per second and have a power window around 0.x-0.y w per ghs. Your integrators final performance will then likely vary depending on their cooling performance and hardware they build with.

right. I need more coffee :-)

Says so in the same post:
The performance of our product will be unveiled at a later date in a joint statement with our partners, expected before the end of this month.

Will You disclose the two companies ? If yes, when ?

We have been flooded with RFI's, I apologize if we havent replied to everyone (yet)
(That apology doesnt extend to those still sending personal forum messages, as pointed out numerous times, please use our email address, as we have better things to do than solve captcha's all day.)

Based on contacts and initial feedback so far, we have made a slight adjustment to our initial plan. In a first phase, we will concentrate our time on two selected companies with a proven trackrecord in B2C (manufacturing, logistics, worldwide sales and after sales support) that are willing and able to market a reference implementation. Time to market is critical, so we want to make sure end users will have several choices of providers this summer, even if this goes at the expensive of diversity of technical solutions initially.

In a second phase we will talk to some of the many engineering companies that have contacted us, most of which appear to have the required technical know how, but have little or no trackrecord in medium volume B2C. We will help those companies create custom implementations for our product,  which they can either use themselves, sell themselves, or sell to our customer facing partners.  We expect to enter the second phase in a few weeks, after we are satisfied our first B2C partners will be ready in time with reference implementations. This means custom solutions may or may not hit the market in time for our first production run, but our first priority is to make sure our chips will not sit idle in a warehouse all summer.

To answer a few recurring questions from forum users:

- Is company X or Y using your ASIC?
We have nothing to do with the other recent asic announcements on this forum. We do not know those companies, and they are not using our product. I also couldnt tell if they are real or not.

-Whats the performance of your chip? How much will it cost? When will it ship?
The performance of our product will be unveiled at a later date in a joint statement with our partners, expected before the end of this month. I understand miners want to know now, but the reality is that  early performance figures will not help you much deciding if you should buy from our competitors as long as our numbers are not complemented with a price and rock solid shipping date. Especially the price cant be determined today and we cant yet promise a precise shipping date. All I can say with confidence is that we will have a technically superior product and our price will be competitive; I might also add one of our suppliers plans to sell its products through constant auctions, so you will determine the price yourself.

-What will happen to difficulty when you ship? How many TH will you add this year?
It should come as no big surprise that the aggregate hashrate of our first production wafer run is expected to be in the petahash range, but I cant comment yet on a time table for its arrival on the market.

-Will you partner with competing company X or Y?
We do not exclude the possibility of anyone partnering with us; anyone that buys our chips is first and foremost a customer, rather than a competitor. Since no one else is selling asics at this point, we could consider ourselves without direct competition. However, we are focused on not repeating the mistake some of our colleagues made by underestimating the challenges of B2C.

Many people here must have contacted with him already. Why not say something and confirm if this is real?

Bump this thread a little, this is more likely to be coming months' topic 

I really think it would be in our interests to move away from USB as the primary connection method (as a long term goal). Why not ethernet or stat. At least then we could have the option of daisy chaining devices together.

As to your previous comment about shiny cases... I'd be fine with having to build my own rack to use bleeding edge tech... but when we're talking about a retail product that can be mass produced it has to be presented well... the only serious options imho a stand-alone case like the BFL mini-rig or a rack mounted case if you want to go for purity of function. I really wish somebody would produce a u3 or u5 asic miner that I could be sold in small lots and just stuck into a rack.

he said the reference design has network and USB controllers. But of course an integrator could do something different.
In theory you could have boards connected to each other via USB hub, and one board using the ARM core to run the mining software. So you could have a more expensive "primary" board and numerous "secondary" or "slave" boards that don't have all the extra sub systems required on the "primary" and hence cost a bit less. You could then build your own mini rig!!

The semiconductor industry is having huge problem right now, bitcoin is the savior for ST

I hope the referense design is usb or esata interface, just plug in and mine