ASICMINER: Entering the Future of ASIC Mining by Inventing It

[rudi]

Shouldn't we all be super stoked right now?

We produced the test boards in December 2nd, and got the packaged sample chips last night.
We were busy testing since then and had some preliminary data already.

Technology: TSMC 28nm HPC

Package: FCLGA 5mm x 5mm

On chip efficiency:
    0.56V | 4.8GH/s | 1.95A | 0.2275W/G
    0.60V | 5.2GH/s | 2.14A | 0.2469W/G

We are still testing more voltage-frequency combinations, as well as more chips.
But the test results are stable and solid so far.

Our power solution evolved from Prisma allows very low overhead in supplying very
low voltage levels to chips compared to standard DC/DC converters. Therefore on-wall
power consumption would be very close to (our on-chip power/PSU efficiency).

After the thorough testing of single chip boards, we are going to test boards with chained chips.

[1714071204]

1714071204

[1714071204]

1714071204

[1714071204]

1714071204

[1714071204]

1714071204

[1714071204]

1714071204

[Puppet]

Shouldn't we all be super stoked right now?

Certainly its better news than 'the chip doesnt work', but these are a few samples on a shared multiproject wafer. The idea is to be able to tweak the design before producing a full mask set. That means that mask set isnt done yet, and it will probably be 3+ months before production chips arrive if they were ordered today.

At least bitfury and spondoolies promised 0.2J/GH at the wall late this year or early next year (and twice that efficiency next year). Assuming they, or bitmain or anyone else, can achieve that, AM will be a fews months behind and with worse efficiency (*). Hardly a better competitive situation than BE200 was in, and that wasnt exactly a commercial slam dunk.

Its a shame friedcat didnt trust his design enough to go for full mask sets directly, that would have been something to be "super stoked" about. Now, its mostly a matter of hoping the competition doesnt meet its goals.

(*) numbers are still being updated and probably relate to few, possibly just 1 chip, so they may still change.

[raskul]

Shouldn't we all be super stoked right now?

Certainly its better news than 'the chip doesnt work', but these are a few samples on a shared multiproject wafer. The idea is to be able to tweak the design before producing a full mask set. That means that mask set isnt done yet, and it will probably be 3+ months before production chips arrive if they were ordered today.

At least bitfury and spondoolies promised 0.2J/GH at the wall late this year or early next year (and twice that efficiency next year). Assuming they, or bitmain or anyone else, can achieve that, AM will be a fews months behind and with worse efficiency (*). Hardly a better competitive situation than BE200 was in, and that wasnt exactly a commercial slam dunk.

Its a shame friedcat didnt trust his design enough to go for full mask sets directly, that would have been something to be "super stoked" about. Now, its mostly a matter of hoping the competition doesnt meet its goals.

(*) numbers are still being updated and probably relate to few, possibly just 1 chip, so they may still change.

invariably, as we all know, this industry is never a one-horse race. All good points made above, but for this GEN however, it may well boil down to who gets first out the traps. That said, having samples to test does no way go the length to offer a best buy, just yet.

interesting times ahead.

[bones]

Gen 4 looking great so far, bulk chips due feb ish if I recall correctly.
And prepping for smaller node size already underway...

Im a happy camper.

[bitdude]

FC previously announced news might come around this date and now he really is back with the news. Definitely, this is an improvement

[Mabsark]

invariably, as we all know, this industry is never a one-horse race. All good points made above, but for this GEN however, it may well boil down to who gets first out the traps. That said, having samples to test does no way go the length to offer a best buy, just yet.

interesting times ahead.

I think AM will be first on the scene with their 4th gen chips, with Spondoolies about a month or two behind with their next gen.

Update
14nm/16nm projects are at pre-evaluation and pre-design stage. It has to be done
in 2015 but the starting time of placing orders depends on overall gain vs (NRE/R&D/risks).
...
The mass production time of BE300 in terms of chip-out date is February to March, 2015.

From this CCN interview with Spondoolies:

Could you tell CCN about what we can expect in the future from Spondoolies-Tech?

We have a very ambitious schedule in front of us.  In Q1, we will bring our next 28nm chip.  It has very ambitious specifications. Then by the end of q2 we will have our 16nm chips, with out of dimension specification. We are looking to be the most significant manufacturer in the SHA-256 industry by the end of 2015.

So, it seems like AM will be first out with their next gen but it looks like Spondoolies or KNC will likely be first to 16nm.

[Blazed]

invariably, as we all know, this industry is never a one-horse race. All good points made above, but for this GEN however, it may well boil down to who gets first out the traps. That said, having samples to test does no way go the length to offer a best buy, just yet.

interesting times ahead.

I think AM will be first on the scene with their 4th gen chips, with Spondoolies about a month or two behind with their next gen.

Update
14nm/16nm projects are at pre-evaluation and pre-design stage. It has to be done
in 2015 but the starting time of placing orders depends on overall gain vs (NRE/R&D/risks).
...
The mass production time of BE300 in terms of chip-out date is February to March, 2015.

From this CCN interview with Spondoolies:

Could you tell CCN about what we can expect in the future from Spondoolies-Tech?

We have a very ambitious schedule in front of us.  In Q1, we will bring our next 28nm chip.  It has very ambitious specifications. Then by the end of q2 we will have our 16nm chips, with out of dimension specification. We are looking to be the most significant manufacturer in the SHA-256 industry by the end of 2015.

So, it seems like AM will be first out with their next gen but it looks like Spondoolies or KNC will likely be first to 16nm.

I hope they stick with bulk chip sales this time and not bother to make miners. It seems like the miners made by RM and xBtec were pretty good quality wise.

[raskul]

invariably, as we all know, this industry is never a one-horse race. All good points made above, but for this GEN however, it may well boil down to who gets first out the traps. That said, having samples to test does no way go the length to offer a best buy, just yet.

interesting times ahead.

I think AM will be first on the scene with their 4th gen chips, with Spondoolies about a month or two behind with their next gen.

On this occasion, i have no insider knowledge as I did with KfC bungled attempt at 20nm, but here is my take on it.

FC was always going to want to be first to announce here, due to the fact that their shares are of public interest to members here and early announcements are beneficial to the price of the shares.

Other companies will not need to give out any info prior to an actual miner being launched, and because of this, as ever, i'd expect a very close time scale between Bitmain, Bitfury, AM and SP-Tech to market. (forget KNC - theirs specs will not be competitive).

Batch chips are likely to be baked and ready for all four, by latest, March 2015 and from that point, the real race will be to bring a miner to market.
If AM decide to just sell chips and not build, this may or may not work in their favour, depending on whether they are pushing for pre-orders of those chips.

IMHO, if AM does push manufacturers to pre-purchase, i feel that will show a slight lack of confidence that they can bring the best chip to market.
If they do not push pre-orders, i wish them well and look forwards to a closely run contest.

[hdbuck]

I sold off the last of my holdings today at BTC0.098 for a sad loss, and based on what I've read from customers (not to mention what I've not read from friedcat&co. due to an utter lack of communication), I've set my expectations for lower future sales based on the current pricing strategy accordingly.

Should have waited, could have sold for 50% more. AM1 sits at 0.15BTC/share as we speak.

I have a feeling someone was waiting for the type of volume I laid down...reeks of market manipulation.  Again, not to sound like RoadStress et al., but just read the Prisma thread to see where you should be pricing shares.  None of those ~0.15BTC asks were up earlier this morning...looks like someone cleared out the ask sheet to try and pick off a few gullible newbs for a profit.

Or.. Somebody knows something..  

well well.. it seems somebody did know something afterall  

[FUR11]

Don't stick to solely selling chips! Try and do everything: Build and sell miners, sell chips, self-mine en masse, and sell cloud mining. Of course please delegate those things to competent people, maybe pursue the coalition with Rockminer or hire some new people who take care of the customers and sell the products so FC and the core-team can focus on the technical issues and deploy the hash rate.

[Chris_Sabian]

Don't stick to solely selling chips! Try and do everything: Build and sell miners, sell chips, self-mine en masse, and sell cloud mining. Of course please delegate those things to competent people, maybe pursue the coalition with Rockminer or hire some new people who take care of the customers and sell the products so FC and the core-team can focus on the technical issues and deploy the hash rate.

Self-mine and cloud mine is the way to go.  If you cloud mine, then you don't have to deal with shipping hardware 1/2 across the planet, waiting for hardware to arrive, and/or fix bad hardware remotely.  With cloudmining, you have all of your equipment in China/Hong Kong were you can easily fix, maintain, and upgrade easily.

[sngwinner]

Please please please start self mining with these chips en mass. Retain a healthy amount of bitcoin for future product development and then start distributing dividends again.

[Dexter770221]

Performance of new toy is nicely climbing 7.2GH with 0.25W/GH. And thats are MPW samples, friedcat said that mass production chips should do better...
New miners will be cold...

[MichaelBliss]

invariably, as we all know, this industry is never a one-horse race. All good points made above, but for this GEN however, it may well boil down to who gets first out the traps. That said, having samples to test does no way go the length to offer a best buy, just yet.

interesting times ahead.

I think AM will be first on the scene with their 4th gen chips, with Spondoolies about a month or two behind with their next gen.

Update
14nm/16nm projects are at pre-evaluation and pre-design stage. It has to be done
in 2015 but the starting time of placing orders depends on overall gain vs (NRE/R&D/risks).
...
The mass production time of BE300 in terms of chip-out date is February to March, 2015.

From this CCN interview with Spondoolies:

Could you tell CCN about what we can expect in the future from Spondoolies-Tech?

We have a very ambitious schedule in front of us.  In Q1, we will bring our next 28nm chip.  It has very ambitious specifications. Then by the end of q2 we will have our 16nm chips, with out of dimension specification. We are looking to be the most significant manufacturer in the SHA-256 industry by the end of 2015.

So, it seems like AM will be first out with their next gen but it looks like Spondoolies or KNC will likely be first to 16nm.

A bird in the hand is worth two in the bush.  About the "competition": Spoondoolies has the added liability of being based in Israel/Palestine, with all of the stability and ethics that entails.   I think it's funny peeps keep bringing up Spoondoolies as fierce competition, without mention that they are doing business out of a war-torn, failed state.  How many work days do they lose when they are cowering under cars etc from the Palestinian fireworks? That should factor, especially given all the shots at Havelock being located in Panama....

I wouldn't worry too much about Afghanistan based competition either.  Same reasons.

[raskul]

invariably, as we all know, this industry is never a one-horse race. All good points made above, but for this GEN however, it may well boil down to who gets first out the traps. That said, having samples to test does no way go the length to offer a best buy, just yet.

interesting times ahead.

I think AM will be first on the scene with their 4th gen chips, with Spondoolies about a month or two behind with their next gen.

Update
14nm/16nm projects are at pre-evaluation and pre-design stage. It has to be done
in 2015 but the starting time of placing orders depends on overall gain vs (NRE/R&D/risks).
...
The mass production time of BE300 in terms of chip-out date is February to March, 2015.

From this CCN interview with Spondoolies:

Could you tell CCN about what we can expect in the future from Spondoolies-Tech?

We have a very ambitious schedule in front of us.  In Q1, we will bring our next 28nm chip.  It has very ambitious specifications. Then by the end of q2 we will have our 16nm chips, with out of dimension specification. We are looking to be the most significant manufacturer in the SHA-256 industry by the end of 2015.

So, it seems like AM will be first out with their next gen but it looks like Spondoolies or KNC will likely be first to 16nm.

A bird in the hand is worth two in the bush.  About the "competition": Spoondoolies has the added liability of being based in Israel/Palestine, with all of the stability and ethics that entails.   I think it's funny peeps keep bringing up Spoondoolies as fierce competition, without mention that they are doing business out of a war-torn, failed state.  How many work days do they lose when they are cowering under cars etc from the Palestinian fireworks? That should factor, especially given all the shots at Havelock being located in Panama....

I wouldn't worry too much about Afghanistan based competition either.  Same reasons.

I fear your political views may be a tad biased.
all it would take would be for China to finish that road (tank route) to Tibet and your safe haven of industry goes up in smoke.

you clearly know not much about the the real political and military situation in China, and if you are tarnishing a reputable company with the evils of it's government, then you perhaps should move to the moon, where politics won't affect you.

what a ridiculous post to make!

[freedomno1]

Gen4 reporting on board:
https://bitcointalk.org/index.php?topic=888260.0

But why bother to announce it on hardware section since friendcat don't want to provide samples. Maybe there would be more miners based on BE200 not only foulty prismas...

Just noticed this myself when I went to check the hardware section
Anyways it's still in testing but some nice image porn in there.

[keystroke]

So how long until ASIC manufacture is lock step with Moore's law? Intel is fabbing at 14nm now. Looks like we will get there sometime in 2015?

[stompysteve]

for those that dont look at the hardware thread

Till now we received enough returned Prismas to get a statistical conclusion.
We decide that the old Prisma design has some flaws in both firmware and hardware.
Therefore manually fixing them and shipping them back take too long and are
not good for both our customers and us.

So the customers can choose one in the two offerings below:

1. We buy back the Prismas at the original price the customers pay.
The customers only need to ship the controller and the hashing boards back.
The fans and heatsinks are not needed. Customers who already shipped them back
can also switch to this option.

2. We use revised version of Prismas based on our new power management chips
to replace old ones, instead of returning old designs or replacing. This version is very
stable and has no problem of hashrate loss/overheat/can't identify boards/etc.
The compensation method for mining time loss is the same as before.

We cover shipping fee of both, if you can provide the shipping label.

We are very sorry for all the problems brought with the old version Prisma. The revised Prisma with our
own power management chips is much more stable. So if you are still able to put some trust on it and
accept the waiting of compensation (it accumulates with each delayed date anyway), you can choose
the second.

Please email us to sale@bitquan.com with the following information and format:

* Personal info,shipping address,tel number
* Date of buying our Prisma
* Your previous individual/exclusive btc payment address
* Number of broken Prsima boards (have to be fit well with the number you shipped back)
* Date you received your miners,date of shipping back
* The picture of the shipping label with the shipping price on it
* Your btc receivable address. it will be best to be the one you paid to us before. If the btc
receiving address is different, please sign the message with your address used for payment if you could.

For customers who already shipped back, please mail to us again (sale@bitquan.com) about the choice
you want to take.

Phasebird will reply afterwards with the new shipping address within US and EU. It
allows quicker turn-around time.

[chriswilmer]

for those that dont look at the hardware thread

Till now we received enough returned Prismas to get a statistical conclusion.
We decide that the old Prisma design has some flaws in both firmware and hardware.
Therefore manually fixing them and shipping them back take too long and are
not good for both our customers and us.

So the customers can choose one in the two offerings below:

1. We buy back the Prismas at the original price the customers pay.
The customers only need to ship the controller and the hashing boards back.
The fans and heatsinks are not needed. Customers who already shipped them back
can also switch to this option.

2. We use revised version of Prismas based on our new power management chips
to replace old ones, instead of returning old designs or replacing. This version is very
stable and has no problem of hashrate loss/overheat/can't identify boards/etc.
The compensation method for mining time loss is the same as before.

We cover shipping fee of both, if you can provide the shipping label.

We are very sorry for all the problems brought with the old version Prisma. The revised Prisma with our
own power management chips is much more stable. So if you are still able to put some trust on it and
accept the waiting of compensation (it accumulates with each delayed date anyway), you can choose
the second.

Please email us to sale@bitquan.com with the following information and format:

* Personal info,shipping address,tel number
* Date of buying our Prisma
* Your previous individual/exclusive btc payment address
* Number of broken Prsima boards (have to be fit well with the number you shipped back)
* Date you received your miners,date of shipping back
* The picture of the shipping label with the shipping price on it
* Your btc receivable address. it will be best to be the one you paid to us before. If the btc
receiving address is different, please sign the message with your address used for payment if you could.

For customers who already shipped back, please mail to us again (sale@bitquan.com) about the choice
you want to take.

Phasebird will reply afterwards with the new shipping address within US and EU. It
allows quicker turn-around time.

Friedcat is a standup guy.

[John (John K.)]

for those that dont look at the hardware thread

Till now we received enough returned Prismas to get a statistical conclusion.
We decide that the old Prisma design has some flaws in both firmware and hardware.
Therefore manually fixing them and shipping them back take too long and are
not good for both our customers and us.

So the customers can choose one in the two offerings below:

1. We buy back the Prismas at the original price the customers pay.
The customers only need to ship the controller and the hashing boards back.
The fans and heatsinks are not needed. Customers who already shipped them back
can also switch to this option.

2. We use revised version of Prismas based on our new power management chips
to replace old ones, instead of returning old designs or replacing. This version is very
stable and has no problem of hashrate loss/overheat/can't identify boards/etc.
The compensation method for mining time loss is the same as before.

We cover shipping fee of both, if you can provide the shipping label.

We are very sorry for all the problems brought with the old version Prisma. The revised Prisma with our
own power management chips is much more stable. So if you are still able to put some trust on it and
accept the waiting of compensation (it accumulates with each delayed date anyway), you can choose
the second.

Please email us to sale@bitquan.com with the following information and format:

* Personal info,shipping address,tel number
* Date of buying our Prisma
* Your previous individual/exclusive btc payment address
* Number of broken Prsima boards (have to be fit well with the number you shipped back)
* Date you received your miners,date of shipping back
* The picture of the shipping label with the shipping price on it
* Your btc receivable address. it will be best to be the one you paid to us before. If the btc
receiving address is different, please sign the message with your address used for payment if you could.

For customers who already shipped back, please mail to us again (sale@bitquan.com) about the choice
you want to take.

Phasebird will reply afterwards with the new shipping address within US and EU. It
allows quicker turn-around time.

His credibility and reputation in my eyes just grew tenfold.