Does anyone think supercomputers are pointless?
Miscellaneous Forums/General Discussion/Does anyone think supercomputers are pointless?
http://news.bbc.co.uk/1/hi/technology/5322704.stmsee basically they become obsolete pretty quickly, and they must cost a bomb with thousands of CPUS, and then they just run some crappy simulation on them. Is anything learning from making these, like sending men/robots into space, or is it just pointless?
The article mentions the potential of such a computer in aiding research into Alzeimers disease. If it can do this, and make progress in discovering a cure, then then building the computer will be worthwhile. I'd hardly say that such an effort is "crappy", would you?
I figure on such a system one could use raytrace @ 1280x1024 and whatever depth you want to play any 3d shooter, in landscapes filled with thousands of recursive trees and plants while every glassy object has correct refractions and photons, oh.. and real atmospheric media ofcourse.
he new computer is a "hybrid" design, using both conventional supercomputer processors and the new "cell" chip designed for Sony's PlayStation 3.
Yep, that's going to be one expensive $^@*$!.
Think it'll be fast enough to do raytracing graphics faster/better-looking then our current polygon-based graphics cards?
this is now fastest comp on the planet..just few years back it was BlueGene..BlueGene use to be builted also with astronomical expenses..so, we having now Roadrunner, also damn expensive, but still nothing out of it worth for typical human being on this planet, as well as happen with BlueGene..i doubt they using this stuff for 'humanitarian' projects and simulations..maybe new good chess program capable to understand that still is long way to defeat human brain...useless computers if you ask me..
Yeah, I would hardly call decoding the human genome faster than they ever thought possible "crappy"...
And as for the chess program... Well, Kasparov might tell you that a computer isn't smarter than him... But I think his record against computers speaks for itself. 1 out of 6 of the games in his last match against Deep Blue. He won the first match by the skin of his teeth. And there's so few people in the world that can match Kasparov on the chess board that I really have no trouble saying that a computer can beat a human at chess any time it feels like it. :)
most of this supercomputers are used for really useless or destructive purposes
who cares if a computer can beat the world champion of chess, it is a huge waste of money. now if they use it to analyze a complex disease or bio-weapon to help them create a cure then it is totaly worth the money
It's going to be used for weather which I think is great. Just about any step forward in human ability is a bonus in my opinion. You don't have to equalize all humans to be humanitarian. When one human being reaches for the stars another is lifted from the gutter via social tether.
When one human being reaches for the stars another is lifted from the gutter via social tether.
Hmm.. Reagan :|
the weird thing about supercomputers is that it's not uncommon for it to be faster/cheaper to get the results of major (extremely time consuming and complicated) calculations to wait a acouple of years first and THEN buy the system.
You'd get a much faster 'machine' than the original one would have been, at less cost (especially when factoring in operating costs such as A/C requirements and MASSIVE power consumption by a cluster itself). When you're talking projects that can easily take years of work to yield results in some cases, the results could still be in sooner from another state-of-the-art supercomputer that you bought and deployed two years later than the original one...
Interesting dillema when planning such long-term projects. Buy now, or wait? counterintuitive as it may seem, waiting can be by far the best option both in cost and time-to-results.
I made that up actually, I'm sure someone else has said the samething in different words but that's mine! Copyright 2006 or what have you ;)
Hmm if you think that research and simulations for fields like weather/climate/physics/medicine/... are worthless then well, go with it.
I've never understood the attitude sending some humble beeings into space is of so much importance when there is still so much to do on earth.
Anyway it's discussable how much pointless it is to use them for military research. For the good i would say they make less tests in real. For the bad they shouldn't do this at all.
And if there is a more economical way to do this then go for it.
xlsior: Yeah you could wait forever and never buy.
So basically if they use the computer fo medical research etc fine, but couldn't it be done on normal PCs running a SETI-like program and like the PS3 shared task thing?
I guess what I meant is does building a computer like that further the knowledge of general computers at all (like space technology filtering down to us all) or just the knowledge of super computers?
Well, i'm not much into this subject these days as the times i was impressed by something like a cray are a bit over but i guess there are several issues like:
a) Do you want your data/client to be shared (also if it's encrypted on the internet)?
b) After the one time investements for building up such a system how large are the costs for paying the energy bill, the space where you store the system, the people who have to maintain the infrastructure and the complexity to administer it?
c) How reliable is the system?
d) How fast is the interconnection of the different cpus to each other and the data needed for solving the problem best?
e) How good are your contacts if you don't work for IBM or Lockheed Martin?
I guess it's just as pointless as people going out paying wads of dough for the best PC computer for it to be old within the end of the month.
I can remember a Pentium 166 costing exactly 300 pounds more than a Pentium 133 back in say 1998. I mean 300 pounds? It was barely any faster, especially by today's standards.
Heheh yeah I'll bet the PCs we own today, we all own a super computer of say 1995 or 2000. Just a shame that Windows almost drags it down to be super average.
Sigh... I always wanted a TMI CM-5, but they do seem hard to come by these days.
Andy
they must cost a bomb with thousands of CPUS, and then they just run some crappy simulation on them. Is anything learning from making these, like sending men/robots into space, or is it just pointless?
I think this is the demo machine for MS Vista !!
who cares if a computer can beat the world champion of chess, it is a huge waste of money. now if they use it to analyze a complex disease or bio-weapon to help them create a cure then it is totaly worth the money
The Chess match was to raise awareness, bit of advertising, bit of prize money but, mainly, a testbed for new technology.
Taken from some gumpf...
While Deep Blue was a computer system designed to play chess at the grandmaster level, the subsequent underlying technology of Deep Blue currently solves many real world challenges, including weather forecasting, financial market modeling, automotive design, and medical research and development.
Hello.
I'd really, really like to play Larn on it :o)
Goodbye.
yeah think how big the dungeons could be :-)
Perhaps it could do a real-time text adventure...?
Ah come on...you know that this is impossible!
Scientists tried this since aeons and it never happened. Also the military was so much after this but even they never got it right.
I remember there were rumours that germans scientists during the world war two found something out about this and it had to do with an artifact found by the team of Dr. Steinheimer...but you know rumours...
> I guess what I meant is does building a computer like that further the knowledge of general computers at all
YES! A couple months ago I attended a discussion of the Apollo spacecraft computer. The number of advancements they had making the computer for that was incredible.
..UberSolider..
or is it just pointless?
I don't think so, while they can be very expensive to build, they are part of the whole history of cumputers in general and as such they can be considered as somewhat "necessary".
interresting read :http://www.thocp.net/hardware/supercomputers.htm
Do you think we would have gone up to where we are today without them ?
If people really wanted to they could spend more money doing more "useful" things but why at the same time discarding the making of SuperComputers ?
Super computers aren't just one computer. It's many put together. Thousands of times faster then home pcs. They are used for simulations that need to be rerun for thousands of different scenerios. It's like having a render farm for scientific data.
I'm still not convinced ;-) ...
without computers?? You kidding me?? Where we are now is not becouse of computers but becouse of hardwork and knowledge of peoples whos spent their lives to discover such a 'trivial' things we are messing with today..to be honest, computers is the one of the main reason why society today is full of peoples with asocial behaving..peoples less meeting each other in live etc...of course, computers is necessary today regarding lifestyle, and i am okay with it, but, dont sell me stuff like 'where we will be without them'...well, I'll tell you .. I'll be now with my girl doing some more constructive stuff, than chatting now with her over crappy MSN and posting funny posts here...yiiikkkk
@Naughty Alien
Don't misinterpret me, computers (and super computers as well) help saving lifes, help people share knowledge, help people communiacate more, ...
Now it's all up to you, you are free to use them or not,
why don't you just stop using your computer and spend more time doing other less "trivial" things ?
People who are associal will always be, don't blame machines for that. You kinda luddite or what ?
where we will be without them'
i don't know.
Takes me back to my childhood :
"Why Don't You Just Switch Off Your Television Set And Go Out And Do Something Less Boring Instead?" Next on BBC1 the Double-Deckers.
.
<edit> I was going to argue about the 'Cult' rating but [a http://en.wikipedia.org/wiki/Why_Don't_You%3F] 42 series over 22 years? [/a]. Blimey!
yes I remember that, "and go out and do something boring instead (of coding)" was what I felt about it though ;-)
I think computers in general are boring, but they have their uses. Same goes for super computers. I used to watch Why don't you with all those kids from Cardiff in. Used to prefer the science stuff by Jonnie Ball or whatever his name is :)
yeah Jonnie was alright, that Cardiff show was the pits though.
but couldn't it be done on normal PCs running a SETI-like program
That greatly depends on the specific task: SETI-at-home is only suitable for very repetitive tasks where you have a small data packet that needs to be processed. If you need a lot of communication and rapidly changing data that affects large quantities of other data, it's pretty much impossible to run in an off-line fashion.
pumping around petabytes of data that needs to be available very quickly is simply not going to work over the internet.
Using supercomputers to simulate how nuclear materials age negates arguments for the resumption of underground nuclear testing.
Reason enough to produce them.
> Now it's all up to you, you are free to use them or not
But the list of things you could not do would be extensive. Every modern thing has computers embedded at some point these days. Even reading a book (it was most likely produced with the help of computers, or gotten tracked down to the library it is stored at)
@ ENAy, I agree - I reckon Microsoft are being paid by computer chip manufacturers to make their applications UNnecessarily processor intnensive. One hand washes the other. Remember the Aimga 1200 with a huge 10MBytes of ram capable of running 1200 by 512 resolution applications with NO graphics cards. That was a super computer.
But, on the not so negative side progress is good...
I reckon Microsoft are being paid by computer chip manufacturers to make their applications UNnecessarily processor intnensive
As they say: "Never attribute to mallice that which can satisfactorily be explained by incompetence"
The windows API and codebase has been bogged down by two decades of backwards compatibility, new paradigms, and thousands of programmers, each with their own programming style.
I bet that there are a bazillion instances of "don't touch this or it would break, just slap another layer of crud on top" all across the microsoft product lines...
This thread makes me want to revisit building a relay computer...
Hehe, perhaps a massively multithreaded one:)
Andy
Nowadays with full sharing (Or whatever tech word it is), you do just as well with 2000 pc ans with a supercomputer.
Look at that wheather prediction thing that the MET is doing, whereby you leave you computer open to their network and it runs subproceses on your PC, thats a super super computer.
>Nowadays with full sharing (Or whatever tech word it is),
>you do just as well with 2000 pc ans with a supercomputer.
Not quite. Although some problems are suitable for a system with many nodes with large local memory and low speed IO, the majority of problems still need many nodes with small local memory but fast IO, which is where the supercomputer excells.
Andy
SuperComputers are needed for research. They have their limitations though; not capable of running microsoft programs.
>SuperComputers are needed for research. They have their
>limitations though; not capable of running microsoft
>programs.
Of course they can, through emulation...
They're perfectly suited for their specialty, and poorly suited for everything else, which, unfortunately includes running Windows.
Andy