Bitcoin Forum

Difficulty History
Q: Is there a replacement for the dead link . "http://nullvoid.org/bitcoin/difficultiez.php" ?

I bet it will drop once summer kicks in,  :)

Here : http://blockexplorer.com/q/nethash

Just a bit of information I came up with while talking with a friend of mine today. Based on the last blocks found by slush's pool that fit on my screen at one time (I think it was 15), the value of a share was .0006 BTC. (Since the score based model evens out over time to match the share based, I think it's fine to talk about the value of a share.) That means that the new value of a share will be about .000378 BTC.

My miner does 300-310Mhash/s, and found 469 shares in 2 hours, which is average for me. This is 3-4 shares per minute. At 100 Mhash/s, you should expect 1 share/min. You can use this as a way to see how many BTC you can expect to mine. Here's the table I made to share with my friend, hope I did the math right, and hopefully somebody can use this to see if it's worth it to them to mine.

Mhash/s      Estimated Hourly Payout      Estimated Average Daily Payout
10               0.002268                          0.054432
20               0.004536                          0.108864
100             0.02268                            0.54432
300             0.06804                            1.63296
500             0.1134                              2.7216

I think something is wrong here. I calculated and tested on the course of a week; my average was 0.2 BTC per hour, with the new difficulty it's 0.145 per hour. Your averages are too low for some reason.

I was thinking something was off, but the numbers bore out. Maybe I was just at an unlucky patch when I took my readings. My average was .2/hour before the difficulty change.

So, I'm going to do the math out here, and maybe we can see where I went wrong.

Statistics taken from mining.bitcoin.cz.

So far it seems that the value is holding the same, and there isn't enough blocks with the new difficulty to throw the averages off yet. Or maybe the average will always be the same, but the rate that shares will be found by the same miner will be slower.


This is where my problem was - I used the rate of share discovery from a block with the new difficulty to figure out averages in the old difficulty. Since the share rate is slower with the new difficulty, the payout looks lower.

So, my payout previously was about .2/hour. With a 37% difficulty increase I expect it to be .126/hour. With that (rounded to 4 decimals)...

10Mhash/s - .0042/hour - .1008/day
20Mhash/s -  .0084/hour - .2016/day
100Mhash/s - .042/hour - 1.008/day
300Mhahs/s - .126/hour - 3.024/day
500Mhahs/s - .21/hour - 5.041/day

That looks more accurate to me. Same with everybody else?

why so complicated?

i would say, it isn't a good idea anyway to base your average expectations on the number of blocks a given pool found a few days/weeks back, but rather on your hashrate, the pools comissions and the current difficulty.

at 10Mhash/s you'll need an average of 9090.5hours to create a block,
let's just say the pool takes 3% and due to network-overhead/-connectivity and invalid or stale hashes you lose another 2%,
then you'll need 9090.5hours to create 47.5BTC on the pool, that's what you can and should expect on average.

10M: 47.5/9090.5 = 0.00522BTC/hour or 0.125BTC/day

I simply calculate it by multiplying the share price, by the number of shares per hour i submit on average (doesn't change much).

0.00058 * 250 = 0.148 BTC/hour

The previous number was very accurate for me using that method, and this number is accurate so far, also.

I just remember that everything goes in a cycle, and whatever spikes will drop. The difficulty spiked, so it will drop eventually. And I'm not mining for the money, it's just a nice benefit.

More (probable) math problems a friend and I are trying to do.

Looking at the 5770, you have a 156.83 Mhash/s, and based on the Mhash/w rate, that's .688705234 watts per Mhash. So, this is what the friend emailed to me:


Clearly this isn't right, but I'm not sure how to calculate actual video card power usage, but where in the above are we off?

A watt is a joule per second.

It's not 108 watts per second, its just 108 watts.  108 watts for ten hours is about 1kwh.

with $43.22/hour, you'll end up with $29,000 electricity bill to pay, fortunately for your wallet, copper wires will melt in the very first 2 minutes :D

So, it's sounding like you are saying that a watt is a unit that's already been converted to an hour. So, if I had a 1000w power supply running at full capacity, it would be using 1kwh.

Your friend used "watts per second" where he should have used "watts per hour". As a result his numbers are far, far off.

I just realized how dumb I was - kwh would be kilowatts * hours. Or hours * watts / 1000. Wow - it must be Friday.

Watts per hour also does not make sense. As others have noted watts are a measure of energy per time already, so "watts per <insert time unit here>" would be some funky kind of energy acceleration measure that nobody uses :P.

It really is as simple as:
100 W = 0.1 kW

If we use 0.1 kW for 5 hours we would get:
0.1 kW * 5 h = 0.5 kWh

Don't feel bad though, I've even seen electricians getting confused by this :)

take your power usage and multiply by time.

for example, my computer, running both GPU (8800GTS) and CPU (core2 duo e6850) miners, draws 270 watts (2.25 amps at 120 volts)

270W*24 hours= 6.48 kilowatt-hours (not kilowatts per hour, it's a draw of 1 kilowatt for an hour or 500W for 2 hours or anything that multiplies out to that same product) per day or 194.4 kilowatt-hours per month.  alternatively, 23.3 megajoules and 700 megajoules if you want the actual SI units.

at current electricity rates here (9 cents per KW-hr) that's $17.50/month.

ordinarily, not profitable for me (prior to the difficulty spike, i estimated i was mining about 12 BTC/month in the deepbit pool), but;

a. 170W of that is idle operation, as this machine is also a web server, so it already needs to be running 24x7, so it really only costs me an extra $6.50 per month.
b. I don't even pay that as my electricity is included in my rent.

Or it might pick up in Australia....

Not for me ;-)

http://www.reddit.com/r/Bitcoin/comments/fxrse/generating_bitcoins_now_financially_impractical/

Unless I have the maths wrong too, I'll be waiting ten years...

My GPU is not supported in my media centre PC so I could not try.

How many khash/sec do you get on average from a good GPU, and at what wattage?

Two AMD Opteron 6128s (https://en.bitcoin.it/wiki/Mining_Hardware_Comparison#CPUs) will get you 32.4 Mhash/s: that's the best result for CPUs listed on the wiki's mining hardware comparison list. One AMD Phenom II X6 1055T will get you 15.84 Mhash/s.  In contrast, 1 AMD 5870 GPU will easily get you 324 Mhash/s - 10x what the 2 6128s would deliver, and 20x what 1 1055T delivers. A 5870 is probably considered entry-level now for GPU mining - a 5970 will be almost twice as fast. And that's before you consider that it's relatively easy to use more than one GPU.

Hmm, $800 to buy each. Not cheap :) (5970s)

You may find the price comes down once 6990s start hitting the gamers' rigs and they start selling off their old 5970s!

What are the $ prices of 5870s like? I got a 5870 in the UK not that long ago, and there were some real bargains to be had if you were prepared to wait for them (I paid £130 second-hard from a dealer, but I've seen prices range from £110 to £210 on ebay).

Used they go for 400~500USD ea. a pair of 5970s can be found for 800$, and with OC, achieving 1.2 GH/s isn't hard... now 1600$ worth of GPU gives you ~2.5GH/s, the rest of the system is like 150$ MAX, so 1750$ for a 2.5GH/s MONSTER

What kind of PSU are you figuring in with that though? A good PSU is going to set you back and for those 4 cards you're going to need 1500watt or more probably.

I think there is no psu more then 1500w

Ultra makes decent 1600watts and I think there might be a couple 2k. The problem is in the US with 110v you might start tripping breakers about there.

My two 5970's draw 770Watts when overclocked to 900Mhz, or around 650 Watts when at factory clocks. So 1500Watts if left at factory clocks is going to be plenty, but two 1000Watt PSU's would give him the advantage of overclocking: Each 50Mhz yields 50Mhash/s boost on each card (25Mhash/s per core). So overclocking both my cards by 150Mhz to 900Mhz gained me 300Mhash/s more. However, your going to need some serious cooling to get these clocks (Clocking the memory to 500Mhz helps).

There are simple and cheap adapters you can get that allow you to easily use more than one PSU on your system: See here http://www.kustompcs.co.uk/acatalog/info_1919.html (http://www.kustompcs.co.uk/acatalog/info_1919.html)

I'm not sure about the best way to go about it though. The power isn't shared evenly by doing that right? Whichever PSU goes to the mobo is going to use 300watts for the pci-e (75 each right?) and the board / cpu. Then you've gotta split the cards up between the two PSU's. I'm guessing you could go with 1 card / mobo / cpu / other on PSU1 and 3 cards on PSU2. I'm just debating this or 1 big PSU, not sure which is the best way to go, 1PSU gives you better air flow etc.