6990s are a stupid buy for mining anyway. $729.99 + 8.50 shipping for ~800Mhash/sec? .93 $/Mhash, or 1.08MHash/$.

Even if you manage to cram 4 into a computer somehow (which is as I understand it not really feasible) you're pulling less than 1Mhash/$, and generating $.017/MHash (per day).

If you buy it for a casual mining/gaming machine, it's still a crappy buy as 2x 6970s are cheaper and run better, but I'm not here to argue that I guess.

And this is part of why Vegas has been suffering HORRIBLY in this recession. Jobs and wages drying up, casinos scrambling to keep up profits, etc. To the OP though, in MY ideal world people wouldn't just throw their money away on stupid frivolities, but I don't know if I'd call that the ideal world.

$10 would be a reasonable return, or possible a little below. This is the value that BTC had before the MASSIVE price pump that occured (most likely from someone with large capital to invest in what people are calling a 'pump and dump'). It will probably dip below 10 as people panic and panic and sell and sell. If bitcoin can survive the runoff, then it will likely return to around $10 and difficulty will drop from its crazy current levels as the people who bought during the $30 / 500k difficulty point return their hardware to stores in droves.

That's my prediction, and we'll see how close I am to the mark.

Please stop making these threads. There is a "beginners" that links to a thread that makes the exact same "analysis" pinned to the top of this forum.

First things first, 14 days difficulty requires ZERO increase in difficulty. Difficulty increase occurs when block rate occurs faster than predicted average should be. 50% difficulty increase requires ~7-9 days between difficulties.

Next, the fact is that these analysis make no sense because a static difficulty increase makes 0 sense. Retargets have been at 50% for... 1 difficulty change. Before that it was a 25% retarget. Before that 65-75%, before that 40% before that 30ish%.
Not to mention a static PERCENTAGE increase means an exponential numerical increase (of difficulty/hashing power). This is completely unfeasible unless price continues to rise at a similar rate.

In fact, in 2 difficulties it is likely your analysis will fall through like a brick on a wet paper napkin. Next difficulty is set to bump over 1million difficulty, but price has just fallen through the floor to around $13 (probably end at $10 or so). This makes mining a complete flip of the $30 / coin 500k difficulty point at which all the machines that are coming online now to bump the hash rate up were bought. Some people will try to stick it out because they sunk their money in so foolishly, many people will run up a massive sell-off to try to recoup the losses they suddenly find themselves taking.

Anyway, that's all short term, long term no one knows, which is why I don't like these threads. The future is not a line on a chart.

Not to be a jerk, but you seem like you must be new to the concept of large groups of people. The world is made of a pretty standard gaussian distribution. In specialized small communities you can get around this, but as with anything that becomes popularized you begin to see the more standard distribution arise again. And unfortunately that big bulk in the middle tends to be people who aren't interested in logic.

I wonder if tux ever has a hand in these runups / downs. Already almost tripled the previous days volume in the past few hours.

June 17, 2011 | Published by DWOLLA.



Pushed by Spots, GRID, and social connections, we've passed $1M a week in transactions.

Jerks don't give credit where credit is due. Right, the GRID service and SPOTS are responsible for that 

When the price of bitcoins wasn't crashing, the "Volume" was running about 30k. Even assuming that was all straight trades and not bots buying and selling back and forth, 7,200 / 30,000 = 25%. How is 1/4 of the market not significant?

Someone crushed the 1400 coin buywall and now down it goes

Someone should make a faq about the fallacy of miners and bitcoin prices. It's like the square and rectangle concept, all rectangles are squares but not all squares are rectangles.

To whit: All mining is related to bitcoin price, but bitcoin price is not related to mining.

Also: Why would anyone be mining @ 90Days in your scenario? You would be earning .755 BTC / 10 days with 1GHash/sec. At 11.84/Coin that is $8.94. Assuming 1GHash takes about 650Watts that's $17.16 in electricity costs @ $0.11 / kWh or a net loss of almost $10.

A few things of note

1) 1560Watts is Wall power. PSUs are rated to deliver their labelled rating *to the system* (this is after AC/DC conversion). 1560watts (which is higher than the real amount used, I just used it for ease of demonstration) from the wall is ~1300Watts to the system, these are non-underclocked cards to boot, so the real deal will definitely use less power. But regardless, AX1200s are overdesigned to be able to produce that. Yes it is overloading, but if you really really worry about that, drop a $15 cx430 in the mix, solved.

2) lol @ 50% load. That's potentially the most cost inefficient thing I can imagine. Not only are quality PSUs designed to produce 100% of their labelled power (well read the label, 12v will be slightly lower than total), they are designed to do it continuously. That is 24/7/365. That's why a quality PSU comes with a 5 YEAR warranty. The manufacturer knows that they won't have to make good on very many of those warranty claims.

Why not underclock your CPU on your main computer to half clocks so it will last longer? Makes about as much sense to me.

That's a fair point if you meant net. Lots of people want a magical money fountain, but if you are looking at this in terms of real investment (its more work than most real investments, but still) the returns have far far far far to go before they are so untenable as to be considered horrible in comparison to other investments.

for 8 5770s? You're crazy. You don't need 2400 watts to power 8 5870s. A full system with CF 5770s was measured by anandtech as pulling 390W from the wall. Even ignoring the system overhead, 390W * 4 = 1560W. Many people have reported their AX1200s as being able to do that without too much fuss. Not a cheap PSU by any stretch of the imagination, but certainly not $500. More like $250.


Again, you run the potential risk of burning out the mobo if you don't provide power bypass for some PCIE slots. They will also need to be single slot 5770s or else you will need to use extenders anyway. If they are not single slot you will need some sort of custom built rig to hold all the cards in place as well. Some things to consider.

Also I still don't have confirmation one way or another, but some users report issues with getting windows to recognize 5+ physical video cards. If you planned on linux anyway not as much an issue.

I spent thousands of dollars on hardware, and I'm currently sitting on thousands of dollars worth of hardware and more money in the bank than I started with. Lots of people looked at me with bemusement back then as well. Fact is we don't know what's coming up, and which people to envy and which to pity.

My only hope is as said above, people aren't gambling with more than they can afford. Don't risk your home/wife/kids over bitcoin.

Maybe not, that was an exceptionally efficient build that I'm proud of, but it still stands that you need to be very efficient if you want to profit, and < 1Mhash/$ is very inefficient. Even at stable difficulty and $20 coin prices, it wouldn't pay itself off for almost 1.5 months. A bad goal to shoot for.

My point isn't to build my build, I was lucky, it was just that you need to be as efficient as possible, and have numbers to approach, if not asymptotically.

You can still build somewhat efficiently, for example a 4x 6870 rig pulling 1.2GHash/sec can be built for maybe $900 if you build smartly. Already up to 1.333mhash/$. There may be other builds that are more efficient, I'm currently amp capped though so I haven't been looking to build more lately

While it's true that "750watts" is enough, _s3v3n_ makes a good assertion that we don't know WHAT MAKE/MODEL of PSU he is using. If it's an HEC crapPSU750 it's more than possible that it is totally f-ing his rig. Hell some much older standards (ATX 1.0 or whatever) would dump like 60-70% of the amps onto the 5v and 3.3v, so yeah you'd have 750W but hardly any on the 12v where modern comps need it. It's also possible it's old and dying, bad capacitors, etc. Diagnostics require information, don't just discount any possibility.

I'm wondering about overheating issues however. What kind of temps are you seeing on your monitors? Closed case? Open case?

A quick calculation shows that is a horrible rig efficiency. 1.06 $/Mhash ratio (or .94 Mhash/$ ratio)? With declining price, increasing difficulty, you need to become as efficient as possible in your builds, and this is far from it.

As a comparison, back in april I built a 1.7GHash machine, and I spent around $985. Even at 1.73Mhash/$ I still considered it somewhat risky, though it certainly paid off in the end.

Also, be wary of running 8xPCIE cards on a single board, unless you are using risers with power splitters. Even lowly 5770s can pull a lot of power through the board if you use 8 of them, potentially destroying it.

$10

So I'm not totally sure what forum to put this under, but it sort of seems appropriate here.

I was looking through the block chain recently, for reasons of my own and I noticed something sort of surprising. http://blockexplorer.com/b/131227

Block 131227 had moved 133,232 bitcoins! And this was in a few mostly concentrated transactions, 49,000 in one, 29,000 in another, etc.

Then I checked back a little bit and again http://blockexplorer.com/b/131223 block 131223 130,015 bitcoins moved. 49k in one, 29k in another, etc.A lot of the nearby blocks have these huge transactions, 79,000, 125,000, 157,000 etc. So I was curious, maybe bitcoin has always been like that and I just haven't noticed, so I skipped back in time a bit. In the past day or so thre have been a ton of 100,000+ BTC transfer blocks, but in previous days past it has been almost totally unseen a volume of transfer that high.

Curious what people think of this, or if I'm just not familiar enough with block history that I'm missing something.

Increased difficulty correlates to more hashing power being put on the network. Hashing power is expensive, even AMD. A GOOD MHash/$ ratio is about 2.2, or $/Mhash of .45; meaning an increase of 50% difficulty is an increase of 50% hashing power. Currently we're at about 6.5THash? So 50% diff inc is about 3.25THash added to the network, or $1,462,500 in gpus alone (not counting the psus and cases and mobos, and whatnot). And that's IF you can even find such efficient miners (stock is thinning, most people will only be able to find 6xxx series cards that are about .55 $/MHash). Compound the fact that a 50% increase in difficulty means you would produce 67% as many coins as you made last difficulty round (and if you see another 50% difficulty increase, thats 45%), and you can quickly see why without an increase in coin price, it is very unreasonable to expect the network to continue to grow.

This can be a good thing or a bad thing. But as profits diminish you will definitely see something big changing, rather than a static model like you'd like.