However, hashing isn't actually a FLOPs operation. The network doesn't really do floating point calculations.
You are correct, but it's the closest equivalent we have data on. FLOPS roughly kinda sorta correspond to hash speed.
Topic: Total network processing power (Read 4156 times)
However, hashing isn't actually a FLOPs operation. The network doesn't really do floating point calculations.
You are correct, but it's the closest equivalent we have data on. FLOPS roughly kinda sorta correspond to hash speed.
A 5830 make 1.79TFLOPs and make like 300Mhash if properly clocked, so you can suppose every 300mhash there are 1.79TFLOPs...
However, hashing isn't actually a FLOPs operation. The network doesn't really do floating point calculations.
As Binary said, i suppose bitcoinwatch just checked how much mhash a video card do and checked his TFLOPs.
A 5830 make 1.79TFLOPs and make like 300Mhash if properly clocked, so you can suppose every 300mhash there are 1.79TFLOPs...
A 5830 make 1.79TFLOPs and make like 300Mhash if properly clocked, so you can suppose every 300mhash there are 1.79TFLOPs...
The number Bitcoin Watch gives is remarkable if it has any basis in reality. I am just looking for a way to verify that it has a basis in reality.
Well, videocards contribute the vast majority of FLOPS to the Bitcoin network. You can look on this page to see the FLOPS of AMD videocards, which are the most commonly used for mining. Maybe Bitcoin could do a card survey to figure out who has which cards, and then it would be trivial to calculate FLOPS.
The number Bitcoin Watch gives is remarkable if it has any basis in reality. I am just looking for a way to verify that it has a basis in reality.
It should be a little over in the hashing domain. Note that you should only compared the total hashing power hash/s instead of FLOP/s and there is no direct conversion. There are unknown fraction of people even using the FPGA to do the job, so it is difficult to compare.
OK... my understanding now is that the Bitcoin software does not even do floating point calculations, so the correct measure of its power would be integer operations per second and the flops number is some kind of estimate. Does anyone know how these numbers are arrived at by Bitcoin Watch or at least have another source with similar numbers?
I'm not surprised that "omg I get rich fast!" beats "If i waste enough electricity, I perhaps somehow, somewhen contribute to something I barely understand because I'm no PhD in biology, but hey, I think it makes the world better" 
Well, I think they're into double digits now. It was 5.1 petaFLOPS avg. on April 21st, 2010 (see here). By now it's probably into the low 10s, but nonetheless, Bitcoin easily trumps BOINC.
Appears to be only about 5.6 petaFLOPs. If that is the next largest then Bitcoin is more then 20 times more powerful then 2nd place.
Right but even compared to other distributed computing projects Bitcoin is huge. The largest other project that I am aware of is BOINC and I believe it still only has single digit petaFLOPs at its disposal.
Well, I think they're into double digits now. It was 5.1 petaFLOPS avg. on April 21st, 2010 (see here). By now it's probably into the low 10s, but nonetheless, Bitcoin easily trumps BOINC.
The top supercomputer in the world has processing power of 8.162 petaflops, see here. It's a lot easier to generate computing power with a distributed network than one computer.
Right but even compared to other distributed computing projects Bitcoin is huge. The largest other project that I am aware of is BOINC and I believe it still only has single digit petaFLOPs at its disposal.
No, he divided by 1000. Metric prefixes are confusing, but he is correct.
No, they really aren't. If I'm too tired to see that, I should probably go to bed
According to Bitcoin Watch the total processing power of the Bitcoin network is over 100 petaFLOPs. If this is correct then that is more power then the top 500 supercomputers in the world combined. Very cool if it is true.
The top supercomputer in the world has processing power of 8.162 petaflops, see here. It's a lot easier to generate computing power with a distributed network than one computer.
That's TeraFLOP/s..
No, he divided by 1000. Metric prefixes are confusing, but he is correct.
That's TeraFLOP/s..
According to Bitcoin Watch the total processing power of the Bitcoin network is over 100 petaFLOPs. If this is correct then that is more power then the top 500 supercomputers in the world combined. Very cool if it is true.
Jump to: