Sip Posted April 21, 2002 Report Share Posted April 21, 2002 I just saw THIS ARTICLE ... about the current fastest computer in the world ... pretty amazing!!! It's made by NEC and is called the Earth Simulator. It is intended "to analyze climate changes, including global warming, as well as weather and earthquake patterns". It has 5,104 processors and could reach a theoretical speed of 35,600 gigaflops. They said yesterday it reached 87% of the peak theoretical speed Just mind boggeling!!! In the time it takes you to blink, this thing can do about 30 TRILLION OPERATIONS ... just to put it in perspective ... I have been alive for 26 years ... if I had done one operation every second of my life, I would have done less than 1 billion things so far ... it would take me about 1 million years to do as many things as that computer can do in 1 second ... that's without any rests!!!! Quote Link to comment Share on other sites More sharing options...
Azat Posted April 21, 2002 Report Share Posted April 21, 2002 Sip,Your comparison of the speed is not very accurate. Suppose that computer has a face recognition software in it with about 100 faces in the database. And suppose you had 100 friends and relatives that you could recognize easily. I would guess(maybe I am wrong), but you would recognize the face of one of those 100 faster than this computer could, if not faster than the speed would not be that much different. Also, something simple getting in your car and driving to school. Just think what an easy task it is for you. Can this computer with all its powers be able to perform that task if it was mounted in a car with all the things that are necessary to drive a car.(I don't want to list the details) I bet that if it could drive, it would have a super hard time trying to keep up with everything that a normal human being does while driving a car or doing simple tasks. But other than that I do think that it is cool what they have accomplished. I worked on a system management tool that we installed on a 500 proc IBM AIX machine at Cornell University 7-8 years ago and at that time that was one of the fastest in the world, but I imagine that a 4 proc P4 machines are just as fast now as that system was then. Quote Link to comment Share on other sites More sharing options...
Garo Posted April 21, 2002 Report Share Posted April 21, 2002 Azat I think Sip is correct. It is the fastest computer for calculations. As for recognizing faces or driving a car I think its mostly software problems rather than hardware.Eventually those kind of softwares will become more effective (eg. Deep Blue for playing chess) and will do the job. And perhaphs it'll happen by using neuro-computers and artifical intelligence systems that have more similarity with the way human brain functions.As for now I think it is correct to compare computers with humans only in the fields the computers are intended for (calculationa, analyzing data using strictly defined logics and so on). Quote Link to comment Share on other sites More sharing options...
Azat Posted April 21, 2002 Report Share Posted April 21, 2002 I completely understand what you mean Garo. I am not saying that I or any human can calculate faster than that computer or for that fact any computer. But comparing it to the humans ability to do the simple things(and some may say that you are constantly calculating the things to do while driving, or recognizing faces or anything) computers have a long long long way to go. I actually think that it is not just software by the way. I agree that software needs to get better, but computers have to get MUCH faster as well. Quote Link to comment Share on other sites More sharing options...
Sip Posted April 21, 2002 Author Report Share Posted April 21, 2002 It does 30 TRILLION THINGS a second!!! Sure, it's still no match for the human brain for performing complex tasks because as Garo said, it's hard to "program" those things. The problem is that current chip hardware are static. They internals don't change during the lifetime. In our brains, the neuron connections keep changing and updating throughout our lives. Now we have billions of neurons and they have many many connections which "update". These computers are soon going to have billions of transistors but they will still be no where near the adaptability and "learning" of a human brain since their connections will still not change (much) as they run. But for "specific" tasks .. Wheeeeeeooooooooo ... 30 Trillion things a second!!!! And it's only going to grow exponentially as we know Quote Link to comment Share on other sites More sharing options...
Guest Fadi Posted April 21, 2002 Report Share Posted April 21, 2002 Sip O.K 30 trillions, but still this don't mean 30 trillions of things, correct me if I am wrong, but this only mean, 30 trillions of calculations, how many calculations would it takes for each average "things" I remember about the Deep Blue vs Gasparov, they were talking that Gasparov could think of 3 possibilities per second at max(If I remember right) against billions for the computer, but still the moves that Gasparov was thinking of were good ones, while the billions of probabilities from Deep blue, the majorities were not good ones, because the computer was just selecting all these probabilities everytime a new move was playing, when Gasparov during the game, his adaptation was in a way, that at the next play, the 3 moves per second he was thinking were selective moves that his brain automatically selected without giving all the possibilities... These are also "things" that worth billions. Quote Link to comment Share on other sites More sharing options...
Azat Posted April 22, 2002 Report Share Posted April 22, 2002 One more thing to throw in there. It was not because Deep Blue used better software to beat Kasparov last time they played in comparison to the time before. It was that Deep Blue was something like 5-6 times faster and could think of all the moves and best probabilities in the allowed time. Quote Link to comment Share on other sites More sharing options...
HaYaStAnRuLeS Posted April 23, 2002 Report Share Posted April 23, 2002 Deep Blue beep Kasparov becuz Kasparov made a mistake... unlike humans, computers rarely make mistakes!!! i say "rarely" becuz hardware or software glitches r possible, but very unlikely. Now as 2 the reason why Deep blue won... If you were to analyse Kasparov's moves, you would have observed a slight error in his calculations, something he l8er admitted to have done... It wuz a mistake, & the computer caught it after subjective analysis of the play... Now the amaizng thing here is that a human brain that is capable of doing only 3 calculations a second is not only as good as the supercomputer, if not better, but it also shows a sense of intuition that immediately eliminated unnecessary calculations... that's the lession we have 2 learn here... n e wayz, being an avid chess enthusiast myself, i could confidently state that Kasparov is more than capable of beeting any computer chess program at least once... And anything that happens once can happen again, especially when computers subjectively play "perfect" matches every time, cuz their calculations r alwayz right, no matter if they win or loose... in other wordz, they never make misatakes, but a human can still beet them... whatever, i hope this helps Quote Link to comment Share on other sites More sharing options...
Sip Posted April 23, 2002 Author Report Share Posted April 23, 2002 Well, the fact of the mater is that computers are getting faster and faster and faster ... at some point, they will be able to consider sooooo many possibilities within the "allotted" time, that no human will be able to beat them. The best they can hope is probably a draw. I don't know too much about chess (I know the basics and I can beat the computer in level 1 about 10% of the time). BUT I am assuming that there is no "solution" to chess. Which means there is no game that you can play which will guarantee that you will win. From what I have learned, I think that it is a valid assumption. I don't remember seeing any proofs for this so there could I guess be a "solution" to chess... anyone know if there are any existance proofs regarding chess solutions? In any case, if the above assumption is ture, there is little doubt in my mind that the computer will easily WIN or get a DRAW against any human chess player ALL THE TIME. Just think, if the computer can exhaustively search ALL outcomes for lets say 10 moves, there is not much that a human player can do! I don't think any human has the capacity to see ALL POSSIBILITIES anywhere near 10 or more moves much less even try to plan for it [ April 22, 2002, 09:02 PM: Message edited by: Sip ] Quote Link to comment Share on other sites More sharing options...
Sip Posted April 23, 2002 Author Report Share Posted April 23, 2002 Just to clarify something ... computers right now are not capable of considering enough possibilities to be able to beat humans all the time ... they have to rely on sophisticated "heuristics" and play based on strategies ... I am not sure what deep blue uses but the old alpha-beta algorithms come to mind. However, with such power (when we start talking trillions of things a second) then starting to look at every option exhaustively, all of a sudden becomes reality! Throw out the old "smart" and "intelligent" algorithms and just look at every possible outcome like a dumb machine Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.