Which ASIC device will ship first? [Poll and Discussion]

[Merrick]

Trouble is BFL has been expected to ship first all along. If they aren't first then someone's ROI will be measured in days or weeks and BFL's customers' ROI will be lengthened considerably.

I ordered late, so my ROI is already much longer. I like what I've seen from them, but if I were to order another, I'd consider hedging my bets with one from another manufacturer as well.

[Unacceptable]

Its a little late to hedge any bets,if you haven't already 

All preorders (first batch's) are full up from all ASIC manufacturer's,so you'll still be "late to the party" 

[LiteBit]

As of yesterday 10.30.2012: (from BFL_Josh on BFL Forums)

Guys (and gals), I am really sorry things are delayed. I want these things to ship more than anyone else (seriously, I promise you, I want these to ship) and the delays are just as hurtful and awful to us as they are to you. We would love nothing more than to ship them all out yesterday.

The first batch should be about 20,000 chips or so. More than enough for all pre-orders and then some. With regards to bASIC, given the fact that Tom has shown basically nothing, no renders, no prototype, not even any power estimates, it seems exceedingly unlikely he will ship before us. I'm not saying it's impossible of course, but there are many "fishy" things about the bASIC offering. Of course, the whole power issue, which Tom refuses to acknowledge is one of the biggest clues that things are not right in the bASIC world. If he were as far along as he implies, he would at least have a fairly reasonable and accurate power estimate, but his little "slip" of less than 100w for 54 GH/s is pretty unrealistic at 90nm, especially if it's Cell ASIC or something similar. Again, I'm not saying it's impossible, maybe Tom has come up with major surprise and be able to ship on his currently stated timeline, but it seems unlikely at this point. Be that as it may, I absolutely promise we will be shipping our units as fast as we possibly can.

For the record, we do not use TSMC, which I suspect Tom uses. I really wish the Bitcoin Magazine article would come out so I can talk more openly about our technology /sigh. Even if Tom were to ship a few days before us (which seems exceedingly unlikely at the moment), the long term ROI on our units vs the bASIC is vastly superior, so you'd make up the difference in the long run. Is that ideal? No, of course not, but it does make a difference, no matter how much the bASIC people try to convince themselves that power efficiency doesn't matter. A year or two down the road when your Bitforce unit is still earning money, the bASIC owners will have a sad panda face when they look at their worthless hardware that costs more to power than it earns and thus has absolutely no use what so ever. Then, on top of that, you'll likely be able to send us your old unit for credit towards a new unit.

Just sayin'

And this little nugget: (from BFL_Josh on BFL Forums)

Some aspects of the chip production ended up taking longer than expected to get rolling, which is why we have ended up paying for a bullet run and we're looking at ~25 days or so. We still intend to do the 1/3 shipping plan, I'm just saying we are getting that many chips in our first production batch... as in, we are getting enough to satisfy a lot of orders, but obviously we can't physically assemble 20,000 chips worth of orders in a few days. Not without hiring like 40 more people at any rate.

[scrybe]

But until something else has been established you guys have to at least acknowledge that doing business with BFL and their imitators is significantly more risky than lets say buying socks.

Have you seen the resale value on used socks lately? Horrible investment advice

[scrybe]

As of yesterday 10.30.2012: (from BFL_Josh on BFL Forums)

Guys (and gals), I am really sorry things are delayed. I want these things to ship more than anyone else (seriously, I promise you, I want these to ship) and the delays are just as hurtful and awful to us as they are to you. We would love nothing more than to ship them all out yesterday.

The first batch should be about 20,000 chips or so. More than enough for all pre-orders and then some. With regards to bASIC, given the fact that Tom has shown basically nothing, no renders, no prototype, not even any power estimates, it seems exceedingly unlikely he will ship before us. I'm not saying it's impossible of course, but there are many "fishy" things about the bASIC offering. Of course, the whole power issue, which Tom refuses to acknowledge is one of the biggest clues that things are not right in the bASIC world. If he were as far along as he implies, he would at least have a fairly reasonable and accurate power estimate, but his little "slip" of less than 100w for 54 GH/s is pretty unrealistic at 90nm, especially if it's Cell ASIC or something similar. Again, I'm not saying it's impossible, maybe Tom has come up with major surprise and be able to ship on his currently stated timeline, but it seems unlikely at this point. Be that as it may, I absolutely promise we will be shipping our units as fast as we possibly can.

For the record, we do not use TSMC, which I suspect Tom uses. I really wish the Bitcoin Magazine article would come out so I can talk more openly about our technology /sigh. Even if Tom were to ship a few days before us (which seems exceedingly unlikely at the moment), the long term ROI on our units vs the bASIC is vastly superior, so you'd make up the difference in the long run. Is that ideal? No, of course not, but it does make a difference, no matter how much the bASIC people try to convince themselves that power efficiency doesn't matter. A year or two down the road when your Bitforce unit is still earning money, the bASIC owners will have a sad panda face when they look at their worthless hardware that costs more to power than it earns and thus has absolutely no use what so ever. Then, on top of that, you'll likely be able to send us your old unit for credit towards a new unit.

Just sayin'

And this little nugget: (from BFL_Josh on BFL Forums)

Some aspects of the chip production ended up taking longer than expected to get rolling, which is why we have ended up paying for a bullet run and we're looking at ~25 days or so. We still intend to do the 1/3 shipping plan, I'm just saying we are getting that many chips in our first production batch... as in, we are getting enough to satisfy a lot of orders, but obviously we can't physically assemble 20,000 chips worth of orders in a few days. Not without hiring like 40 more people at any rate.

Thanks for the info. For all the forecasters out there we just got some ASIC shipping volume confirmation! (20,000 chips, enough to cover all pre-orders and then some)

Assuming all chips were put into singles and run at 7.5GH/s (60/8=7.5) then we are looking at a grand total of 150TH of total capability from this initial batch at release speeds.

Given that we have heard over and over about multiple batches of preorders, I'm assuming that all 20k will be not be hitting the streets at once but will be spread out over a 2-3 months.

[SgtSpike]

As of yesterday 10.30.2012: (from BFL_Josh on BFL Forums)

Guys (and gals), I am really sorry things are delayed. I want these things to ship more than anyone else (seriously, I promise you, I want these to ship) and the delays are just as hurtful and awful to us as they are to you. We would love nothing more than to ship them all out yesterday.

The first batch should be about 20,000 chips or so. More than enough for all pre-orders and then some. With regards to bASIC, given the fact that Tom has shown basically nothing, no renders, no prototype, not even any power estimates, it seems exceedingly unlikely he will ship before us. I'm not saying it's impossible of course, but there are many "fishy" things about the bASIC offering. Of course, the whole power issue, which Tom refuses to acknowledge is one of the biggest clues that things are not right in the bASIC world. If he were as far along as he implies, he would at least have a fairly reasonable and accurate power estimate, but his little "slip" of less than 100w for 54 GH/s is pretty unrealistic at 90nm, especially if it's Cell ASIC or something similar. Again, I'm not saying it's impossible, maybe Tom has come up with major surprise and be able to ship on his currently stated timeline, but it seems unlikely at this point. Be that as it may, I absolutely promise we will be shipping our units as fast as we possibly can.

For the record, we do not use TSMC, which I suspect Tom uses. I really wish the Bitcoin Magazine article would come out so I can talk more openly about our technology /sigh. Even if Tom were to ship a few days before us (which seems exceedingly unlikely at the moment), the long term ROI on our units vs the bASIC is vastly superior, so you'd make up the difference in the long run. Is that ideal? No, of course not, but it does make a difference, no matter how much the bASIC people try to convince themselves that power efficiency doesn't matter. A year or two down the road when your Bitforce unit is still earning money, the bASIC owners will have a sad panda face when they look at their worthless hardware that costs more to power than it earns and thus has absolutely no use what so ever. Then, on top of that, you'll likely be able to send us your old unit for credit towards a new unit.

Just sayin'

And this little nugget: (from BFL_Josh on BFL Forums)

Some aspects of the chip production ended up taking longer than expected to get rolling, which is why we have ended up paying for a bullet run and we're looking at ~25 days or so. We still intend to do the 1/3 shipping plan, I'm just saying we are getting that many chips in our first production batch... as in, we are getting enough to satisfy a lot of orders, but obviously we can't physically assemble 20,000 chips worth of orders in a few days. Not without hiring like 40 more people at any rate.

Thanks for updating us with the update...  

I first thought that he meant the first batch to be assembled would be all of the preorders..!

Lol, scrybe beat me to it.

Interesting though.  If 20,000 chips are more than enough for the preorders, then let's assume the preorders are 15,000 chips.  15,000 chips @ 8gh/s each is 120TH/s.  That would fit right in on the low side of estimates people have been making.

I like it.  The singles would still carry an ROI of a few months.  Might have to order more...

[MrTeal]

I thought Tom has repeatedly said that he isn't assembling the bASIC units himself. He's going to get them from the assembly house, flash and test them, the package them for shipping. Managing inventories should be much simpler than for BFL.

If I had to guess, I'd say his margins are likely to be much lower than BFL's, but that's not really much of a concern for us as customers. If true it might come into play more in half a year when ASIC vendors start slashing the price of their units to keep them moving.

[creativex]

I thought Tom has repeatedly said that he isn't assembling the bASIC units himself. He's going to get them from the assembly house, flash and test them, the package them for shipping. Managing inventories should be much simpler than for BFL.

If I had to guess, I'd say his margins are likely to be much lower than BFL's, but that's not really much of a concern for us as customers. If true it might come into play more in half a year when ASIC vendors start slashing the price of their units to keep them moving.

This is my understanding as well. No included case(or PSU?) further reduces bASIC construction times and potential for further delay(see BFL minirig screen). It seems as though Tom's process is about as streamlined as such a complex project can get, though at the cost of margins.

Speaking strictly for myself, as someone with multiple nekked mobos running on desktops, DAMN the case...just put the working board in a box and ship me that suckah!

[scrybe]

I first thought that he meant the first batch to be assembled would be all of the preorders..!

Lol, scrybe beat me to it.

Sorry about that, I I was so happy to see new data I could not resist!

Don't confuse ASIC FAB Batch from Assembly/shipping Batch. 20,000 is a pretty small order for a fab, but it's huge for in-house assembly.

I'm betting on a daily limit of 100-200 boards (750-1500 ASICs) for at least the first week, and it could be substantially lower. The first shipping batch (1/3rd's) is supposed to be able to go out 1 week after they get the ASIC's, we'll see what volume they hit.

Personally I think Tom made a smart bet outsourcing almost the whole thing, but I understand BFL has had issues with that mode before, so hopefully they will do something cool with their in-house capability.

We might actually see the scenario where BFL "ships first" by putting out daily production, but Tom gets his whole first batch out before BFL is done with theirs.

[LiteBit]

Speaking strictly for myself, as someone with multiple nekked mobos running on desktops, DAMN the case...just put the working board in a box and ship me that suckah!

Amen Brotha!

[abeaulieu]

There are a lot of people who marked other on the poll. Are you who marked this skeptical of ASICs shipping at all or is there another that you're more confident in?

[creativex]

ASICMiner investors I guess, though I thought they were planning to mine with their creations first.

[Blackasaurus]

Anagrams....

bASIC = ISCAB
Butterfly Labs = Beta Blurts Fly
AVALON = ?
ASICMINER = Re Scam I In

AVALON = NO LAVA

IT ALL MAKES SENSE!
Scams, scams everywhere!

[scrybe]

ASICMiner investors I guess, though I thought they were planning to mine with their creations first.

Good point, the poll is on shipping, not mining.

Also, am I the only one that keeps typing ASCIIMINER by mistake?

[jjiimm_64]

Speaking strictly for myself, as someone with multiple nekked mobos running on desktops, DAMN the case...just put the working board in a box and ship me that suckah!

Amen Brotha!

triple amen to this...  bugs and dust be damned, let me plug the damn thing in.

[Odi]

I thought Tom has repeatedly said that he isn't assembling the bASIC units himself. He's going to get them from the assembly house, flash and test them, the package them for shipping. Managing inventories should be much simpler than for BFL.

If I had to guess, I'd say his margins are likely to be much lower than BFL's, but that's not really much of a concern for us as customers. If true it might come into play more in half a year when ASIC vendors start slashing the price of their units to keep them moving.

Yup, this is what I read when I was deciding who to pre-order with in mid October, everything except firmware loading and final testing is outsourced, which will reduce his margins but make things much easier for him to manage compared to his FPGA products:

What has changed?

(a) The ModMiner quad goes through a lot of steps that take time
- The PCB boards are fabricated in one place in china
- The FPGA Chips are purchased from another place
- Everything is shipped to me and then
- Everything is shipped to the assembly house
- I purchase some final parts and have them shipped here
- Everything gets nearly assembled and shipped here
- I put them together and test them, package them up myself and ship them out myself
- All while trying to handle support issues, customer service and the logistics of the whole thing

So with the bASIC I decided to streamline a lot of these things, for one I have a very good relationship with the place that assembles the ModMinerQuad and they have taken over a lot more responsibility with this project:
- Their engineers are designing the final production board
- The PCB's will be fabricated by them
- The Boards will be assembled by them to completion


Then final boards will be shipped to me


And here is where I have made another change and upgraded my operation:

I have also leased a 2500 square foot office space / store front which I am in the process of moving into and will be hiring full time employees to help when the bASIC units ship out, their duties will include  opening up boxes from the assembly house - loading firmware - testing for 24 hours - boxing them back up and shipping them out to customers.

I was going to wait and announce this later this week but now seems like a fitting time:

If anyone lives in the Central/Northern NY area (has experience enough and time to do this job efficiently/effectively) and would like to apply for a job please call me any time to arrange an appointment (seriously)

when the bASIC units come in from the assembly house you will get a lot of overtime - we will probably need about 5 people working 8-12 hours a day, 6 days a week.

As of yesterday 10.30.2012: (from BFL_Josh on BFL Forums)

With regards to bASIC, given the fact that Tom has shown basically nothing, no renders, no prototype, not even any power estimates, it seems exceedingly unlikely he will ship before us. I'm not saying it's impossible of course, but there are many "fishy" things about the bASIC offering. Of course, the whole power issue, which Tom refuses to acknowledge is one of the biggest clues that things are not right in the bASIC world. If he were as far along as he implies, he would at least have a fairly reasonable and accurate power estimate, but his little "slip" of less than 100w for 54 GH/s is pretty unrealistic at 90nm, especially if it's Cell ASIC or something similar. Again, I'm not saying it's impossible, maybe Tom has come up with major surprise and be able to ship on his currently stated timeline, but it seems unlikely at this point.

I think that BFL_Josh's statement is not accurate in that particular paragraph I left quoted.  Tom claimed to have a prototype using Cell ASIC from a university wafer run before even announcing his offering, but made no statements about his final ASICs other than being on a 90nm process:

anyways here are some more specific details for those who are interested on how BTCFPGA produced our first 4th gen prototype

Like the ModMiner quad I leased the design from someone else. Also like the ModMiner quad this project is being partially funded by lenders in the Bitcointalk lending syndicate. We contracted a well known overseas company to provide us with a SHA256 hashing core that could be applied to Bitcoin mining, this company is one of the top design firms in the world when it comes to SHA256 IP cores. We leased what is called an RTL design which allows us to use any target ASIC process technology of our choice, once we had the core design we spent some money to have some made (this took some time) and turned those over to the hardware development team. We have two guys which I have no problem admitting are much smarter than I am who have come up with a working development board using multiple instances of the hashing core I indicated above. They have a very very (did I mention very?) primitive hacked together mining program that is being used to test, testing has been going on for about 2 weeks now with positive enough results that encouraged me to go public with our next generation of Bitcoin mining product. The hardware guys are still working on finalizing the pcb board design and there is a lot of firmware and software work that needs to be completed.

A lot of people are asking about specific details about the chips and also about power usage, and the only reason I am at all reluctant to answer these questions is because I dont want to say something now and be wrong about or have it change.

These would be considered Cell Based ASIC

I am not going into specific details about certain aspects of the prototype asics / mining unit because some of this may change from prototype to production

Please note I do not claim to be the technical mind behind all of this, this protoype has been achieved through leases, and contractors

the SHA256 hashing core was (not designed by me) it was leased through a overseas design firm

the prototype asics themselves were made in a multi project wafer run program through a university

we are still shopping around for a long term manufacturer of these asics

the mining units will be assembled at the same place that assembles the modminer quad.

BTCFPGA went with an IP design, made some prototype chips in a university run, hacked together a prototype, then outsourced the whole thing to fabricate and assemble.

[creativex]

I think that BFL_Josh's statement is not accurate in that particular paragraph I left quoted.

What's new. Tell me when that guy says something that is accurate. Not to worry, I won't be holding my breath.