Bitcoin Forum

With the ASICMiner blade auction edging into nose bleed territory for some of us, I started wondering what the most a reasonable person should pay per gigahash of mining power, ASIC or otherwise.

With GPUs and FPGAs, you can build a 2Gh/s machine for around $2000, or around 25 btc at 80$/btc.

Of course, you should really use the exchange rate that you believe will be the average conversion rate over the life if your miner, but that's way beyond the scope of this back of the envelope (actually, iPad at the gym) analysis.

Anyways, at 80$/btc, one should not pay more than 12.5 btc per gigahash of power.

And if you take into account the relatively easy 50% eBay rebate you can get for the gpus once you're done mingling with them, which you probably cannot get for an obsolete ASIC, perhaps the reasonable cap per gigahash of mining power is more like 6 or 7 btc per gigahash.

So, does that reasoning make sense?

As much as will return in 180 days or less.

You forgot the electricity, you need a room full of GPU rigs to match the performance of a single ASIC rig, and your circuit break won't stand it

That's what ended my mining expansion back in 2011. Circuit capacity reached (very very quickly too). Thinking back on it, that might have saved my life though, my room was at least 95F with windows open, doors open, and multiple room fans blowing, as well as cross-current windows open.

That said, there is no real way to price ASICs at the moment, super low supply, super high demand, super efficient product. I feel the pricing people are paying for ASICMiners is retarded, but if I had the BTC to splash, I might feel differently. There is a premium to be paid for an "off the shelf" product rather than a "will be made soon, we promise" product.

7BTC seems like a fair limit, but I wouldn't be surprised to see people surpass that.

if it doesn't ROI in less then 3 months, DO NOT WANT!

count your gigahash price from the above formula.

Bingo.  ONE Avalon is like having TWO-HUNDRED FORTY Radeon 5830s.

That would be about 24KW just for video cards (sans host computers).  Or $57/day in electricity at $0.10/KWH - more like $80/day with host PC requirements.

Honestly NO ONE should be buying GPUs right now.  Do a group buy of raw Avalon chips or something!

While that is certainly true, that only raises the cost per gigahash of gpus by around $1.80 per day at .15$/kWh (going by the power usage for a complete 3x 7970 pc).

When you consider a fair portion of the initial investment in gpus can be recouped in three or four months and reinvested into whatever asics are ready to ship at that time, at most likely much lower prices, I just don't think it's reasonable to pay exorbitant prices for current asics. I think this is true even if you have to pay server hosting charges for a big mining operation.

So you say first that the auction has a price thats too high but then say that at$80/btc one should not pay more than 12.5BTC per GH? I mean that would say that the auction could go till 125BTC per blade.

I wasn't directly commenting on the asicminer auction. I was wondering what the maximum anyone should pay is. I think it's hard to argue that the max is any higher than 12.5, but I certainly believe that it's lower.

Interestingly the current top bids are actually in the 6 or 7 btc per gigahash range that I think is the gpu equivalency point. At least they'll save themselves the trouble of converting to asics in four to six months.

Limit price BTC/GH in general is not connected with the exchange rate $/BTC, and is connected only with the difficulty of the network.

With the announcement of Burnin's Avalon Chip mining boards (https://bitcointalk.org/index.php?topic=179769.0), we have a better way of calculating the maximum reasonable price per gigahash.

10 Avalon Chips = 0.86 BTC * 1.1 = 0.946 BTC (per zefir's group buy (https://bitcointalk.org/index.php?topic=177827.0))
1 Mining Board Mounting = 80 EUR (~0.91 BTC right now)

So, for about 1.85 BTC, you can get 2.82 Gh/s, delivered in mid July, or 0.658 BTC per Gh/s.

Using organofconti's predictions for the earnings per gigahash from this thread (https://bitcointalk.org/index.php?topic=178051.0), we can calculate how much hashing between now and mid July is worth:

Cumulative earnings from now to August = 2.897 BTC
Cumulative earnings from July 19 to August = 0.120 BTC

Difference is what would be earned between now and the mid July delivery of Burnin's boards = 2.777 BTC.

So, that puts the reasonable cap per gigahash at 2.777 BTC + 0.658 BTC = 3.435 BTC, if you could start hashing right now. If it takes a week for delivery, deduct ~0.5 BTC.

Edit: Correct transposed Gh/s / BTC calc, and updated with current BTC/Euro price.

Is it set in stone that Burnin's boards will be delivered in mid July? You have to add "risk premium".

Fair enough. If you were not completely confident that Burnin will deliver in mid July, you could add a risk premium based on when you think he *will* ship by.

For every two weeks past mid July you think Burnin will ship, add 0.120 BTC (the earnings for one gigahash from mid July to August), which is reasonable if you think hash rate will level off in August. It's an overestimate if you think hash rate will continue growing at the current rate past July.

You'd have to believe he won't ship for over 7 months to be willing to pay 6 BTC per gigahash now.

Also the 80 eur is liability which is not sure right now. Also, Avalon chips eat more energy (although only slightly).

I mean, people who are bidding in AM auction surely have done their math. It's not like you find 5-6 thousand of dollars on street everyday.

The underlying tech is Avalon's proven chips. Burnin's project is the first one announcing something for sale, but there will certainly be others. I think it's fairly likely that at least one of them will succeed.

As far as the AM bidders, I'm sure they've done the math to ensure that they'll make back their investment eventually, but I think they put a more significant premium on a bird in the hand than I do.

No. That's 1.41 Gh/s per BTC or 0.709 BTC per Gh/s.

Oops. Thanks for the correction.

3.435 BTC as a reasonable cap per gigahash with the fixed math.