Topic: What's your magic number? (or, Should I keep mining?) (Read 8752 times)

They will have to start sharing them eventually. If they don't there will be no point in mining, and which means nobody will use their pool.

No, it won't, and you'd better stay away from business if you think so.

It will also create pressure for significantly higher fees. If people aren't willing to pay them a lot of miners will stop mining, and the network will get much less secure.
[/quote]

Right now something like 95% of miners mine on pools, and none of the pools payout from tx fees. Unless this changes, there is no pressure for fees to do anything.
I tend to agree, I don't really feel like there is much need for a magic number. Work out how much profit you are comfortable making and decide that way.

I'm personally considering dropping out already even though I'm way above/below a magic number, and I'm still making more than electricity costs, simply because electricity costs an arm and a leg here, and the hassle and noise and heat of running mining is approaching a level of not being worth it for me when half the money made pays for electricity, even though techincally I'm still getting a fair amount of "free money".

Not necessarily. Hardware is a one time cost. It may be more beneficial to run it even when the sale of coins doesn't pay for electricity and upkeep, just to keep the hardware generating. Obviously this is less attractive if you expect the value of bitcoins to fall instead of climb.

We've been over this many times. If you keep them it's speculation, and the potential gain should be attributed to this, not the mining. Also, if you are operating at a loss you should stop mining and buy coins instead.

It will also create pressure for significantly higher fees. If people aren't willing to pay them a lot of miners will stop mining, and the network will get much less secure.

Thanks, that's a good analysis. Except there's one other consideration. Your number is for instantaneous profitability. That may apply if you are selling all mined coins immediately and saving in other currency. However, if you expect in the longer run bitcoin may appreciate, then it can be worth it to accumulate as many as you can now in anticipation for that day, even if you have to pay by operating at an instantaneous loss in the mean time.

I think it will be an interesting dynamic at the point in time when the bitcoin block bounty halves to 25, and again later as it halves again and again towards zero. The market will have to absorb a lot of new bitcoins that are created in the meantime, and they can drop a lot. But at each of these discontinuities in block bounty, the supply/demand situation changes. A lot. When there are half as many per unit time being created, that will create pressure for appreciation.

The current profitability is around 2-3% / month (compared to hardware costs). I don't see why would anyone start mining at this moment. Am I wrong maybe?

Ok so I set up this site with a simple dashboard based off this magic number equation
http://bitcoindashboards.com
  Please check for any errors.  I will keep building on it.  The next step will be to factor in some what ifs for $ projections, ROI %, fee incorporation etc. etc. Any suggestions welcome.  Also I will pass along 10% of any donations to the original poster of this equation, Ian Maxwell .

I find it interesting just to look at the numbers. I don't even mine as it's totally unprofitable for me.

I am not sure why people are overgeeking this into obscure numbers.

1. Take your 7 day average of BTC generation. While difficulty remains stable, this is your current income estimate in BTC. As it adjusts, you get a new figure that should ideally remain vaguely proportional.

2. Figure out the Watt difference of your system between generating and not. This is either idle vs load if using the PC anyway, or off vs full GPU load while running if dedicated.

3. Calculate your extra / total kwh per day. Multiply by xx cent per kwh. Have a daily cost.

4. You can now calculate: At xx BTC per day income and xx cent or USD cost per day, I need the exchange rate to remain above xx USD to break even or make a profit.

5. Forget magic numbers.

I went a step further and did a profitability chart. Profitability = Magic number/cutoff
Ian, I used you magic number of 9700 shares/USD as an example for this chart.



If I'm interpreting it right this number shows the ratio of money you are generating by mining vs the money spent on electricity. So a number of 10 means you are making 10 times as much in bitcoins by mining as you are spending in running costs. If the number goes below 1 then mining is unprofitable. There is no consideration for initial hardware costs or depreciation or anything like that.

Inspired by this thread and bitcoinBull's thread http://forum.bitcoin.org/index.php?topic=7427.msg109079#msg109079 I created a graph of the cutoff number over time.



It uses the daily average price from MtGox and the difficulty level on that day.

Great, thanks guys!! I get it now.  With the given numbers, is there then a way to calculate what you are generating btc/hr from your magic number?

1E6 / 2³² = 0.000233

It is a calculation given a difficulty of 1. It is an approximation, because in practice you might get more or less.

If you have a six sided die, the odds of getting three is 1/6. But in practice, you might only have to roll the die 4 times to get a 3 instead of 6. Or maybe you'll need to roll it 13 times. But if you do try this billions of times, the number of threes as a percentage of the total number of rolls will approach 1/6.

Given a difficulty of 1, 233 is the approximate number of valid hashes you would generate with 1 Thash.

At least, that's how I understand it =)

Thanks for the replies.  What I'm asking is where did the 233 in the first post come from?  How do we know it's a constant?  Is it a calculation related to the difficulty?  If the difficulty goes up, does the shares per Thash go down?
   

frozen, what you wrote seems pretty much correct. It's worth pointing out that the 8.5 seconds is an average, and there's a significant chance that once in a while it will take a minute or longer (or a second or less). In fact, if you're mining 24/7, that probably will happen once in a while.

From the first post: "shares is the (constant) rate of "share" generation per hash, which is about 0.000233 shares/Mhash or 233 shares/Thash"

For every Thash, approximately 233 "shares" are generated. A "share" is a hash that would validate if difficulty were set to 1.

Suppose that you have a machine generating 500Mhash per second. Multiply by 60 seconds per minute = 30000Mhash per minute. Multiplying again by 60 gives 1800000Mhash per hour, or 1.8Thash per hour. Approximately 1.8*233 (419) shares should have been generated during that hour, 7 shares per minute. It will take approximately 8.5 seconds to generate a share. Double this to 1000Mhash, and it will take about 4 seconds to generate a share.

If anyone finds a mistake please let me know.

Oops, I messed up the comparison. It should be

Where did the shares =233Thash/hr come from?  I can't find where that is calculated.

Bitter Tea: Thanks for writing the script, but something seems backwards, when I put 15mhash it says profitable, when I put 1500Mhash its says not profitable.

Yeah, some resources are definitely shared.
For my machine it looks much more extreme:

Just CPU: 9.5 Mh/s
Just GPU: 85.5 Mh/s

Both: 68Mh/s(GPU)+9Mh/s(CPU)=77Mh/s - so it seems their competition for resources takes up some time. I got the highest M using GPU only.

Very useful post for those intending to get into mining and/or purchasing more hardware. As I'm using some existing equipment a while ago I calculated the extra power use as about $1 a day and just keep my eye on if it's profitible from time to time based on my BTC generated per day, not as though in my case having it running a week when it would have been cheaper to purchase is a particular financial disaster.

Besides it's coming into winter in Australia and I'm having a go at growing herbs inside, so if I stop mining it'll probably get too cold and kill my Thai Basil .

Don't forget that this is assuming that you sell your BTC at the exact rate that is used to calculate (BTC/USD), if you stop once your "Magic Number" falls below, you'll stop earning BTC. Essentially betting that BTC rates will not go up.

If enough people does this and pulls out, others that are still mining will get more for their shares since less share are distributed during the solving time.

Also, you're ignoring the fact that (if) you invested capital into buying hardware and have not compensated, those are just sitting around. AND if you're just waiting for the rates to go up, you won't have BTC to sell (assuming you sold them when you've got them) when they do and have to start mining for them.

But of course, during this 'downtime' it's obviously better to use your 'cost' to buy bitcoins instead that'd bring a different dynamic as people who are still mining would earn more BTC while people who are buying bitcoins are bringing the rates up.

The important question is then how long does it take for the rates to go up high enough for people to start selling/mining again, if it takes long enough, people who are mining might make more than those who are buying as the prices go higher. If this time period is too short then question is did you buy enough to cover what you've missed out.

Then again, this is assuming if the economy get's back up. if it doesn't, then those who invested into buying more than their 'cost' would lose more, while those mining will have BTC in their wallets and maybe those will worth something in the future after the collapse.

Just my 2 cents worth, I don't really know which is better or the real dynamic. I'm new and I just think that calling this a 'magic number' is over-reaching and it actually doesn't reflect enough of what goes on. It just shows that right at this point you're probably paying more for electricity than what you're mining. Not necessarily a decisive factor to stop operation altogether.

Just out of curiosity I did the calculations for my Netbook (HP Mini 311c) which has an Atom N270 and Nvidia ION chipset (Geforce 9400M - hell yeah, netbook with Dx10 and Cuda  ) while they have very low hash rates I was curious to see if the silly low power usage would counter that. The hash rate testing was performed in Ubuntu 11.04, using the SSE CPU Miner and Phoenix GPU miner.

I pay about £0.10 per kWh for power.

CPU:

Hashes at 1.52Mhash/s
Draws 5W at full load

GPU:

Hashes at 1.92Mhash/s
Draws 6W at full load

However when I tested this I noticed that they cannot do these hash rates while the other is loaded as they must share some of the same components in the ION chipset, the combined hash rate was only 2.52Mhash/s.

Once again discounting all extras such as the screen, HDD, speakers, etc, etc we get the following:

Just CPU: 2549.95
Just GPU: 2684.16
Combined: 1921.61

So in this case, surprisingly in theory it is technically profitable right now to using my netbook; however once the power used by everything else is factored in (29W maximum with screen on full brightness at 100% load) with the obscene low hash rates it is completely unfeasible.

that sounds about right, mining on any CPU is pretty much worthless, they are very inefficient.  mining on my 5GHz 2500K yields 8 Mhash/s where on my 6950 is 360 Mhash/s, ATI video cards are where its at, nVidia cards are about 5-10x slower due to their GPU architecture being more like a CPU than ATIs.  You can read about it on the bitcoin FAQ.  It doesn't sound like it would be worth it for you on either the CPU or GPU.

Based on the information here I can make a few of these estimates for your system under ideal circumstances.

Using the highly optimized SSE2 miner yours Q6600 should be able to do 11Mhash/s using all 4 cores for 105W, now if you only want to use two cores then we can roughly half all of those numbers for the total calculation (very roughly)

So thats 5.5Mhash/s @ 52.5W.

Next we add your graphics card in, which is sadly only capable of about 3.37 Mhash/sec at a power usage of 50W.

So the following magic number apply for you (assuming no other power use, which of is not true):

Just CPU = 585.83
Just GPU = 376.90
Both CPU & GPU = 518.43

Now as you can see I'm afraid even under ideal circumstances, not counting the cost of power for your other components/cooling/peripherals you are well below the break-even point of around 700 shares/USD right now, with your GPU being far more inefficient than your CPU.

I'm new to this (just saw the write-up on lifehacker), and curious if this would be profitable. I did a test run on my q6600 using only two cores and got 2269 khash/s. Now at .14 kWh and a 450W ps I got .008391698 as my magic number. Now assuming my math was right (.000233*2.269)/(.45*.14) I should just forget it.

My questions are:
1. Did I get the math right?
2. Would using a mining program vs bitcoin make a difference?
3. Would using a mining program on my 9400GT card make it worthwhile?

Thanks for the help.

If mining for a pool that has a fee one should use an adjustment for the reward: For example if the fee is 2% (like Slush's one) the reward should be 49 not 50.

Values for me:

M=317
Threshold=275

So I should mine even if my comp has nothing else to do.  

Careful with the units.

(1) 1.44 Gh/s is 0.00144 Th/s, not 1440. So divide by 1,000,000.

(2) Your hashrate is a rate per second, but your power usage is a rate per hour. So multiply by 3600.

(3) I'm assuming by "KW/hr" you mean "dollars/kWh" (or whatever currency you prefer).

This gives you an actual magic number of 16,043.4286 shares/dollar.


You're good until the difficulty divided by (50x the exchange rate) reaches 16,043.4286. But this number is falsely precise and you'd be better off looking out for the number 16,000.

1.44 Gh/s 820Watts .092 KW/hr

(1 440 * 233) / (.820 * .092) = 4 447 507.95

So I'm good until the difficulty reaches 4,447,507.95 ?

Subsidy is another name for the block reward. I prefer the term subsidy because it implies that it is a temporary measure while the network is smaller and transaction fees are not a sufficient enticement for mining.

few months/weeks later mining can't be profitable even with GPU.
and after hash change[SHA256 isn't bulletproof enough, IMO], even CPU's mining goes in history too.

I just noticed something: If you're paying $0.15/kWh or less for electricity, mining with a very high-end CPU, such as a Phenom 2 X6, is actually marginally profitable at the moment. This even with the recent drop in value and fast rise in difficulty, which implies that it was more-than-marginally profitable a week ago. (This only applies if you already own the CPU, of course.)

"42" (c) ?

What is the subsidy variable you added intended to be?

as long as complexity rise, only guys with cheap electricity stay in business.
half-year ? maybe faster.
ie, guys with access to industrial area, living in Russia or under Nuke power station

Well, it's not exactly a calculator, but maybe this is useful. All the numbers scale nicely: that is, if you want to look at the 1M-10M difficulty range just multiply all the magic numbers by 10, or if you want to look at exchange rates in the 10-100 range just divide them all by 10.

currently the exchange rate is tanking and the difficulty is on track to go up quite a bit.

For those who have difficulty converting units...

MHash/sec * 60 seconds/minute * 60 minutes/hour * 1 THash/1000000 Mhash = THash/hour

All the numerator units except THash cancel, and all the denominator units except hour cancel.

Btw, someone should throw this up on a website to calculate all this stuff. Exchange rate and difficulty could probably be filled in automatically with a couple of calls to BlockExplorer and MtGox.

No that's right (Well your Thash/hour is a little out but its close), your just paying twice the amount per kWh in power than most of the other results in this thread which is forcing your Magic number to be about half the size (give or take).

A theoretical 2x6870 rig @ 600MHash/s combined.

(233 * 2.16) / (0.450 * 0.21) = 5325.71...
244139.48158254 / (6.44 * 50) = 758.197...

Something doesn't seem right (It is 4am, though... nothing seems right when it's 4am)

Mine is 16,776

Absolutely right! A share is just a hash that would be valid at difficulty 1. I've edited the post to use this terminology.

My magic number is around 16,731 using a single 5870 with a hash rate of around 375Mhash/s (1.35 Thash/hour) - I have discounted the power requirements for the other components since the PC would be on pretty much 24/7 anyway (its not a dedicated miner) and as such the mining is only costing me the extra GPU power (0.188 kW) draw at a rate of £0.10 kWh. (Infact it is probably reducing my heating costs  )

The current difficulty value for GBP (based off Britcoin ask rates) 244139 / (£3.50 * 50 BTC) = 1395.08 (shouldn't this be Shares/GBP?).

Last Update: 18:40 BST 21^st May

Edit: I've decided to calculate my magic number including all other components and internals, cooling etc, which all in all uses around 350W (Core 2 Duo E5200 @ 2.75, 4x1Gb DDR2, Assorted drives).

( 1.35 Thash/hour * 233 shares/Thash ) / ( 0.35 W * £0.10 ) = 8987.14

A much lower number once I included all the internals and USB stuff also drawing power; however I would probably keep mining up until the 16,000 mark simply because the other components would be turned on regardless  since this is a desktop machine and I don't count their cost off my BTC profits.

When calculating the electricity cost, don't forget: everything that uses power, that wouldn't use power otherwise, is a mining cost. If my machine weren't mining, it would be sleeping and wouldn't have a window fan pointed at it.

Excellent post!

It appears my magic number is ~15,000. I'm mining with a single 5850, see if my numbers check out.

0.972 Thash/hr, 0.151 kW (based on the mining hardware comparison), $0.10/kWh.

Watching the difficulty go up, I realized I needed a way to quickly recognize when the day comes that mining is no longer profitable. So I worked out the following. Maybe someone else will find it useful.

If you are a miner, your your "magic number" is M = (shares * hashes) / (power * cost), where

• shares is the (constant) rate of "share" generation per hash, which is about 0.000233 shares/Mhash or 233 shares/Thash;
• hashes is your hashing rate versus time (say 780 Mhash/sec or 2.81 Thash/hr);
• power is your rate of energy consumption for mining (say 450W or 0.450kW); and
• cost is your cost of electricity (say 0.150 USD/kWh).

All of these numbers are relatively fixed, so go ahead and calculate your M now. Be sure you have the units straight---this "number" is actually denominated in shares per . In fact, it has a specific meaning: it's the number of difficulty 1 shares you are generating per spent on electricity. Using the data above, for example, my magic number is about 9700 shares/USD.

The point of this exercise is that your mining is profitable on the margin if and only if M > diff / (xcg * reward), where diff and xcg are the current difficulty and exchange rate and reward is the block reward (50 BTC right now). So when difficulty increases and you're wondering whether to keep mining, or the exchange rate soars and you're wondering whether it's time to fire up the ol' CPU again, you now have an easy way of deciding. Just divide difficulty by exchange rate and reward, and compare it to your magic number.

For example, right now we have a cutoff of (244139 shares/block) / (6.63 USD/BTC * 50 BTC/block) ≈ 735 shares/USD. Anyone whose magic number is lower than this shouldn't be mining. Fortunately, mine is still well above it! Unless the exchange rate tanks or the difficulty soars, I should still be in the game for at least a couple of months.