Message boards : Number crunching : Petaflop record - combined Boinc push
Author | Message |
---|---|
Russell Rukin Send message Joined: 22 Jun 06 Posts: 5 Credit: 1,417 RAC: 0 |
Sorry if this subject has been brought up before (I did a search but couldn't find anything). Wiki (http://en.wikipedia.org/wiki/BOINC) lists that Boinc as a whole has reached an average of 418 TFLOPS as of as of April 13th 2006. My question is, does the Boinc community ever try for a crunching day (24hrs) when everyone trys to reconfigure all available computer resources for maximum output over a day? Would petaflop performance be completely unrealistic even with such a push on that day? In my example my new main computer is only recently connected. The rosetta program runs when the screen saver is on and the dual processor is set to 1 core only. I would certainly be interested if it was merely for 24hrs to break a record and create a news story by letting the computer use full resources all day and might even hook up my old 2.4Ghz 1GB Athlon to help out. From news reports I've seen the petaflop record is still 2 to 4 years away from being broken by a single supercomputer. I'm sure it would make news items on the net and maybe in traditional media if it could be broken by distributed network. If a date could be fixed for a day in a month or couple of months then people would have time to plan, get the word round and hopefully encourage new crunchers to get involved to break the petaflop barrier? |
Feet1st Send message Joined: 30 Dec 05 Posts: 1755 Credit: 4,690,520 RAC: 0 |
I can appreciate your enthusiasm... and also the idea to promote BOINC projects... but it just doesn't work that way. The FLOPS are recorded when WUs are reported back. You could save up a week of work and report it all back on one day (the competitive teams do that all the time). But for BOINC as a whole it would not really demonstrait the PetaFLOP you intend. Also, it's very difficult to get users to change their operations, even if only for a day. So your participation rate is such an endeavor would be disappointing. What might be interesting though is to monitor the theorectical power of the projects, and see when that number reaches the petaFLOP. You can monitor all the numbers here on BOINCstats. Add this signature to your EMail: Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might! https://boinc.bakerlab.org/rosetta/ |
dcdc Send message Joined: 3 Nov 05 Posts: 1831 Credit: 119,627,225 RAC: 11,586 |
and most of the people that would participate are already maxing their resources as it is. Worth pushing for permanently though ;) |
jjgb10 Send message Joined: 29 Sep 05 Posts: 21 Credit: 6,152,959 RAC: 0 |
Russell, I think you should run Rosetta all the time on your computer. I run it on my Pentium T2500 dual core laptop 24/7 and it doesn't slow it down a bit. Every program still opens as if I'm not using the processor. I also run it 24/7 on an AMD X2 4200+ dual core processor and again, it doesn't even phase the computer. You aren't goint to lose much by running it 24/7. Your power bill may be a little more but it is for a good cause. |
Russell Rukin Send message Joined: 22 Jun 06 Posts: 5 Credit: 1,417 RAC: 0 |
Russell, I think you should run Rosetta all the time on your computer. I run it on my Pentium T2500 dual core laptop 24/7 and it doesn't slow it down a bit. Every program still opens as if I'm not using the processor. The power bill does bother me, I'm a poor artist. I don't keep my computers on 24/7 for a start. I hope to make more cash this year, if I do I'll up the CPU time. As for not slowing down processes that might be so for regular use but I'm a 3D renderer, power user and my comp uses its dual 3Ghz core and 4GB of ram to the max when I'm in session. I do understand how important this work is though. For me its not just the goal of the individual project but the collaborative distributed net resource itself, I'd like to see it grow and grow, it has such potential. |
Russell Rukin Send message Joined: 22 Jun 06 Posts: 5 Credit: 1,417 RAC: 0 |
I can appreciate your enthusiasm... and also the idea to promote BOINC projects... but it just doesn't work that way. The FLOPS are recorded when WUs are reported back. You could save up a week of work and report it all back on one day (the competitive teams do that all the time). But for BOINC as a whole it would not really demonstrait the PetaFLOP you intend. I have to admit I'm not sure what you really mean? the 400+ TFlops a cumulative figure over ALL the years Bonic has been going? If thats not the case and its merely a case of collecting data over a span of time (a week/2 weeks) then so be it, have the max out time to over a week/2 weeks. I understand dcdc's point that most crunchers are maxed out all the time but is that "most" the core users, what are the statistics of the people that only do light work? I have no knowledge of your figures so they could be in a dramatic minority but if not they could significantly move figures for a short period! |
Feet1st Send message Joined: 30 Dec 05 Posts: 1755 Credit: 4,690,520 RAC: 0 |
Ah ha, so that's where you were coming from. I see now. FLOPS is floating point operations per second. So... per second. It is a measure of the speed of your computing. Like your PC having 2 processors each running at 3Ghz. The Ghz is a measure of speed... not floating point calculations exactly, but speed none the less. So, when they talk about collective BOINC TeraFLOPS, they are talking about a sustained SPEED of computing... not the total of all the work done. So to truely reach a petaflop, you would have to start at a given time and demonstrait you've crunched (collectively) a thousand trillion floating point caculations, PER SECOND, throughout some period of time after that point. In the case of BOINC, it is basically a measure saying that if you wanted to crunch more than BOINC, you'd need a supercomputer of THIS size to match us. Since BOINC isn't truely one computer, and isn't actually running all the time, it's an approximation of the crunching power of all the participating machines. Add this signature to your EMail: Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might! https://boinc.bakerlab.org/rosetta/ |
Tenemag Send message Joined: 3 Jan 06 Posts: 1 Credit: 144,003 RAC: 0 |
i'm all for this... I'll run my server, my Dual-Core, my laptop, my 2 Pentium 3's, my P4HT and everything else I have all day and all night with full resources... i'll even see If i could get them on some of my friends machines... to break a world record would be awesome! |
Message boards :
Number crunching :
Petaflop record - combined Boinc push
©2024 University of Washington
https://www.bakerlab.org