ASICMINER: Entering the Future of ASIC Mining by Inventing It

[jimmothy]

Does anyone know the specs of gen1 chips vs gen3? What I want to know is what we can expect the solomining hashrate to be if every current gen chip was swapped with 28nm.

[MrTeal]

Though interpreting this... The cubes are nearly sold all... so the next divs might be a bit higher because of the sales income but then there is nothing moving until mid february? I mean january the 20th tapeout. Producing miners take probably 2 more weeks, probably 3. Then we are mid february. Sounds to me like no hope for the shareprice. It will go down and down the next months lower than ever.

I think you're being a little aggressive if you think it will be 2-3 weeks from tape-out to working miners.

I wanted to go higher first but then i thought that friedcat most probably will know when the chips will be ready from the foundry so that he can order the timeframe at the assembler beforehand. No waiting for the place in queue. Of course the development of the miner has to be done beforehand too.

Though i think about it... doesnt tapeout mean thats the first chip too? In that case the miner design has to be tested first because otherwise it would be a too high risk to produce the miners without doing that... which means it can easily take way longer then...

What do you guys think when the divs from the cubes will come in?

I wish friedcat would have said a word about the divs he keeps. Is he still keeping?

No, tape-out means that they send the design to the foundry, who then start the process of turning raw silicon wafers into ASICs. They won't even have the wafers finished 3 weeks after tape-out, let alone have miners built.

[stripykitteh]

http://en.wikipedia.org/wiki/Tape-out

Ready to burn chips.

No, that's not right. Tape-out only means the design is finished. The design is sent 'on tape' (now it would be emailed I guess) to the foundry to make the masks. I believe making masks takes several weeks, though someone with IC manufacturing experience (Vycid?) may correct me on this. Making the masks is a big ticket item; you have to book out the specialized photolithography equipment for weeks to have the mask built, one layer at a time.

So if AM hit their target on tape-out (January 20) they won't have masks until late February at the earliest. Sample chips early March, chips in volume late March at the earliest. Allowing for slippage here and there it's probably more realistic to say April for chips in volume.

It's good that friedcat gave us an update, but the news isn't that startling, in my opinion. In fact, I sold on the news.

[SebastianJu]

Though interpreting this... The cubes are nearly sold all... so the next divs might be a bit higher because of the sales income but then there is nothing moving until mid february? I mean january the 20th tapeout. Producing miners take probably 2 more weeks, probably 3. Then we are mid february. Sounds to me like no hope for the shareprice. It will go down and down the next months lower than ever.

I think you're being a little aggressive if you think it will be 2-3 weeks from tape-out to working miners.

I wanted to go higher first but then i thought that friedcat most probably will know when the chips will be ready from the foundry so that he can order the timeframe at the assembler beforehand. No waiting for the place in queue. Of course the development of the miner has to be done beforehand too.

Though i think about it... doesnt tapeout mean thats the first chip too? In that case the miner design has to be tested first because otherwise it would be a too high risk to produce the miners without doing that... which means it can easily take way longer then...

What do you guys think when the divs from the cubes will come in?

I wish friedcat would have said a word about the divs he keeps. Is he still keeping?

No, tape-out means that they send the design to the foundry, who then start the process of turning raw silicon wafers into ASICs. They won't even have the wafers finished 3 weeks after tape-out, let alone have miners built.

Hm... somehow i thought taping out means its done then... i really would like to hear what friedcat plans for the meantime... i mean he has a datacentre with immersion cooling for what?

photolithography

I believe at 28nm you cant use photolithography anymore. It should be e-beam lithography then.

[Vycid]

http://en.wikipedia.org/wiki/Tape-out

Ready to burn chips.

No, that's not right. Tape-out only means the design is finished. The design is sent 'on tape' (now it would be emailed I guess) to the foundry to make the masks. I believe making masks takes several weeks, though someone with IC manufacturing experience (Vycid?) may correct me on this. Making the masks is a big ticket item; you have to book out the specialized photolithography equipment for weeks to have the mask built, one layer at a time.

So if AM hit their target on tape-out (January 20) they won't have masks until late February at the earliest. Sample chips early March, chips in volume late March at the earliest. Allowing for slippage here and there it's probably more realistic to say April for chips in volume.

It's good that friedcat gave us an update, but the news isn't that startling, in my opinion. In fact, I sold on the news.

This is a pretty sober analysis. Correct: there is a long lead time between tape out and chip delivery, especially at modern process nodes. The photomasks have not yet been made, although FC may already have paid for them.

photolithography

I believe at 28nm you cant use photolithography anymore. It should be e-beam lithography then.

You are wrong. These days we use immersion litho and double patterning, but e-beam is not cost effective for anything except samples and making the masks themselves.

[SebastianJu]

I believe at 28nm you cant use photolithography anymore. It should be e-beam lithography then.

You are wrong. These days we use immersion litho and double patterning, but e-beam is not cost effective for anything except samples and making the masks themselves.

Ah ok... the it was for the masks... at least thats part of taping out...

[KS]

More like ready In Two Weeks^(TM)

It could still take 10-20 weeks.

Or it could be never. Or all life on earth could be obliterated on News Years.

Seriously, Tapeout is a huge deal. It costs an unbelievable amount of money and effort to get there. It is possible FC could start reclaiming global hash percentages in March. A million years in the bitcoin universe to be sure. But obviously not something one can do in a few weeks. It takes a ton of planning and effort to be able to stand up in March and proclaim "we have 15% of the global hash" especially after everyone else has deployed the best that they have.

Seriously, Tape Out just means your chip is ready _in software_ and that's the code you send to the fab which then needs to process with the _other_ steps before even etching it on silicon.

It's an important milestone, but then having a full (working) mask is another and then your first run is yet another etc.

[KS]

...

You are wrong. These days we use immersion litho and double patterning, but e-beam is not cost effective for anything except samples and making the masks themselves.

Them engineers with them tricks. Can't even make a propah 28nm litho thingamagic.

[jimmothy]

More like ready In Two Weeks^(TM)

It could still take 10-20 weeks.

Or it could be never. Or all life on earth could be obliterated on News Years.

Seriously, Tapeout is a huge deal. It costs an unbelievable amount of money and effort to get there. It is possible FC could start reclaiming global hash percentages in March. A million years in the bitcoin universe to be sure. But obviously not something one can do in a few weeks. It takes a ton of planning and effort to be able to stand up in March and proclaim "we have 15% of the global hash" especially after everyone else has deployed the best that they have.

+1

First everyone begs FC to tell them when next gen will be here and when he gives them his best estimate it's still not enough to silence the FUD.

[Kouye]

...

Yo KS. I'm looking to buy more cheap shares, care to dump yours in my stash?

[KS]

...

Yo KS. I'm looking to buy more cheap shares, care to dump yours in my stash?

Nope.

edit: it's not like I'm trying to scare ppl into selling their shares, but Tape Out is really not as important as having a working mask, and when you do, you still have to book a run. Months.

[itamidensha]

Looking forward to the New Year then

That reminds me, The Chinese new year holiday is at the end of January. That might slow the production prep/process down a bit.
I was in China once over that period - it's frigging nuts! People were setting off fireworks in the streets for a whole week!

[jimmothy]

...

Yo KS. I'm looking to buy more cheap shares, care to dump yours in my stash?

Nope.

edit: it's not like I'm trying to scare ppl into selling their shares, but Tape Out is really not as important as having a working mask, and when you do, you still have to book a run. Months.

He could have estimated having a working mask in Feb and you would complain that it is not as important as having a working product. We have seen that AM can in fact build asics and deliver on their promises.

But why even base your investment on past history. Lets all just assume that after tape out FC is going blow it and or take unreasonably long. Because FUD.

[Vycid]

But why even base your investment on past history. Lets all just assume that after tape out FC is going blow it and or take unreasonably long. Because FUD.

I'm basing my investment decisions on present facts. For example, the fact that the price-to-book ratio is worse than it was in March when there was no real competition in sight. I need to make no assumptions about FC failing.

If you need to counter every bear argument with "what FUD" you might want to re-evaluate your own biases toward FC/ASICMiner. I assure you that if FC's news today was that the masks were done and they are currently producing Gen3 chips, my reaction would be quite different.

When I predicted 0.3 BTC/share a few months ago, I expected pretty much this exact situation. Nothing has changed.

Keep in mind that successful investing comes down to superior information. If you don't know anything about semiconductor manufacturing, you might want to reconsider if you really have an edge here.

[KS]

...

Yo KS. I'm looking to buy more cheap shares, care to dump yours in my stash?

Nope.

edit: it's not like I'm trying to scare ppl into selling their shares, but Tape Out is really not as important as having a working mask, and when you do, you still have to book a run. Months.

He could have estimated having a working mask in Feb and you would complain that it is not as important as having a working product. We have seen that AM can in fact build asics and deliver on their promises.

But why even base your investment on past history. Lets all just assume that after tape out FC is going blow it and or take unreasonably long. Because FUD.

Are you just trolling or sth?

Tape Out signals the end of the _design phase_. NOT the same as _manufacturing_ (fabricating/"fabbing"). When you have *all* the requisites for fabbing, you still need to book a run. A rocket run can be as little as 45 days when a regular run is 10-26 weeks.

So no, Tape Out in January doesn't mean "chips in 3 weeks".

edit: and when the chips are done in silicon, you need to cut them up and put them in a package etc. Please lurk MOAR.

[jimmothy]

But why even base your investment on past history. Lets all just assume that after tape out FC is going blow it and or take unreasonably long. Because FUD.

When I predicted 0.3 BTC/share a few months ago, I expected pretty much this exact situation. Nothing has changed.

Keep in mind that successful investing comes down to superior information. If you don't know anything about semiconductor manufacturing, you might want to reconsider if you really have an edge here.

Besides the fact that it is no longer selling at 0.3 btc/share and that was only for a day on a very small volume which in no way represents the actual value of the shares. Also ignoring the fact that the previous week was extremely unlucky mining and this week is looking to be nearly double the dividend of last week. Even on havelock where it was dipping down to 0.3 it is now trading at around 0.5btc.

I don't need experience about semiconductor manufacturing as history has already shown us that Asicminer does know very much about it. Tell me why you don't think AM will succeed in creating 28nm asics where as the competition can?

[Kouye]

Are you just trolling or sth?

How much do you want for your shares?

[jimmothy]

Tape Out signals the end of the _design phase_. NOT the same as _manufacturing_ (fabricating/"fabbing"). When you have *all* the requisites for fabbing, you still need to book a run. A rocket run can be as little as 45 days when a regular run is 10-26 weeks.

So no, Tape Out in January doesn't mean "chips in 3 weeks".

edit: and when the chips are done in silicon, you need to cut them up and put them in a package etc. Please lurk MOAR.

We get it. Nobody misunderstood that FC's estimation was not for completed 28nm hardware. What we do know is 28nm chips are being developed and this is great news. Just because you can point out every step of the manufacturing process does not take away from the fact that 28nm chips are coming. So you can stop saying that they wont have working 28nm hardware by January because we know this already.

[jimmothy]

Are you just trolling or sth?

How much do you want for your shares?

He doesn't have any shares, he is obviously trying to pick some up for cheap.

[Kouye]

he is obviously trying to pick some up for cheap.

This is what I was trying to do, too.