The General idea of what Yochdog said is about right. You could do it but it be very expensive.
Bitcoin "mining" started on CPU's, but it then moved on to GPU's once code was modified and specialised to work on them.
FPGA's in the last year or so, improved upon what GPU's could do, with less power used, again modifying the code to work on them.
ASIC are this year just starting to improve on what FPGA's did. Each time getting faster and using less power.
Each change, made them more efficient. CPU's now compared to ASIC's is like comparing a Bicycle to a Jet Fighter in terms of speed.
ASIC is the latest Tech, going back even 2 steps in mining would be horribly expensive to do those sort of numbers, with GPU's.
Even a modern rendering (ATI gpu) farm would not be that efficient, other than it would be more likely to have a half decent resale value.
An ASIC or FPGA now a days is still the best thing to get, going backwards is not cost effective unless under unusual situations that effect your bottom line of costs.
Before spending any money, especially a lot, the wiki is a worthwhile read.
https://en.bitcoin.it/wiki/Main_Page
Topic: can we use a big monster server to reach 1.5 terahash/sec? (Read 811 times)
Quote
"do any of you have ideas about if we use a super server to make it work"
Even the worlds fastest super computer Titan is just made out of graphics cards like every other non-ASIC person is doing.
http://www.olcf.ornl.gov/titan/
Your question should be, "how can I get free electricity"
because all we are doing is turning electricity into currency.
Don't forget to add the cost for it's own powerplant.
You need to be more specific.
Are you talking CPU based servers? GPU clusters?
Are you talking CPU based servers? GPU clusters?
hi something similar to this
http://www.liquidnitrogenoverclocking.com/HSSM_3960X.shtml
Assuming BEST case scenarios, the expense to reach 1.5 TH/s would be utterly absurd.
Assuming 70 MH/s per core i7 processor, you would need 21,428 processors.
Lets boil that down a little more. Based on the cited figure that the server in question costs $289 per Ghz, and 1 GHz generates approximately 14.5 Mh/s ( 70/4.8 ), we can calculate the cost of this "super server".
1,500,000 / 14.5 = 103,448 <--- this is GHz needed to reach the desired hashing rate
103,448 * 289 = $29,896,551 <---- dollars needed to buy the super server
Seems like a bargain!
You need to be more specific.
Are you talking CPU based servers? GPU clusters?
Are you talking CPU based servers? GPU clusters?
hi something similar to this
http://www.liquidnitrogenoverclocking.com/HSSM_3960X.shtml
You need to be more specific.
Are you talking CPU based servers? GPU clusters?
Are you talking CPU based servers? GPU clusters?
hi guys,
do any of you have ideas about if we use a super server to make it work ,do you think we ll be able to arrive at a speed of the big one of butterfly?
meaning at least 1500 GHash/sec?
if anyone have an idea about that?
thanks
do any of you have ideas about if we use a super server to make it work ,do you think we ll be able to arrive at a speed of the big one of butterfly?
meaning at least 1500 GHash/sec?
if anyone have an idea about that?
thanks
Jump to: