Bitcoins
Did he get to keep the 20 he "mined"?
The US National Science Foundation has suspended a researcher for using two university computers to mine Bitcoins. In its biannual report [PDF] to the US Congress, the government agency revealed the misdoing and recommended action: The researcher misused over $150,000 in NSF-supported computer usage at two universities to …
if he was any good he would have written a flowery proposal justifying the mathematical work being carried out, and then saying the bitcoin generator was the application to solve it...
I must say I had a giggle when I saw the value vs electricity measure, which is pretty much what the bit miners themselves say is the in built cost of mining them...
P.
"I must say I had a giggle when I saw the value vs electricity measure"
$150,000 wouldn't have been the cost of the electricity, it will be the wooden dollars the uni assigned to the total cost of buying, building and maintaining the machine (and the lecky) spread over the life of the machine.
"if he was any good he would have written a flowery proposal justifying the mathematical work being carried out, and then saying the bitcoin generator was the application to solve it..."
...and demonstrating that wasting computing power contributes to global warming thus doubling his future grants.
What would you need in terms of regular hardware to mine the same in the same period? I'm trying to get a feel for if this supercomputer was equivalent to thousands of dedicated graphics cards or a handful of mid-range cards on a single box. After all, a GPU will crunch through these far, far faster than a CPU would. Unless the super-computer is designed to do the same sort of calculations it could be a tremendously inefficient way to mine.
Anyone here know much about super-computers?
from what I read bitmining has already moved to ASICs...
Supercomputers are general purpose machines, that are optimised to solve certain classes of parallel algorithms. If you can make an ASIC for a scientific problem (google ANTON or GRAPE with keyword molecular) and you can get an appreciation of the mathematical problem solved, and the data transport and storage issues. FYI the supercomputers at ORNL (and other places) are chock full of GPUs as well as CPUs. The biggest challenge is cooling!!!
I was sort of being tongue in cheek above about rewriting the application, but in all seriousness, if you can formulate your problem to exploit the mathematical sweet spot of your computer, you cannot do better!!
Hence, LINPACK. It might not be the best benchmark for my application interest (molecular dynamics), but it is an objective test that is relevant to a great deal of scientific problems, and one can gauge the machines effectiveness.
P.