Message boards :
Number crunching :
Congratulations. We're the 171st fastest supercomputer
Message board moderation
Author | Message |
---|---|
tbret Send message Joined: 28 May 99 Posts: 3380 Credit: 296,162,071 RAC: 40 |
I have watched Dr. Korpela's recent presentation at the EBI 2014 conference and in it he says the estimate of the real computing power of SETI@Home is about like a 237 TFLOP computer. https://connect.arc.nasa.gov/p4nb5vxclik/?launcher=false&fcsContent=true&pbMode=normal According to the November 2013 list of the 500 most powerful supercomputers on the planet, that makes our collective efforts about #171. We would be just behind the "Lindgren" Cray XE6 which, when it was new, was #31 in the top 500 supercomputers on the planet back in June of 2011. Reference: https://www.pdc.kth.se/resources/computers/lindgren/hardware Of course, I have no idea how Dr. Korpela knows this, I'm just repeating what he said. Compare that to the 15.7 Tflops he estimated in October of 2000, and you can see that we've improved a lot. Still... #171... I guess it's time to upgrade some computers. reference: http://penguin.ewu.edu/cscd543/Fall-2000/Seti@Home.html We'd better get crunching! |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
WoW, cool! But only 171? not the first? ;) And 237TFLOPS not looking so unreachable actually. Am I wrong but estimation for modern GPUs are in 1-2TFLOPs range? That is, only ~100-200 modern GPUs on the project??? From such approach 237TFLOPS looks like little low... SETI apps news We're not gonna fight them. We're gonna transcend them. |
tbret Send message Joined: 28 May 99 Posts: 3380 Credit: 296,162,071 RAC: 40 |
From such approach 237TFLOPS looks like little low... Perhaps he's estimating using Recent Average Credit? That would look a little low, I would think. |
yo2013 Send message Joined: 9 Mar 14 Posts: 173 Credit: 50,837 RAC: 0 |
According to BAM!, we have 676 TFLOPS. The top BOINC project is DistrRTgen, with 1,607 TFLOPS. The top supercomputer is Tianhe-2, with 33,863 TFLOPS. And the top computer grid is Folding@Home, with 40,323 TFLOPS. |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
Last stat at E@H shows: 1161.5 TFLOPS and we are far beyound that point here on SETI, or i miss something or the numbers are totaly wrong. Maybe someone could explain that better. That could be the side effect of miscalculations due the bad estimates in creditscrew, who knows? |
Thomas Send message Joined: 9 Dec 11 Posts: 1499 Credit: 1,345,576 RAC: 0 |
Thanks tbret ! :) Who will find the right TFLOPS ? :p |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
From such approach 237TFLOPS looks like little low... ROFL, indeed :DDD SETI apps news We're not gonna fight them. We're gonna transcend them. |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
I got tflops to spare. LOL. Wonder what the little kitty farm in Wisconsin is doing in terms of flops. The kitties flop on the waterbed all the time. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
According to BAM!, we have 676 TFLOPS. The top BOINC project is DistrRTgen, with 1,607 TFLOPS. The top supercomputer is Tianhe-2, with 33,863 TFLOPS. And the top computer grid is Folding@Home, with 40,323 TFLOPS. Folding at home... gr... they still in front of us :) I participated in Folding@home once when I bought my first Ati HD4870.. that time no BOINC project supported ATi. Then MilkyWay added ATi support (via 3rd party app initially) and then I thought, why SETI behind? :) ... Interesting, what allows them (Folding@home) accumulate so bigger power? more than x40 than us. Some discrete support from Univercities? one or 2 supercomputers behind them as behind Einstein@home? Or what? ... SETI apps news We're not gonna fight them. We're gonna transcend them. |
Thomas Send message Joined: 9 Dec 11 Posts: 1499 Credit: 1,345,576 RAC: 0 |
I got tflops to spare. TFLOPS to share, Mark ? BRIGADE DU COSMOS is agree :p |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
According to BAM!, we have 676 TFLOPS. The top BOINC project is DistrRTgen, with 1,607 TFLOPS. The top supercomputer is Tianhe-2, with 33,863 TFLOPS. And the top computer grid is Folding@Home, with 40,323 TFLOPS. It may just be the way it is calculated and not really the true picture. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
According to BAM!, we have 676 TFLOPS. The top BOINC project is DistrRTgen, with 1,607 TFLOPS. The top supercomputer is Tianhe-2, with 33,863 TFLOPS. And the top computer grid is Folding@Home, with 40,323 TFLOPS. The BOINCstats (BAM) figure is most certainly reverse-calculated from credits - specifically, the project-supplied total RAC. A RAC of 200 is assumed to have been generated by 1 GigaFLOP of cunching power (put the two figures into a calculator). If we look in the top500 list (current edition - November 2013 figures), we would be at about #50 on BAM's figure, displacing Edinburgh's Cray. If CreditNew was fixed, and RAC was for the sake of argument doubled, we would be at #21, displacing the Chinese National Supercomputing Centre. Build a new supercomputing centre by fixing BOINC's credit system? Er, no - I don't think so. If you listen to Eric's talk (link in first post - at about 8 minutes in), he is very careful to describe his 237 TFLOP claim as "actual" computing power, and to distinguish it from ... theoretical computing power, which is what a lot of people report in their distributed projects. Given the nature of the talk he was giving, on the record, and the sort of people he would expect to be in the audience, I suspect that those were very, very, carefully chosen words. |
yo2013 Send message Joined: 9 Mar 14 Posts: 173 Credit: 50,837 RAC: 0 |
The BOINCstats (BAM) figure is most certainly reverse-calculated from credits - specifically, the project-supplied total RAC. A RAC of 200 is assumed to have been generated by 1 GigaFLOP of cunching power (put the two figures into a calculator). Yes, it's calculated that way. |
Thomas Send message Joined: 9 Dec 11 Posts: 1499 Credit: 1,345,576 RAC: 0 |
According to BAM!, we have 676 TFLOPS. The top BOINC project is DistrRTgen, with 1,607 TFLOPS. The top supercomputer is Tianhe-2, with 33,863 TFLOPS. And the top computer grid is Folding@Home, with 40,323 TFLOPS. Thanks Richard |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
The BOINCstats (BAM) figure is most certainly reverse-calculated from credits - specifically, the project-supplied total RAC. A RAC of 200 is assumed to have been generated by 1 GigaFLOP of cunching power (put the two figures into a calculator). If this is right then we could see "hipercomputers" in GPUGrid... you see there host with >2MM RAC - 10000GFlops??? Insane. Instead on SETI our actual RAC is about 1/2 to 1/4 of the real one... Another good reason to fix creditscrew!. BTW FYI one of this GPUGrid top host who produces there >2MM produces about 130K of RAC here in SETI, that i could said with confidence, i know the host and his owner. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
That made me think of another little bit of research I could do. Here is a table of the highest individual computer RAC at each of the main BOINC GPU projects. Project Highest RAC Collatz Conjecture 5,675,105.68 DistrRTgen 4,818,867.42 GPUGrid.net 2,327,376.94 PrimeGrid 2,147,330.57 Moo! Wrapper 967,421.13 Milkyway@home 638,370.73 Einstein@home 273,284.26 SETI@home 205,836.67 Just a little snapshot this afternoon. They have little in common (except all being based on an i7 CPU, with multiple GPUs). But I think it's safe to assume that each one is its owner's pride and joy, the best they can afford to buy, and fine-tuned within an inch of its life. You don't get to be Top Dog without having that much in common. Now, is anybody going to tell me that the top computer builder, or the top programmer, at Collatz or DistrRTgen can get 27 or 23 times more GFlops out of the same consumer-grade silicon, as their opposite numbers here at SETI? I think not. Out of that list, I would trust Einstein - which issues fixed credits, not using CreditNew - to have made an honest attempt to verify their credit awards against actual scientific GFlops. I would also trust Eric to have a scientific justification for the figure he quoted in that talk last night of 237 actual TFlops, though the time available in that slot was far too short for explanations of that level of detail. The rest, frankly, I consider to be works of fiction, dreamed up by the respective project administrators - those are perhaps the "theoretical GFlops" that Eric also mentioned. All of which means that I also consider the claim of 8.398 PetaFLOPS currently displayed on the BOINC home page for the total power of all BOINC projects to be deeply suspect, and likewise the individual GFlops values displayed on Top 100 multi-project BOINC participants. |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
It's all relative, Richard. What one project considers a 'flop', I think. Seti's work requires a most active system. Even the GPUs need CPU support. RAM access seems to be particularly important, although I have never quite nailed down just why. The way the program runs just does not seem to support that theory. But it does work that way. Meow. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
SETI@home 205,836.67 That is the highest RAC I have seen on a single host. I too now have the highest RAC I have ever had. It was the recent AP feast that did it. |
ExchangeMan Send message Joined: 9 Jan 00 Posts: 115 Credit: 157,719,104 RAC: 0 |
SETI@home 205,836.67 I have that machine. When V6 multibeam was running, I topped out on this machine with the same configuration at 270,000. It would be nice if I could get a month's supply of AP work to see where the top really is. With V7, the number is meaningless. My present RAC on this machine has just topped 207K and still going up. Next time the top host display updates, it will be higher. |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
I know I have been trying to catch you all winter. How do you run 5 GPU cards on one machine? That is your secret is it not? |
©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.