I recently had a conversation in which someone asked me to help build them a bitcoin mining machine. I explained that looking at the mining configuration hash rates, they should not bother unless they had access to a new ASIC miner.
The problem is even if they do not have to pay for electricity (e.g., it is “free” somehow) the difficulty rating is rising so fast (double digits at every adjustment) that even a top of the line ATI GPU is equivalent to “old school” CPU mining at this point. They would get mere fractions of BTC even if they set up two or three ATI 7970s.
For example, check out this bitcoin mining profitability calculator and plug in the numbers… only gigahash (and now terahash) producing ASICs will get anything useful out of it.
A souped-up, overclocked Radeon 7970 will do around 700 Mhash/s for BTC. So let’s say you connect three of the best Sapphire Vapor-X, they are listed at $440 a piece. Factor in the rest of the system (cheap hard drive, ram, cpu, stable power supply, milk crate) and you are pushing at least $1,700 yet only producing 2100 Mhash/s.
Plug that into the calculator, the break even will be >10,000 days because you are only mining 0.0017 BTC a day. Assuming you did it for 365 days and the rating difficulty didn’t increase (which it does at double digits every reset) you would end up with 0.6205 BTC in one year. But again, this wouldn’t happen, more than likely you would get less 0.1 BTC the first year. And nothing at all the year after. Why? Click here to see the visualization of the bitcoin difficulty rating this past year.
Again, you would be much more profitable mining Litecoin, for example, if we use that same system you would end up doing about 2100 KHash/s and generate about 1.42 LTC / day (here is a litecoin estimation calculator). Note: LTC mining is extremely memory intensive, hence the substantially fewer hash/sec.
But even then, this would decline as the difficulty rating increases each reset. Best case scenario, that rig would get about 400 LTC/year. The highest a LTC has sold for thus far this year has been about $20, so you could earn $8,000 in a year (assuming you sold the coins and didn’t save or reinvest them).
Thus, it is far more profitable to mine LTC than BTC.
But let’s go back to the 0.0017 BTC you would make if you set up that system today. Even in the rosiest of conditions, where one BTC was worth $1 million, .1 would be $100k, .01 would be worth $10k and .001 would be worth $1k.
Therefore, even in the most optimal scenario, there is no GPU solution that is profitable mining BTC, even if the electrical costs were free.