ASICMINER: Entering the Future of ASIC Mining by Inventing It

[Vycid]

Actually, I don't think that's one unknown. I think that's just a lot of people solo mining.

IPs like these, though, are probably guys who bought the illegitimate Avalon chips that were made by a disgruntled employee. Those shipped on time, unlike the legitimate ones.

http://blockchain.info/blocks/129.132.188.124

http://blockchain.info/blocks/75.102.1.116

 

Goddamn, it's like some shit out of a cyberpunk novel.

[felente]

there are not only avalons for now.

bitfury going live incrementally;
BFL is delivering as they should did it times ago;
sure, avalons too;

people are getting enough hash-power to form independent "unknown" pools.

[velacreations]

Interesting discussion on the ActiveMining thread about some of the "competition" out there in mining land.

This is a very interesting topic.   That is why I wonder.   It is highly likely that the network is 1,000 Th/s (1,000,000 GH/s) in 45 days from now.
So, say you want to be a player like friedcat and have 20%.   With avalon chips drawing 9W per Gh/s, you need 900 kw just to power the avalons to get 10% of the market.  How you going to do that?  lot's of buildings with that power coming in?   How you going to distribute the power?   Oh, and guess how much heat 900 kW of power puts out when it goes into something that basically just heats up?   But to get 20% you need 1800kW PLUS cooling.    Even at 1W, it is an enormous undertaking.    It will be cool to see how people over come these hurdles.   Someone is going to have 500 TH/s rolled out by November?   They best call the electric company now because ordering new transformers takes a while. 

What they are saying is right.  If you want 10% of the network in September, you will need almost 900kW of power lined up and ready to go.  I don't think people realize how big of a deal that really is to get that much power online in that time frame.  It's not going to be easy and/or cheap.

If you already have a datacenter and power infrastructure in place, then expanding is much easier than building from scratch.  AM's first mover advantage may last longer than anyone anticipates because of these limitations.

At the very least, you won't see a new entity taking 10% of the network in the next 2 months.

tl;dr - getting the chips is the easy part (and they still don't even have that)

[Vycid]

Interesting discussion on the ActiveMining thread about some of the "competition" out there in mining land.

This is a very interesting topic.   That is why I wonder.   It is highly likely that the network is 1,000 Th/s (1,000,000 GH/s) in 45 days from now.
So, say you want to be a player like friedcat and have 20%.   With avalon chips drawing 9W per Gh/s, you need 900 kw just to power the avalons to get 10% of the market.  How you going to do that?  lot's of buildings with that power coming in?   How you going to distribute the power?   Oh, and guess how much heat 900 kW of power puts out when it goes into something that basically just heats up?   But to get 20% you need 1800kW PLUS cooling.    Even at 1W, it is an enormous undertaking.    It will be cool to see how people over come these hurdles.   Someone is going to have 500 TH/s rolled out by November?   They best call the electric company now because ordering new transformers takes a while.  

What they are saying is right.  If you want 10% of the network in September, you will need almost 900kW of power lined up and ready to go.  I don't think people realize how big of a deal that really is to get that much power online in that time frame.  It's not going to be easy and/or cheap.

If you already have a datacenter and power infrastructure in place, then expanding is much easier than building from scratch.  AM's first mover advantage may last longer than anyone anticipates because of these limitations.

At the very least, you won't see a new entity taking 10% of the network in the next 2 months.

tl;dr - getting the chips is the easy part (and they still don't even have that)

A major player would just rent datacenter space. I fail to see the infrastructure problems, since datacenters already exist.

Here's a simple example - http://os-legacy-images.s3.amazonaws.com/uploads/Seattle/wastatedc/wastatedc.pdf. 2.25 MW critical load.

If I had $10M and I wanted to be a serious first mover, I'd probably talk to Cointerra about getting 50% of the hashrate on silicon and get a deal worked out there, rent a data center in Washington where electricity is very cheap (as low as $0.01/kW-h in parts of the state), and strike a deal with the Winklevoss ETF for volume coin sales.

[felente]

Interesting discussion on the ActiveMining thread about some of the "competition" out there in mining land.

What they are saying is right...

A major player would just rent datacenter space. I fail to see the infrastructure problems, since datacenters already exist.

yes, they exist, but are not intended to consume +/- 1MW energy in such a concentrated way.
there are not many rack-devices sucking several kilowatts at once + on a continuous basis + in such quantities(ie devices self) + THE cooling of ~1MW...
all that is no joke man! so there is a truth somewhere...

a major player maybe will need in the fact build / energetically wire / cool such a datacentre, or actually PAY for an upgrade of rented one.

EDIT:
checked the pdf now. nice datacenter.
but this is WASHINGTON STATE datacenter, right?
we should ask them to permit us mine BTC on more than a half of it's area

[velacreations]

A major player would just rent datacenter space.

hence, FC's franchising plans.

I fail to see the infrastructure problems, since datacenters already exist.

they don't spring up overnight, though.  And although renting a datacenter might seem like an easy option, that basically means that you are adding significant overhead to your mining costs, because no one is going to rent you a datacenter for the cost of power/infrastructure alone.  You're gonna have to pay a premium to get that level of service, regardless of where you go.

If I had $10M and I wanted to be a serious first mover, I'd probably talk to Cointerra about getting 50% of the hashrate on silicon and get a deal worked out there, rent a data center in Washington where electricity is very cheap (as low as $0.01/kW-h in parts of the state), and strike a deal with the Winklevoss ETF for volume coin sales.

and you're still looking at 2-3 months before you mine your first block, if everything goes right.

[velacreations]

a major player maybe will need in the fact build / energetically wire / cool such a datacentre, or actually PAY for an upgrade of rented one.

assuming it's cost effective to rent a datacenter like that for mining bitcoins

EDIT: looking at that pdf above to rent that datacenter, you are looking at $125/kW per month.  If you want 10% of the network in 45 days, like the example maths above, then that's 900kW, so $1,125,00 per month PLUS energy use (hopefully you can get the good rate of $0.01/kWh)

Of course, by the end of the month, you will probably have double the power consumption/rental cost if you try and maintain your percentage of the network.

So, the bigger question is, which IPOs are considering these issues in their estimates?

[DeathAndTaxes]

A major player would just rent datacenter space. I fail to see the infrastructure problems, since datacenters already exist.

Here's a simple example - http://os-legacy-images.s3.amazonaws.com/uploads/Seattle/wastatedc/wastatedc.pdf. 2.25 MW critical load.

Not at any reasonable costs.  Make sure to look into amperage details in datacenter contract.  Most datacenters drop 20A @ 120V (20% derate) = ~2.4W per rack.  Now you usually can get 20A @ 208V single phase for not much more which gives you ~4KW.  Going to 30A @ 208V bumps you into an entire different cost range and still "only" gives you 6KW per rack.  Datacenters are carefully designed to avoid hotspots so they really aren't wired for higher heat loads then that. Nobody is going to let you dump 20KW into a rack just because it "fits".  It might fit physically but it doesn't fit from a power or thermal standpoint.  If your heat or power load compromises other clients the datacenter is going to suffer massive fines and it is simply not a risk worth taking.  Plus the datacenter only has so much space, cooling capacity, and power.  They want to get top dollar on all three.  Running out of cooling or power capacity and having half the datacenter empty isn't going to make any beancounter happy.

Try pricing out what 150 racks with 30A 208V single phase costs.   I am not saying it can't be done but I think most people have casually looked around and saw "deals" for $400 for 40U and are greatly underestimating what a datacenter will charge for higher density loads.

[Vycid]

A major player would just rent datacenter space.

hence, FC's franchising plans.

I fail to see the infrastructure problems, since datacenters already exist.

they don't spring up overnight, though.  And although renting a datacenter might seem like an easy option, that basically means that you are adding significant overhead to your mining costs, because no one is going to rent you a datacenter for the cost of power/infrastructure alone.  You're gonna have to pay a premium to get that level of service, regardless of where you go.

If I had $10M and I wanted to be a serious first mover, I'd probably talk to Cointerra about getting 50% of the hashrate on silicon and get a deal worked out there, rent a data center in Washington where electricity is very cheap (as low as $0.01/kW-h in parts of the state), and strike a deal with the Winklevoss ETF for volume coin sales.

and you're still looking at 2-3 months before you mine your first block, if everything goes right.

Well, a major player would not be interested in FC's franchising - why would they? They'd be hurting their own margin for no real reason, since there are cheaper options for hardware. Franchising makes more sense for a little guy trying to get in on the action.

No, datacenters don't spring up overnight, but you don't need them to. Options exist already. The one I linked is a very nice facility, so it would probably not be the one actually used by a serious investor - it's just an example. All that is really needed is a temporary location while the difficulty is low - they can move their hardware to a purpose-built facility later (and Washington State is perfect for that, what with the $0.01/kW-h electricity).

You're right about 2-3 months, but I don't see that as an issue either. VCs will invest in startups and often wait 5 years or more before they see their first return.

[Vycid]

A major player would just rent datacenter space. I fail to see the infrastructure problems, since datacenters already exist.

Here's a simple example - http://os-legacy-images.s3.amazonaws.com/uploads/Seattle/wastatedc/wastatedc.pdf. 2.25 MW critical load.

Not at any reasonable costs.  Make sure to look into amperage details in datacenter contract.  Most datacenters drop 20A @ 120V (20% derate) = ~2.4W per rack.  Now you usually can get 20A @ 208V single phase for not much more which gives you ~4KW.  Going to 30A @ 208V bumps you into an entire different cost range and still "only" gives you 6KW per rack.  Datacenters are carefully designed to avoid hotspots so they really aren't wired for higher heat loads then that. Nobody is going to let you dump 20KW into a rack just because it "fits".  It might fit physically but it doesn't fit from a power or thermal standpoint.  If your heat or power load compromises other clients the datacenter is going to suffer massive fines and it is simply not a risk worth taking.  Plus the datacenter only has so much space, cooling capacity, and power.  They want to get top dollar on all three.  Running out of cooling or power capacity and having half the datacenter empty isn't going to make any beancounter happy.

Try pricing out what 150 racks with 30A 208V single phase costs.   I am not saying it can't be done but I think most people have casually looked around and saw "deals" for $400 for 40U and are greatly underestimating what a datacenter will charge for higher density loads.

Those are co-location "deals". Here, you'd be leasing out the whole damn datacenter. You're making the rules at that point.

Edit: Here's an upcoming datacenter in North Central Washington. http://datacenters.sabey.com/intergate_quincy

Electricity is $0.025/kW-h, with a cap of 48MW. Friedcat (Shenzen, Guangdong province) pays $0.06-0.07 and cannot compete with that.

All of this is just from a cursory google search, by the way. Someone seriously considering a move like this will find the best deal.

I'm honestly a little surprised that people think this can't be done. Do you think Friedcat is a magician or something? I'm sure he had to rent space.

[velacreations]

Those are co-location "deals". Here, you'd be leasing out the whole damn datacenter. You're making the rules at that point.

it's still going to cost a few million $$$ per month just for the datacenter, plus power use.  You can make the rules all you want, but they make the bill.

From your pdf:

High power density – 175+ watts per square foot

what does that translate to in hashrate per square foot?

[velacreations]

You're right about 2-3 months, but I don't see that as an issue either. VCs will invest in startups and often wait 5 years or more before they see their first return.

yeah, I don't disagree that on a longer time scale, it MIGHT make financial sense, but I think a lot of the "competition" to AM are ignoring some of these issues.  It's not as easy as just getting your chips, folks, and truth be told, getting your Avalon chips is probably the easiest part of the whole deal.

In the meantime, for those 2-3 months, AM already has a datacenter up and running, AM is producing and releasing more hardware, and AM is making money.

[Vycid]

Those are co-location "deals". Here, you'd be leasing out the whole damn datacenter. You're making the rules at that point.

it's still going to cost a few million $$$ per month just for the datacenter, plus power use.  You can make the rules all you want, but they make the bill.

You're mad. A few million per month? Where did you get that number? The WHOLE datacenter I linked (and you could probably sublease quite a lot of it) cost $35M to construct. If we assume a typical 10-year payback period for leasing, then that comes out to 120 months, or $3500000/120 ~ $300k/mo.

Not to mention that the state fucked up big time with this project and is willing to lease below market.

http://www.zdnet.com/washington-states-datacenter-fiasco-its-time-to-pay-the-bill-7000011382/

You're right about 2-3 months, but I don't see that as an issue either. VCs will invest in startups and often wait 5 years or more before they see their first return.

In the meantime, for those 2-3 months, AM already has a datacenter up and running, AM is producing and releasing more hardware, and AM is making money.

Yes, indeed. BUT it is trading at a price that assumes AM will continue to dominate the market for years.

This is why I am so bearish, and why I have been actively buying puts.

High power density – 175+ watts per square foot

what does that translate to in hashrate per square foot?

For the existing Avalon chips, about 30 GH/s per square foot.

[velacreations]

You're mad. A few million per month? Where did you get that number?

your pdf said $125/kW per month:

$125.00 per kW per month, all-in except power consumption

Based on the math I quoted from, 10% of the network would require 900 kW, so $1,125,000 per month.  Of course, that's not renting the whole space, and than only gets you 10% for a limited time (a week?).  If you want to keep that 10% rate for the whole month, you'll likely need double that, so $2M a month.

maybe I'm quoting the wrong thing, or the pdf had a mistake, but I assumed it was right.

[Vycid]

You're mad. A few million per month? Where did you get that number?

your pdf said $125/kW per month:

$125.00 per kW per month, all-in except power consumption

Based on the math I quoted from, 10% of the network would require 900 kW, so $1,125,000 per month.  Of course, that's not renting the whole space, and than only gets you 10% for a limited time (a week?).  If you want to keep that 10% rate for the whole month, you'll likely need double that, so $2M a month.

maybe I'm quoting the wrong thing, or the pdf had a mistake, but I assumed it was right.

Oh, right - that's electricity cost, not rent. Understand that Friedcat is on the hook for that too.

That's the biggest issue with that datacenter. It's in Washington, but the electricity isn't steeply discounted. If you want to see why Friedcat is (in my opinion) in serious trouble, look at the link in this post: https://bitcointalk.org/index.php?topic=99497.msg2954417#msg2954417

Like I've said before, I'm not spending the time to find the best datacenter. I'm sure there are even better deals.

Edit: Here's the PDF from the North Central Washington datacenter. http://datacenters.sabey.com/docs/properties/profiles/intergate_quincy_profile.pdf

[Juzzers]

Hi all, I've been trying to transfer my shares to ASICMINER-PT on btct.co and have sent a request to friedcat and burnside several times over the last 2 weeks and have received no reply.

Have I missed a message about some new process for transferring shares?

Any help much appreciated.

[velacreations]

Oh, right - that's electricity cost, not rent. Understand that Friedcat is on the hook for that too.

What makes you say that?  The line says "all-in except power consumption".  I would assume that to be a rental cost. Electricity cost is typically depicted in $/kWh

[Vycid]

Oh, right - that's electricity cost, not rent. Understand that Friedcat is on the hook for that too.

What makes you say that?  The line says "all-in except power consumption".  I would assume that to be a rental cost. Electricity cost is typically depicted in $/kWh

I see what you're saying. I think you're right, and I didn't read that PDF correctly. That would make this particular data center a poor prospect, since the ASICs are very heavy on power consumption. It's possible you could negotiate the price down, though, since you'd be using less floor space than they'd expect per square foot.

Still, the Integrate.Quincy datacenter is another example of what I'm talking about. That's a 52MW facility, easily big enough to accommodate a project like this, and they're leasing now. Power is $0.0225/kW-h.

The particular datacenter isn't the point here - just that this can be done, and done cheaper than AM has managed. Believe it or not, China (particularly Guangdong) is the wrong location. Big business mining will end up in Washington.

If I had the money, I would seriously consider doing this project myself. Shit, these IPOs have been so easy to pull off, maybe I can...

[lophie]

Hi all, I've been trying to transfer my shares to ASICMINER-PT on btct.co and have sent a request to friedcat and burnside several times over the last 2 weeks and have received no reply.

Have I missed a message about some new process for transferring shares?

Any help much appreciated.

planning on doing that as well. Any pointers guys?

[velacreations]

I see what you're saying. I think you're right, and I didn't read that PDF correctly. That would make this particular data center a poor prospect, since the ASICs are very heavy on power consumption. It's possible you could negotiate the price down, though, since you'd be using less floor space than they'd expect per square foot.

possibly, but I think this illustrates the point that it is neither cheap nor as easy as first expected to launch something of the size and power consumption that we are talking about.

None of the IPOs I have seen are taking this into serious consideration, or at least, they are not voicing how they are going to meet these obstacles.  Everyone seems to be focused on getting chips/hardware, which is a major obstacle in and of itself, but meanwhile, they are ignoring the real feat of housing/powering that hardware for a reasonable cost. 

These kinds of obstacles are not fixed overnight.  Even if you have the data center lined up, have your hardware in hand, everything ready to go, getting 100 TH/s online is going to take weeks, at best.

Still, the Integrate.Quincy datacenter is another example of what I'm talking about. That's a 52MW facility, easily big enough to accommodate a project like this, and they're leasing now. Power is $0.0225/kW-h.

yeah, that's a nice datacenter, but what does it cost?  As we saw above, the power cost is miniscule, renting the actual datacenter is several orders of magnitude more expensive.  I think you might be underestimating what top of the line datacenters like this lease for.  I do agree that there are cheaper datacenters out there.

In any case, the rent is going to be a much larger cost than power, and may even make it cost prohibitive to rent a datacenter like that.

Of course, all of this assumes no additional cost for govt permits, licensing, bribing, etc in the US.  I imagine administrative costs like that are going to be a lot more than in countries like China.

The particular datacenter isn't the point here - just that this can be done, and done cheaper than AM has managed.

that MAY be true, but it doesn't look like it will happen in 2013.  

The reality is that to get something of this magnitude up and running is going to take months (at best).  We are going to see a lot of attempts fail to accomplish this goal, that's the nature of the beast.