MINING IS PROFITABLE

[casascius]

Here is my mining spreadsheet.

Assumptions: (PLEASE FEEL FREE TO CHALLENGE):

• Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity
• I am assuming today's BTC's are worth 57 cents, despite 0.85 at MtGox
• I am assuming difficulty will rise 30% every ten days
• I am assuming my two 5970's will always give me 1.2 GHash/sec
• I am assuming my mining rig cost me $2000, and my power is $0.095/kWh

If all these assumptions are even close, then by my calculation, mining should still be super profitable, even after paying for the equipment.

https://spreadsheets.google.com/pub?hl=en&hl=en&key=0AlC40w3H3dxKdFpJQWc3VlhyWUIzN1Y1UnpEaVlyOEE&output=html

Please feel free to suggest which assumptions I should adjust, if any.

EDIT: While these estimates show that one could buy 1227 BTC today and be further ahead, that depends on the difficulty rising by 30% every 10 days.  The point I am making is that mining should be reliably profitable, more so than buying BTC and stashing it away.  If it does not rise that fast (which is very possible), mining will yield more BTC that are worth less, but should still be just as profitable as I say it is.

EDIT 2: added a Sheet2 that assumes difficulty and value only rising 15% every 10 days, which is more likely as a long term average.

[1714933892]

1714933892

[1714933892]

1714933892

[1714933892]

1714933892

[Quip]

Yes, yes it is.

[slush]

mining should still be super profitable

psssst!!! quietly! :-)

(You rised next difficulty to 50000 right now.)

[jakakxl]

># Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity

that is a very wrong assumption; just take a look at your numbers

[riX]

The value will always be what the market decides. If everyone trading on the market was making mathematically optimal decisions, your first assumption would be valid.
This is the same reason as why salary is not equal to the amount of money needed to buy the exact amount of what you produced by working.

[SmokeTooMuch]

• Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity wrong

• I am assuming today's BTC's are worth 57 cents, despite 0.85 at MtGox wrong

• I am assuming difficulty will rise 30% every ten days dunno

• I am assuming my two 5970's will always give me 1.2 GHash/sec could vary through driver updates

[casascius]

• Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity wrong why?

• I am assuming today's BTC's are worth 57 cents, despite 0.85 at MtGox wrongThen mining is more profitable than I say it is

• I am assuming difficulty will rise 30% every ten days dunnoRise less? make more BTC.  Rise more? BTC worth more.  Which?

• I am assuming my two 5970's will always give me 1.2 GHash/sec could vary through driver updatesbut of course, you won't install a driver that reduces output, will you?

[SmokeTooMuch]

• Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity wrong why?

• I am assuming today's BTC's are worth 57 cents, despite 0.85 at MtGox wrongThen mining is more profitable than I say it is

• I am assuming difficulty will rise 30% every ten days dunnoRise less? make more BTC.  Rise more? BTC worth more.  Which?

• I am assuming my two 5970's will always give me 1.2 GHash/sec could vary through driver updatesbut of course, you won't install a driver that reduces output, will you?

1. Bitcoin can fail anytime (see the thread about a google competitor) and electricity prices aren't the same worldwide.
2. I don't see how they can be worth 0.57USD when the market price is way above. Maybe I misunderstood something ?
3. I just can't estimate the growth in difficutly, there could be a jump-in or a jump-out of huge computation power anytime.
4. If you need the cards for other stuff than mining, you maybe do.

[M4v3R]

I don't get it where you get BTC value from. It's 0.57 cent in the beginning, but it get's unusually high over the time, even if your BTC count is not that high (~1200 BTC)

[casascius]

1. Bitcoin can fail anytime (see the thread about a google competitor) and electricity prices aren't the same worldwide.

Very true.  But if someone steals Bitcoin and starts a new block chain, then I'm already equipped to take my mining there too.  And I will be there on the ground floor, instead of two years into it.

2. I don't see hwo they can be worth 0.57USD when the market price is way above.

That's me being conservative.  I think BTC are overvalued in the short term.  0.57 comes from ignoring the huge recent hump in the overall growth chart.  If I am wrong, and BTC are really worth more, then mining is even MORE profitable.

I believe if anyone has a heap of $$$ to bring to the table, and they want a more reliable return, they will be reading this board first, and they'd be nuts not to throw it at MINING right now.  The ONLY thing that will make the value at MtGox rise, is someone bringing moneybags to MtGox, and nothing else.  By making posts and presenting evidence like this, I am very arguably preventing that from happening, singlehandedly.

3. I just can't estimate the growth in difficutly, there could be a jump-in or a jump-out of huge computation power anytime.

You are absolutely right.  Mining controls for that though.  If there's a huge jump-out, mining will produce more BTC per miner.  Isn't that good?  That would make mining either the same profitability, or MORE profitable.

4. If you need the cards for other stuff than mining, you maybe do.

Maybe. but I have made the conservative assumption that I don't.  If I do, I'm profitable faster, since instead of a $2000 outlay in equipment, I am only paying its depreciation, which is MUCH less.  So, in other words, if I'm wrong, mining is even MORE profitable.

[Raulo]

[Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity

Wrong because it is other way round. Difficulty is related to profitability of mining but BTC price is supply and demand on the market. BTC can rise or can drop and difficulty can adjust accordingly.

Nobody knows what the price of BTC will be. Anybody who claims otherwise is either a fool, tries to pump the value of his holdings, or both.

[casascius]

[Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity

Wrong because it is other way round. Difficulty is related to profitability of mining but BTC price is supply and demand on the market. BTC can rise or can drop and difficulty can adjust accordingly.

Nobody knows what the price of BTC will be. Anybody who claims otherwise is either a fool, tries to pump the value of his holdings, or both.

the reason why I believe they are related, is because demand is directly related to the number of people who know about, and care, about Bitcoin.

And so is the number of miners.

If Bitcoin hits the 10 o'clock news, both will rise in step.

Actually, video cards for mining are becoming increasingly rare on the market, most places are sold out of the 5970, or want much more for it when it used to be $500 new.  You can hardly even buy it used on eBay for $500 anymore.  Lack of supply of GPU's will introduce an exception to the growth-in-lockstep assumption... one which would make mining MORE PROFITABLE

[Cryptoman]

Like Raulo, I question your first assumption:

Value of a BTC will always be related to the cost of power to produce it, since that's directly related to difficulty, which in turn is related to Bitcoin's popularity

If someone decides to mine because it is profitable, then at some point they will want to collect their profit by closing their position.  Therefore, if they invested in mining equipment using USD or other traditional currency, then they will want to exchange BTC for USD or EUR or whatever once they generate.  This will create selling pressure and drive the value of bitcoin down.  Difficulty is closely related to mining popularity, but value is more closely related to investment and trading popularity.

[casascius]

Like Raulo, I question your first assumption:

If someone decides to mine because it is profitable, then at some point they will want to collect their profit by closing their position.  Therefore, if they invested in mining equipment using USD or other traditional currency, then they will want to exchange BTC for USD or EUR or whatever once they generate.  This will create selling pressure and drive the value of bitcoin down.  Difficulty is closely related to mining popularity, but value is more closely related to investment and trading popularity.

You are right.  It will drive the value of Bitcoin down to what I think is its real value.  I don't think that the stable true value of Bitcoin can get much higher than the cost of producing new ones.  Why would anyone want to pay $10 for a BTC when it can be currently mined for 5 cents?

When Bitcoin is priced as a function of the goods and services and the productive capacity of the community surrounding it, as well as the cost to mine it, the value is more stable and real.

When Bitcoin's price shoots into the stratosphere after a $10k buy and plummets to the ground after a $10k sell - a consequence of a very few people hoarding it and making it artifically scarce because they expect it to be worth millions down the road - Bitcoin becomes highly volatile and makes it unattractive as a means of trade.  It makes it more likely that someone who wants to give Bitcoin any serious merit will look to dump the block chain (and all the "hoarders" with it), and start anew.

I would rather see 25 cent Bitcoins that stay that way, rather than 1 dollar bitcoins that pop when someone sells a few.  I feel that if I can encourage more people to mine, the day when 1 dollar bitcoins are real and don't pop at the slightest pressure will come sooner.

If Satoshi thought leaving the mining to a privileged few was a good thing, he would not have built the "Generate Coins" option into the client.

[marcus_of_augustus]

I like the way casascius is thinking.

Would it be correct to paraphrase: Mining difficulty is underlying drive for increase in value of bitcoins, not the other way around.
(although there is a certain amount of feedback)

Would be interesting to see a plot correlating network difficulty and bitcoin value (as based on a basket of commodities and currencies).

Only when we get some big cabinets full of racks will the value really head up and stay up. The mining hardware is the concrete under the floor of the BTC.

On that note, has anybody had a go at porting a miner code onto the Tilera chipset?
http://www.tilera.com/products/processors

Linux SMP and other lower-level data-planes are already ported and apparently, they have very good integer performance   , i.e., for network security, QoS and DDoS, digital video.
I'm trying to get hold of an evaluation board but it seems they are held in tight supply unless you want to buy lots of 10,000  

[casascius]

Only when we get some big cabinets full of racks will the value really head up and stay up. The mining hardware is the concrete under the floor of the BTC.

On that note, has anybody had a go at porting a miner code onto the Tilera chipset?
http://www.tilera.com/products/processors

Linux SMP and other lower-level data-planes are already ported and apparently, they have very good integer performance  , i.e., for network security, QoS and DDoS, digital video.
I'm trying to get hold of an evaluation board but it seems they are held in tight supply unless you want to buy lots of 10,000 

That 5970 is going to be pretty hard to beat.

I looked at the hashing code for the 5970 at the assembly language level.  It does a pretty dang good job of parallelizing the workload.

The 5970 has 3,200 "stream processors" which each have 5 distinct ALU's, so that's a total of 16,000 ALU's.  In its most efficient configuration, each set of 5 ALU's is working on 2 hash attempts at a time.  The ALU's are what do the dirty work of hashing, and the OpenCL compiler does a pretty good job of parallelizing the workload to keep all the ALU's busy.

The gross number of ALU's in the 5970 come at the expense of them all being relatively "dumb".  General purpose computing (specifically, if-then-else branching) is very expensive for them.  But bitcoin mining needs negligible branching, so the 5970 favors it.

Without a similar number of available ALU's, I'm not sure what platform will really outshine the 5970.

If they are nothing more than 64 normal cores on steroids, and it can run Linux, then running a CPU miner should already work.

[TheHoboHarvester]

$75 million USD by next year? With 2 5970's?
You're assuming that people are going to pump billions of dollars into btc.

http://ragecomic.appspot.com/packs/laugh/images/Laughing.png

[dust]

$75 million USD by next year? With 2 5970's?
You're assuming that people are going to pump billions of dollars into btc.

This was exactly my face when I read the spreadsheet.  The difficulty rises when the btc price makes it profitable to mine.  The btc/usd exchange rate does not necessarily rise with increased difficulty.

[casascius]

$75 million USD by next year? With 2 5970's?
You're assuming that people are going to pump billions of dollars into btc.

Actually, I'm smarter than that.  The only point I'm trying to make is that you'll quickly do better than breaking even by mining, not that you'll become a millionaire.  Construed literally, my spreadsheet could be read to say that if you buy 1227 BTC today, you'll become a millionaire.  Obviously not very likely.

The bazillions only appear as a result of how many rows I hit paste on.  Obviously BTC can't continue to grow at 15% or 30% every 2 weeks forever, with world population (and the world's budget for GPU's) as an obvious limiting factor.

I deleted the rows dated far into the future to settle that.

[marcus_of_augustus]

That 5970 is going to be pretty hard to beat.

I looked at the hashing code for the 5970 at the assembly language level.  It does a pretty dang good job of parallelizing the workload.

The 5970 has 3,200 "stream processors" which each have 5 distinct ALU's, so that's a total of 16,000 ALU's.  In its most efficient configuration, each set of 5 ALU's is working on 2 hash attempts at a time.  The ALU's are what do the dirty work of hashing, and the OpenCL compiler does a pretty good job of parallelizing the workload to keep all the ALU's busy.

The gross number of ALU's in the 5970 come at the expense of them all being relatively "dumb".  General purpose computing (specifically, if-then-else branching) is very expensive for them.  But bitcoin mining needs negligible branching, so the 5970 favors it.

Without a similar number of available ALU's, I'm not sure what platform will really outshine the 5970.

If they are nothing more than 64 normal cores on steroids, and it can run Linux, then running a CPU miner should already work.

Yeah, I agree. Looks like the Tilera has only 1 ALU per processor, so 64 total or up to 100 on new chip ... not in the same league as the 5970.