Topic: ASICMINER: Entering the Future of ASIC Mining by Inventing It - page 682. (Read 3917468 times)

He says an order is two weeks late. Why? Do they know if it will show up? I don't see how you can say , oh the order is 2 weeks late, but in OCT it will add 500 ths. If it's already late how can you know when it will arrive?

Great news

I'm sorry, but it's very difficult to answer that question.  The industrial prices are usually lower than home prices, but they depend on a lot of things.
1) There are many different companies in different areas, but all of these companies are de-facto national, not private.  These companies are very complicated, and may have very different prices.
2) The companies usually don't post the prices on the internet, suggesting you call them and ask.  I've just tried to find some region-by-region comparison rating or something like that, but there's no such info.
3) The prices are calculated from many parameters: maximum power, average power, consumed energy, power usage ratio (avg. power / by max. power), supplied voltage, location, etc.  The prices are calculated from company's transit price (set by local officials) and federal prices (set by laws), both of them change often, at least every year.
4) The prices depend on who you are: what kind of company (there are several categories), and it may become lower if you are "well connected" to some officials in the energy business.
It's all VERY complex.

As for hooking up, it may also be different: for example they can provide a cable, or you may have to lay the cable yourself.  The closest energy hub may be full so you may need to lay a cable to some more remote hub.  If you're in the city, it will become bureaucratic hell to lay the cable through the streets.  And of course it also depends on corruption, whether you are "well connected" or not.

I've managed to find some numbers at around $300/kW for hook-up, but it's for 2010 and for Moscow (the most expensive region).

Interesting. Haven't had a chance to speak to a local. I assume cooling expenses are lower too in Russia  

How's rent on industrial space? What would it cost for, say, a 1MW hookup (and associated costs per watt at that consumption size)?

I think there is serious first-mover potential for low-margin mining in Russia.

We pay here about $0.01/kWh, but that's for private homes.  Companies pay less, about $0.005/kWh, depending on their volume, voltage and location.  And the big loyal-to-government companies get it for $0.001/kWh in some cases, because government kinda "supports" them, LOL.  Stupid socialists.

It's not Siberia though, only 500 km to the east from Moscow.
But it also varies from region to region, I just happen to live 30 km from a hydroelectric plant.  I heard in other regions prices are significantly different.  For example, some southern regions (Dagestan, Chechnya) get some "special" government support, so people there get $0.003-0.005/kWh for private homes and even less for companies.

Hope it's useful.

The future of asic mining is selling hardware, including cooling system, backplane, maybe a diesel generator, solar cell, phone miner, anything relating to mining. Ltc / xmp asic ,may come in a few years, we might even see a personal nuclear powerplant selling in btcguild. So the income is not limited to  btc left to be mined.

Vycid has it correct. This is an interesting link for a rough comparison -> http://en.wikipedia.org/wiki/Electricity_pricing. In the US, I believe the average is around $0.09-0.11/kWhr with some exceptions(California, extremely high, Pacific Northwest, extremely low).

I'm on an industrial plan, if I can franchise I need to talk with Excel about the ability to drop my kWhr rate even further.

Wow -- I thought we had it cheap in Oklahoma at $0.045/kWhr during non-summer months.

Thanks for the info.

I checked europe when i still thought i would get a reasonable amount of profitable avalon asic chips and found that in europe bulgaria has the best rates with 9Euro Cents.

~ $0.07/kWh in Guangdong (Shenzen - AM is here).

~ $0.01/kWh in parts of Central Washington (state) and Siberia.

~ $0.15/kWh in California (for businesses).

The electricity cost is going to be much a factor in the long haul as the chips are. We will definitely see a migration of old hardware to super-cheap electricity regions. If I was starting a company, it'd be to do just that - buy obsoleted hardware and move it to super-cheap locations.

Any idea what rates are around the world?

I pay about 0.045/kWhr.

Interesting regarding the franchising. I hope more information comes out as I have ~60 kW of electrical capacity sitting idle due to increasing difficulty making any ASIC purchases a risky gamble. Winter is coming and while $0.076/kWhr might not be the most competitive, I'm sure it'll be competitive for a few months.

mean while, cointerra have just got FPGA working, is it good or bad news

This is typically called the "minimum feature size".  It has a bunch of other names too, like "critical dimension", etc, but I feel that "feature size" gives the best immediate understanding.

First - the process size (130nm, 55/45nm, 28nm, etc) has nothing to do with the chip thickness.  That is determined by the thickness of the wafer, which is 400-800 microns thick (it's actually usually more for smaller process nodes, since 300mm wafers are used and those have to be thicker to keep the wafer from flexing/shattering).

The process size instead refers to the standard half-pitch of a memory cell. That doesn't really apply to a SHA256 ASIC, but it is a measure of how small the process can reliably make things. A die shrink (logic's 32nm -> 22nm for example) typically doubles the amount of standard cells that can fit on a die of fixed size. For SHA256 applications, this means precisely double the speed (and probably less power use per GH). Since the assembly is a major part of the costs,  and you can get twice as much out of one chip, you've basically halved your costs per GH, AND made it easier to deploy a large amount of hashpower.

So that's 32->22, one 2x shrink. For 130-> 28 we expect approximately 21.5x performance (feel free to check my math). Probably more, since professionals will be doing the 28 designs and FC's 130 design was basically a hobby project.

The process node matters. Intel's whole business model has relied for decades on staying a process node ahead of their competitors. Seems to be working.

“Risk comes from not knowing what you're doing”
― Warren Buffett

I hope that friedcat will do and sell this extremly overpriced Novec 7000 cheap
https://bitcointalk.org/index.php?topic=255613.0;all

new players (bitfury) are already hashing for months with their gear, they are deploying 8 x ASICminer hashrate, and are selling chips for premium prices.

I would say its quite the opposite: ASICminer is facing higher pressure now then 6 months ago.

  

I have a hunch on what system they are using but it is interesting news

I totally share Strange Vlad's feeling:
Many of you seem to believe that thiner chips instantly make the previous ones obsolete. But from a miner point of view, the equation is simple: see what you can get for which price, see how much it will consume of electricity, and see how much it will earn you in bitcoins. Looking at products with this in mind, 130nm will remain a relevant choice for many more months. I am sure AM pricing policy will constantly remind that to everybody.
When FC says "gen2 will come in time", you are wrong to think it is a way to admit they are late. It simply means "at the most appropriate time". You should really wonder why AM does not agree with you as to what is the most appropriate time...

I also share Franktank's positive view of two aspects:
1) The cooling system. Nobody saw that coming! It may sound a bit of a detail, but for me it's not: if you manage a farm of 100TH/s, that can make a huge difference in terms of electricity/maintenance. Plus, this is going slightly further bitcoin mining alone. Isn't that what many had advise AM to do?
2) The HK data center. AM is moving toward more decentralization (for the farm it manages itself). They spoke about moving partially to HK, now it becomes reality and this is very good news also for resilience to legislation matters.

It sounds to me, like they haven't even pushed the gen2 design to fab yet. Sounds like AM is being stomped by bitfury, knc, cointerra, who else? They probably were about to push out a new design around what, 45-65 dies and then realized their head is about to be crunched by 22 dies and had to make a change.

AM defined BTC ASIC's, however, unless their power is very cheap and they get to the market quickly, the are about to become meaningless.

At least IMO anyway