Ha ha. :rolleyes: 
Since you’re obviously very smart, help me out here. I’m having trouble wrapping my (practical) mind around that kind of processing speed. Give me an idea how long it would take to do something a little complicated, like read a (digital) bible or encyclopedia.
Something like that.
Ah, simulation. That’s what they’re after, huh?
Simulation is a common use of massive computing power, and since Argonne National Lab is getting one[li] of these toys, I’d make a WAG that they’ll be doing weapons simulations (for example stepping through the process of an explosion. More computing is always useful for this sort of thing, as you can simulate smaller and smaller spatial regions in finer and finer time slices.)[/li] correction: got one. I just noticed that the article I’m looking at is from 2007.
Sure, when you tell me how floating-point math relates to reading, and do it using some numeric ratios.
Seriously, this is a rather strange request because it simply isn’t an apples-to-apples comparison. It isn’t even a chalk-to-cheese comparison. It’s more like a starfish-to-pencils comparison.
Yeah, I see that now. I didn’t mean to actually read, just looking for something with a lot of data to process.
I’ve moved on now, from bibles to the Big Bang.
I din’t suppose I could get one of these in my macbook?
OK how about this? It can do over one quadrilion 7th grade math questions a second.
Or let’s assume there are 10 million 7th graders in the US. Let us assume that they are assigned 100 math problems a week for 35 weeks a year (allowing for spring, summer and christmas breaks). That means 7th graders in the US are assigned 35 Billion math problems a year. A Petaflops computer can, every second, complete over 28,571 years worth of all US 7th grade math problems.
Yeah - but can it tell me what time the train from boston will collide with the train from chicago if they left at the same time heading in opposite directions at 72 and 68mph repectivelly?
mangeorge: I think your quest to fully understand these systems by looking at how much math they can do is futile. They do more math than you can comprehend every single second, which means the numbers form up into tight little balls and bounce off your skull. 
You have to go up a few levels and look at the applications of all that math in terms of an end result, such as modelling proteins: Proteins, the building blocks of all cells, are complex devices that are not only made of the usual chemical bonds but are folded in on themselves in very complex ways under the influence of hydrogen bonds. That folding dictates their function: When all goes well, we get a functional cell and a healthy person. When something goes wrong, we get diseases like Huntington’s, mad cow, and certain cancers. Getting a real understanding of how proteins fold and what can go wrong is difficult because there is just so much data to go through. I think you see where this is going.
You can contribute to the understanding of protein folding by running some software.
They won’t collide, of course. They’ll reach vanishing point first.