Message boards :
Number crunching :
AstroPulse computing time
Message board moderation
Author | Message |
---|---|
Victorf Send message Joined: 20 Nov 99 Posts: 22 Credit: 747,484 RAC: 0 |
you need to match up simalar computers (speed). It takes about 3 day for my rig to do one astrpulse work unit @ 750 plus credits, paired up with a rig that takes 3 to 4 weeks. I stopped doing astropulse. |
Leaps-from-Shadows Send message Joined: 11 Aug 08 Posts: 323 Credit: 259,220 RAC: 0 |
You stopped doing Astropulse because you won't get credit right away? Awww, poor baby. My Windows machine Cruiser does Astropulse units in a little over a day (~29 hours) and I still do them. I have waited nearly two months for credit on a couple of them. Then again, I'm not a credit-monger. I figure that Astropulse might actually help find an alien signal some day. Cruiser Gateway GT5692 L-f-S Edition -Phenom X4 9650 CPU -4GB 667MHz DDR2 RAM -500GB SATA HD -Vista x64 SP1 -BOINC 6.2.19 32-bit client -SSE3 optimized 32-bit apps |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
you need to match up simalar computers (speed). My rig knocks out an AP WU in about 12.2 hours. Is it fair that I should have to wait for you? If you look at S@H as a search for interesting astronomical phenomena and/or a gigantic experiment in Distributed Computing then the credits can take care of themselves. F. |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
you need to match up simalar computers (speed). Can't tell if you use them or not since your pc's are hidden but try using the optimized apps for AP if you're not already. They make a significant difference in the time it takes to complete them. If your system qualifies it's worth looking into. |
Victorf Send message Joined: 20 Nov 99 Posts: 22 Credit: 747,484 RAC: 0 |
my rig AMD 6400+ Black Box Edition, zalman 9700 HS (can be upgraded to quad core) ASUS Crosshair MB 8 Gig's ( 2 gig each) xms2 dhx matched pairs 800 Corsair mem 2 Seagate sata's 320 & 500 Gig HP 20x DVD burner nv 8800gts512oc ddr3 vid Sound Blaster X-Fi XtremeGamer Fatal1ty Professional Series sound card antec 650 watt ps antec 900 case, 4 fans (can add 2 more) vista ulta 64 you need to match up simalar computers (speed). It takes about 3 day for my rig to do one astrpulse work unit @ 750 plus credits, paired up with a rig that takes 3 to 4 weeks. I stopped doing astropulse. |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
my rig About 10 hours on the Frozen Penny......but I quit doing them because the credits are not up to par with opti MB........ "Freedom is just Chaos, with better lighting." Alan Dean Foster |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
my rig It is a known issue that AstroPulse takes longer on AMD CPUs than Intel's. Personally, I'm not a big fan of matching up similar speed computers just so people can have their credit faster. |
Aurora Borealis Send message Joined: 14 Jan 01 Posts: 3075 Credit: 5,631,463 RAC: 0 |
I agree. Besides, it would add a bunch more overhead on the database which is already overloaded with needed queeries. Why add even more dependencies on the servers that are already constantly running into bottlenecks. Boinc V7.2.42 Win7 i5 3.33G 4GB, GTX470 |
Leaps-from-Shadows Send message Joined: 11 Aug 08 Posts: 323 Credit: 259,220 RAC: 0 |
you need to match up simalar computers (speed). They will never speed-match crunchers like you want. It adds too much overhead to the process, and their servers are hard-pressed as it is. You do realize that some machines are really slow at regular Multibeam units too, right? Does that mean that you'll quit doing those too? You won't get that credit right away either... If you really want to have someone add this to the features, post something in the Questions and Answers - Wish List section. It's not likely to be added, but it is where this topic should be placed. Cruiser Gateway GT5692 L-f-S Edition -Phenom X4 9650 CPU -4GB 667MHz DDR2 RAM -500GB SATA HD -Vista x64 SP1 -BOINC 6.2.19 32-bit client -SSE3 optimized 32-bit apps |
Zap de Ridder Send message Joined: 9 Jan 00 Posts: 227 Credit: 1,468,844 RAC: 1 |
From the 14th of october I stress tested my computer for 2 weeks 24/7 100% cpu usage. It got me near to 2100 rac with MB only.Nice to see . I'm throteling back to 5/24 50% and I will only do Astropulse for the next month at least because the science looks more interesting. Credits are just a sideshow for me. The first WU I will do full throtle cause the wingman of the guy wich already completed it let him down so he'll get his credits by tomorrow.:-) Looks like it gonna take another 12 hours wich make a total of 18.5 estimated. |
Virtual Boss* Send message Joined: 4 May 08 Posts: 417 Credit: 6,440,287 RAC: 0 |
you need to match up simalar computers (speed). Yeah - Right? And then your fast wingman has a ten day cache, and you still have to wait longer than a slower machine that does it straight away and returns a result in 3-5 days. Its all luck of the draw. Have patience. |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
Certainly crunch speed is the wrong measure to be used in any discussion of credit delays. The host entries in the database have an average turnaround time for each host which would be the sensible way to match hosts, and of course the Scheduler already reads the host entry for each host which makes a request so matching need not increase database load. Joe |
Mumps [MM] Send message Joined: 11 Feb 08 Posts: 4454 Credit: 100,893,853 RAC: 30 |
Certainly crunch speed is the wrong measure to be used in any discussion of credit delays. The host entries in the database have an average turnaround time for each host which would be the sensible way to match hosts, and of course the Scheduler already reads the host entry for each host which makes a request so matching need not increase database load.Joe But wouldn't it increase the load if it had to keep looking for a host that fell within the window of "average turn around time?" So now, I request a WU, and it needs to find a WU that was doled out to another host with roughly the same turn around time, rather than simply giving me what's next in line... |
BigBLueGecko Send message Joined: 3 Nov 08 Posts: 24 Credit: 5,029 RAC: 0 |
I can see why people go for credits, but personally I dont care that much for credits, its cool for seeing what progress you are making but I dont specificaly go for the credits. I just enjoy helping out SETI and watching the data <Yeah I know Geeky>. But I think it is pretty cool. P.S. on my lesser computer, I got a huge spike past the top of my view in the SETI window across the frequency bar, it lasted for about 2 time slots. Is that a glitch or something on the data. This was the seti@home and not the astro pulse. |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
Certainly crunch speed is the wrong measure to be used in any discussion of credit delays. The host entries in the database have an average turnaround time for each host which would be the sensible way to match hosts, and of course the Scheduler already reads the host entry for each host which makes a request so matching need not increase database load.But wouldn't it increase the load if it had to keep looking for a host that fell within the window of "average turn around time?" So now, I request a WU, and it needs to find a WU that was doled out to another host with roughly the same turn around time, rather than simply giving me what's next in line...Joe The feeder puts a limited number of tasks in a shared memory buffer, the Scheduler assigns them from there as hosts request work. So all that's needed is separate buffers for hosts in different turnaround time classes. The Homogeneous Redundancy feature which some projects use is set up that way, and new criteria for classifying hosts can be added. IOW, most of the needed capability already exists in BOINC. OTOH, when Matt made a recent change to the feeder settings to smooth out the bursts of AP work which were causing difficulties, he mentioned some concern about the shared memory buffers. That suggests the scheme might not be practical for this project. Joe |
The Clue isin front of you Send message Joined: 21 Mar 05 Posts: 4 Credit: 2,848,466 RAC: 1 |
I was just wondering what is Astropulse? It seem to take a very long time to process it. I'm using P4 3.6 with HT and run it 24/7 at 100%. It has been running now for over 3 days and its only at 34% is that correct? How big is this work unit. I'm not worrying about the credit but I want to know is there something wrong or is it a bad computation, is that why its taking so long to complete? Thanks Clue :-) |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I was just wondering what is Astropulse? It seem to take a very long time to process it. I'm using P4 3.6 with HT and run it 24/7 at 100%. It has been running now for over 3 days and its only at 34% is that correct? How big is this work unit. I'm not worrying about the credit but I want to know is there something wrong or is it a bad computation, is that why its taking so long to complete? Thanks Clue :-) Hello The Clue is in front of you. The clue in front of you is: Astropulse FAQ. |
tullio Send message Joined: 9 Apr 04 Posts: 8797 Credit: 2,930,782 RAC: 1 |
I was just wondering what is Astropulse? It seem to take a very long time to process it. I'm using P4 3.6 with HT and run it 24/7 at 100%. It has been running now for over 3 days and its only at 34% is that correct? How big is this work unit. I'm not worrying about the credit but I want to know is there something wrong or is it a bad computation, is that why its taking so long to complete? Thanks Clue :-) On my AMD Opteron 1210 at 1.8 GHz running Linux the standard AP takes 115 hours. The optimized version takes 55 hours, but AP is very AMD-unfriendly. Tullio |
The Clue isin front of you Send message Joined: 21 Mar 05 Posts: 4 Credit: 2,848,466 RAC: 1 |
Thanks for the help. :-) |
I C U Send message Joined: 4 Apr 07 Posts: 5 Credit: 277,443 RAC: 0 |
Went to google to see if there was anything concerning AstroPulse AMD and increased time (which led me to this thread). :-/ About a month ago, had an astropulse on my machine, interesting to see one and for curiousities sake, watched as time progressed to see what happened. It originally began at some smaller value like 100 hours and slowly increased until 200+ when done. The end result turned-out to be miscalculated waste of time. :-/ Today, I happened to see my machine processing another Astropulse (I upgraded from boinc 6.2 to 6.4 a few days ago), and I again see that this Astropulse is also slowly increasing in time too. The left side now at 139hours processed while the right side shows 90 hours still to do, but the right side is increasing (time to go). Is that normal behaviour for time to increase? This is for an AMD Sempron machine if that info helps. Stats for this machine are here if that helps: www.joescat.com/boinc/ |
©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.