Message boards :
Number crunching :
Nasa supercomputer..
Message board moderation
Author | Message |
---|---|
sqbell Send message Joined: 5 Apr 03 Posts: 4 Credit: 17,945 RAC: 0 |
Wonder how manu WU's this machine is able to crunch in one minute.. :-) http://news.bbc.co.uk/1/hi/technology/3532706.stm |
Toby Send message Joined: 26 Oct 00 Posts: 1005 Credit: 6,366,949 RAC: 0 |
> Wonder how manu WU's this machine is able to crunch in one minute.. :-) In one minute? none... but give it a couple of hours and it will turn in 10,240 :) mmmmmmmmmm And just think of all the star trek those 500 TB could hold! ------------------------------------------- - A member of The Knights Who Say NI! Possibly the best stats site in the universe: http://boinc-kwsn.no-ip.info |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
I am not a specialist in supercomputing but here what I think. If I am wrong I will soon be corrected... This arcticle does not specify how many TeraFlops this machine is capable of. I have made a web search but didn't find it. The most powerfull supercomputer currently in activity is the earth simulator located in Japan. It has a power of 35 TeraFlops (35 Trillions arithmetical operations per second). I am not sure about that but I think a SETI WU needs about 5 TeraFlops of calculation. In theory the earth simulator could process one SETI WU in 35/5 = 7 1/7 = O.142 seconds Like I said this is theory. In practice It wouldn't work that way because these supercomputers are connected into clusters. The way it works you have a kind of commander computer that split and distribute works to other computers (Nodes) and wait for results. Again I am not sure but I think SETI can't work using clusters computer. You could still use each nodes as a client and achieve spectacular performance. IBM is currently working on its new ASCI supercomputer and it will have 360 TeraFlops. The new NASA supercomputer could well be in that range so to make a long story short and finaly answer your question it would takes 360/5=72 1/72= 0.013 seconds 13 ms (in theory) Impressive nubers compare to our home toys... Regards Marc -.-. --.- -.. -..- . - .-.-. -.- --... ...-- |
sqbell Send message Joined: 5 Apr 03 Posts: 4 Credit: 17,945 RAC: 0 |
1 WU in 0.013 seconds.. so in one minute it will be able to compute 4615 work units.. nice.. :) |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> 1 WU in 0.013 seconds.. so in one minute it will be able to compute 4615 work > units.. nice.. :) > NO 360 / 5 = 72 Work units per second 72X60= 4320 WU per minutes 1/72 = 0.013888888888888 second to complete one WU EDIT it's still nice... |
Darren Send message Joined: 2 Jul 99 Posts: 259 Credit: 280,503 RAC: 0 |
> -.-. --.- -.. -..- . - .-.-. -.- --... ...-- I give up, what's the seventh character (.-.-.)? Now, as for the original question - it's a trick question, right? Obviously, the correct answer is 50 - 50 in a minute, 50 in an hour, 50 in a day - because as soon as it finishes those 50, Berkeley is going to tell it that it has exceeded the daily quota and not give it any more units anyway :o) |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> I give up, what's the seventh character (.-.-.)? LOL AR (end of message/transmition) Actually I will put it at the end... -.-. --.- -.. -..- . - .-.-. -.- --... ...-- |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
@Darren I have rearange it the way it would really be EDIT; OUPS ! |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
@Darren OK now but for some reason it doesn't take double space into account. Witch makes it harder to decode. Now its XXXXXXXXXXXX but should be XX XX X XX XX X X X -.-. --.- -.. -..- . - -.-. --.- -.. -..- . - .-.-. -.- |
Toby Send message Joined: 26 Oct 00 Posts: 1005 Credit: 6,366,949 RAC: 0 |
Well after I wrote my first post I did some searching. It seems an Itanium 2 can do a seti@home classic work unit (don't know about BOINC) in about 1.5 hours. If you ahve 10,240 I2s (each would have to be running its own copy of the client) then that would give you 10,240 work units every 90 minutes which workes out to 113.7wu/min. Definitely something I wouldn't mind having crunching for me although the power bill would be... uh... 'out of this world' :) Nitpickers: I know this doesn't take into account additional system overhead from clustering and ignores some other issues alltogether so just shut up, smile and nod. :p Well, I think we have proved that we are a bunch of nerds now. Time for bed! ------------------------------------------- - A member of The Knights Who Say NI! Possibly the best stats site in the universe: http://boinc-kwsn.no-ip.info |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> Well after I wrote my first post I did some searching. It seems an Itanium 2 > can do a seti@home classic work unit (don't know about BOINC) in about 1.5 > hours. lol Witch Itanium ? The one with 2 4 or 8 Mb of cache.... I've just signed a contract with Silicon Graphics, Inc to have one installed in my back yard. We should be fix in a couple a months -.-. --.- -.. -..- . - .-.-. -.- --... ...-- |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
@Darren again I've put it back the way it was cause it fits the size of my sig. |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
If each node asks for work units individually then each of the 10,000 nodes could do the 50 per day. One of the interesting notes I saw while mousing about is that there is concern with the state of supercomputers in that almost all of the latest type of super computer are collections of nodes. This includes the distributed type that we are participating in also. The field of vector processors has been neglected with the effort concentrated in the construction of parallel procssors. There are some problems that do not lend themselves to parallel processing and we are not making new vector processing machines. It is kind of interesting when you think about it. |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> There are some problems that do not lend themselves to parallel processing and > we are not making new vector processing machines. It is kind of interesting > when you think about it. Hi Paul It's the real beginning of a new era in supercomputing. The Virginia tech's "Big Mac" is a great example I guess. With Unix and the new High I/O architecture machines that are available to public today any one could built his own. With great knowledge and some money of course... -.-. --.- -.. -..- . - .-.-. -.- --... ...-- |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
Petit, > It's the real beginning of a new era in supercomputing. The Virginia > tech's "Big Mac" is a great example I guess. With Unix and the new > High I/O architecture machines that are available to public today any > one could built his own. With great knowledge and some money of course... In the "old" days we had Cray doing a new machine roughly each year. Even more interesting is the field in which Cray had more patents than any other ... I am not sure but NSF once it gets BOINC going well might not start to put more attention into the field of vector machines. |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> I am not sure but NSF once it gets BOINC going well might not start to put > more attention into the field of vector machines. I don't know as much as you do about supercomputing. Even It has always fascinated me I have never really got into all the techs involved in it. Well I know quite a few things about digits being an electronic engineer but It's the the big numbers and sci-fi that got me. I remember have seen a movie when I was a kid that I would really like to watch again. I don't remember the title but it's the story of a supercomputer located under a mountain in the US who is analysing the world geopolital situation and made to take control of the defence systems. Nobody knew it but it has a russian counterpart and they start communicating with each other. It starts to behave badly and its "creator" lost control of it. My favorite Hollywood supercomputer remain Hal 9000...War games was good too. Marc -.-. --.- -.. -..- . - .-.-. -.- --... ...-- |
Honie Send message Joined: 22 Jan 04 Posts: 141 Credit: 29,681,066 RAC: 0 |
Maybe you think of this film: COLOSSUS: THE FORBIN PROJECT (USA 1970) look at this link: http://www.amazon.com/exec/obidos/tg/detail/-/6300987531/103-2895952-8007843 |
Petit Soleil Send message Joined: 17 Feb 03 Posts: 1497 Credit: 70,934 RAC: 0 |
> Maybe you think of this film: > > COLOSSUS: THE FORBIN PROJECT (USA 1970) > Thanks Honie Thats the movie I was talking about. I remeber the name now. Regards Marc -.-. --.- -.. -..- . - .-.-. -.- --... ...-- |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
|
bellamyj_2003 Send message Joined: 15 Jun 03 Posts: 12 Credit: 225,195 RAC: 0 |
> I don't remember the > title but it's the story of a supercomputer located under a mountain in the US > who is analysing the world geopolital situation and made to take control of > the defence systems. I think it was T3:Rise of the Machines :) SDC Buggy |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.