The supercomputer I am researching has 2.2 petaflops and boasts total memory of 1000 terabytes and disk space of 23.5 petabytes.
Is this more computing power than the total of the entire worlds computing power 10 years ago (2004)?
$\begingroup$
$\endgroup$
1
-
$\begingroup$ You might be careful about how you think of "computing power"; see my comment(s) on David Richerby's answer. Are two 2GHz processors more powerful than a single 3GHz processor? Or does it depend? $\endgroup$– Patrick87Jun 20, 2014 at 17:22
Add a comment
|
3 Answers
$\begingroup$
$\endgroup$
3
No, and not by a long way.
Take 25,000,000 ordinary desktop PCs from 2004. To exceed the supercomputer you're talking about, each would need an 88megaflop CPU, 40Mb of memory and a 940Mb hard disk. In 2004, CPU clock frequencies were already in the GHz range, RAM in the hundreds of megabytes and hard drives in the gigabyte range. And there were many more than 25,000,000 computers in the USA, let alone the world.
answered Jun 19, 2014 at 22:47
-
$\begingroup$ While accurate, I wonder whether this fails to address the real question. Today, when they're putting together the Top500 list, they don't just count cores and multiply by frequencies to determine the winner. They run benchmarks (LAPACK, etc.) and see which one does better in practice. I would guess that the interconnect might level the playing field significantly between "all computers" in 2004 and a single supercomputer now... though, of course, I might be mistaken. Do you think there's any merit to this criticism? $\endgroup$ Jun 20, 2014 at 17:17
-
1$\begingroup$ (I admit that this answers the question as asked, but mean to point out that the question, as asked, might not be a meaningful one in the sense the questioner wanted.) $\endgroup$ Jun 20, 2014 at 17:18
-
$\begingroup$ @Patrick87 I agree with your point to some extent. Just adding up megahertz and megabytes as I did is a dumb way of working out what's the most powerful. However, working in this naive way, 25 million computers out-performed today's top supercomputer. Since there were really far, far more than 25 million computers in the world, I figure that, even if you used an intelligent metric, the huge pile of everyday desktops would win. $\endgroup$ Jun 20, 2014 at 19:56
$\begingroup$
$\endgroup$
EDIT: I realize my answer is a bit off-topic, but I let it anyway, it could interest the OP.
I cannot tell you if the most powerful supercomputer today is more powerful than the sum of all computers 10 years ago (I don't think so), but I can tell you that the most powerful supercomputer today is about 30 times more powerful than the sum of the best 500 supercomputers in the world 10 years ago: http://top500.org/statistics/perfdevel/
$\begingroup$
$\endgroup$
2
Here's another way to think about it that might help your intuition a bit. Moore's Law states that transistor density double every 1 to 2 years (you'll find varying numbers here). Let's be optimistic and assume that this correlates to a doubling of computing power every year. In 10 years we should expect supercomputers to be roughly $2^{10}$ (1024) times as powerful as they are now.
-
1$\begingroup$ How does that help? The question is about comparing supercomputers in year Y with other computers in year Y, not about comparing something in year Y with something in year Y+10. $\endgroup$ Jun 20, 2014 at 14:51
-
$\begingroup$ Actually the question is about comparing a supercomputer in year 2014 with all computers in 2004. the Moore's law argument is just to show that we wouldn't expect a single super computer to be nearly as powerful as the collective computing power of 10 years ago. $\endgroup$– WcrousseJun 20, 2014 at 16:02