10 replies to this topic

[Time_Traveller]

IBM's Sequoia has taken the top spot on the list of the world's fastest supercomputers for the US.

The newly installed system trumped Japan's K Computer made by Fujitsu which fell to second place.

It is the first time the US can claim pole position since it was beaten by China two years ago.

From http://www.bbc.co.uk...nology-18457716

Well Done to IBM for overtaking Japan's Fujitsu.

Edited by Time_Traveller, 18 June 2012 - 12:36 PM.

[Raklian]

Sequoia is a little over 50% faster than Japan's K supercomputer and it only uses 60% of the energy K consumes. . That's a remarkable jump.

[wjfox]

It won't be long before Sequoia itself is overtaken by another. I'd say maybe 6-12 months, given how fast things change.

http://www.futuretim...ture-trends.jpg

[Raklian]

Yeah, things are starting to change faster than we can track them. Hurry up with the neural impants!

[Anu]

A pity hat will be for nuclear blasts simulation's

[mic of orion]

It won't be long before Sequoia itself is overtaken by another. I'd say maybe 6-12 months, given how fast things change.

http://www.futuretim...ture-trends.jpg

U know Will, nobody knows how fast or how capable human brain is, this is what nuroscience is all about, reason why we can't repair brain damage now is simply because we know too little about how human brain works, we have some ideas, and we know what some regions of brains do, but to compare human brain to a piece of machine is bit stupid tbh.

How fast is human brain, I have no idea, and I bet no one else does.

Just to illustrate my point, to post this took me only 1-2 minutes, but in that space my brain did well over 10 on 50 calculations. Just perception and vision took half of that.

[wjfox]

Actually, it's surprisingly easy to quantify the brain and its processes, it's less complex than you'd think, and there's tons of redundancy in its structure.

Read The Singularity is Near. Also check out the Blue Brain Project.

[mic of orion]

^^

I've read on blue brain project, it was more of a pr stunt than a real attempt to simulate human brain or answer some of the most basic questions. Guy even said they'll try to make a human brain simulation by 2020, its 2012 and they already saying they think 2030 target is achievable, I bet 2030 comes and goes they'll be nowhere near.


Human brain is way too complex to understand it in few years of computer progress, its like Genom project, it took large number of scientist and research institutes to complete it, it would take similar endeavor to uncover how human brain works, all nitty gritty stuff too. Once we really know how brain works and what makes human brain so unique that perhaps than we'll be able to replicate and build artificial or even grow new brains from stem cells. For that to happen we are at least 40 years away.

[wjfox]

Funny you should mention the Genome Project, as it's precisely this that Kurzweil compares in his book. It was expected to take centuries to complete, but due to exponential progress it was finished in less than 15 years. Also scanning resolution, bandwidth, storage, etc. all are increasing exponentially. We can quantify things like neurons, and on current trends we'll have mapped the entire brain and its regions by 2025. These trends follow a very smooth and predictable path, like Moore's Law.

Understanding all of its processes and nuances will obviously take longer, but we'll at least have a working, physical "framework" available to study in the near future.

[Raklian]

It is likely we will shortly enlist the aid of artifical intelligence to help us understand the brain's processes and nuances a lot faster than without this assistance. All sorts of tools that don't exist today will be created in the next few years or decade - that is the power of exponential progress. It seems a lot of people don't consider it when making predictions about technological advances.

[mic of orion]

Funny you should mention the Genome Project, as it's precisely this that Kurzweil compares in his book. It was expected to take centuries to complete, but due to exponential progress it was finished in less than 15 years. Also scanning resolution, bandwidth, storage, etc. all are increasing exponentially. We can quantify things like neurons, and on current trends we'll have mapped the entire brain and its regions by 2025. These trends follow a very smooth and predictable path, like Moore's Law.

Understanding all of its processes and nuances will obviously take longer, but we'll at least have a working, physical "framework" available to study in the near future.

Yes I used genom project as a prescient, if such scale research is done on human brain, than by 2030 we might be able to map human brain and understand and perhaps even build out of scratch another human brain. but nuroscience is such complicated field (most complicated medical research branch) that we are at the very begging of very long road.