Message boards :
Number crunching :
Astropulse FAQ
Message board moderation
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 14 · Next
Author | Message |
---|---|
![]() Send message Joined: 9 Apr 04 Posts: 8797 Credit: 2,930,782 RAC: 1 ![]() |
Astropulse 5.0 optimized SSE3 for Linux has run in 47,787 hours and given me 749.76 credits, that is 15.69 credits/hour. But I don't care much about credits and the time taken is about one half compared to the standard app. My CPU is an Opteron 1210 at 1.8 GHz running SuSE Linux 10.3, BOINC 5.10.45. No CUDA, thanks God. Tullio |
Gone4good Send message Joined: 21 Dec 08 Posts: 20 Credit: 24,062 RAC: 0 ![]() |
Astropulse 5.0 optimized SSE3 for Linux has run in 47,787 hours and given me 749.76 credits, that is 15.69 credits/hour. But I don't care much about credits and the time taken is about one half compared to the standard app. My CPU is an Opteron 1210 at 1.8 GHz running SuSE Linux 10.3, BOINC 5.10.45. No CUDA, thanks God. Part of the reason I ask over credits is because I just built this new rig from scratch and was going off the normal enhanced units as a kind of reference over the 24 hr period then I started getting Astropulses which kind of changed how I was seeing the actual work done. I didnt want to use a non standard app for now and crunch with the one originally downloaded so I can sort of use it to benchmark and also be of help finding something intelligent I hope before I die.I had a lot of heat issues but want the cpu to be happy at 100% so I installed a water system over the last 2 days to see how effective to system temps it was.Im seeing a drop of approx 20deg off the chipset now and cpu at 34deg when running seti at 100% as opposed to normally(pre water)round 55deg for chipset and cpu roughly 45 - 50.Even though I havent put the cpu on water the overall drop is effecting it so I know for sure I had heat bleeding off from the boards toward the cpu base.Part of this is the graphics cards so I am looking to build an addon module that will take a little from them as well since they run stupidly hot(4800 series ati guys know what I mean)and in use like a game you can physically get burnt off them even though ATI say this is normal.I disagree on this amount of heat since it rises and was overheating the northbridge. Just as a note my old pc would do 2 normal units a day and an Astropulse you could likely walk to the Star its from before it could complete one.Im hoping to use the ATI cards to crunch before long but thats if I can actually make a unit work on its stream input but im not the hottest at that kind of coding so may be a while for me yet. As for units themselves it doesnt matter how many you do as long as you take part is what matters the most I think,and im sure my sytem is a snail still compared to some that may be in use on seti.;) Night |
![]() ![]() Send message Joined: 30 Jul 00 Posts: 1887 Credit: 7,441,278 RAC: 49 ![]() ![]() |
With the Astropulse graphics (compared to the SETI@home graphics), are they supposed to not have text in them? The progress bar shows and the box below it (I don't remember what it's supposed to be called) are there, but no text anywhere, except for the "Astropulse" logo at the bottom. Is this just for the 1st time it's "crunched", or is it every time? Secondary Q: Is there a (non-technical) reason why AP takes (a lot) longer to crunch than S@h? Is it just a larger (overall) file size than S@h, or what? |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 ![]() |
With the Astropulse graphics (compared to the SETI@home graphics), are they supposed to not have text in them? The progress bar shows and the box below it (I don't remember what it's supposed to be called) are there, but no text anywhere, except for the "Astropulse" logo at the bottom. Assuming you're using stock Astropulse, it should have some text but there are blank areas. You may have to wait for awhile after bringing up the graphics display before it is filled in. The program doesn't keep generating information for display while the display is off. Secondary Q: Is there a (non-technical) reason why AP takes (a lot) longer to crunch than S@h? Is it just a larger (overall) file size than S@h, or what? An Astropulse WU has 32 times as much data as a SETI@home Enhanced WU. The program is also looking for a different kind of signal so does different operations on the data. Joe |
![]() ![]() Send message Joined: 30 Jul 00 Posts: 1887 Credit: 7,441,278 RAC: 49 ![]() ![]() |
With the Astropulse graphics (compared to the SETI@home graphics), are they supposed to not have text in them? The progress bar shows and the box below it (I don't remember what it's supposed to be called) are there, but no text anywhere, except for the "Astropulse" logo at the bottom. |
Graham H Send message Joined: 28 May 99 Posts: 21 Credit: 84,774 RAC: 0 ![]() |
There seems to be much discussion in this thread about AP not completing in time on "somebody's" PC because of X or Y reasons and them not getting the credit. Please spare a thought for those that have completed an AP unit but so far no-one else has managed to complete the same one...... It was sent to me on 30 October 2008 and I completed on 28 November. (I only use my PC in the evenings for about 6 hours or less, so it was a tight squeeze even for me) The person who was sent the unit at the same time as me failed to complete in time and so it was sent to someone else. They failed to complete on time also, so it was sent to someone else. They have until 28 January 2009 to complete. Hopefully they will, or I will have to wait at least another few days, weeks, or even MONTHS or YEARS before I get the credit. Can it not be done that AP units are only sent to PC's that stand a far more resonable chance of completing, rather than leaving it to the user to decide that his PC is not capable of completing one on time and having to "delete AP" from the preferences? |
Rob Lilley Send message Joined: 20 May 99 Posts: 6 Credit: 514,059 RAC: 1 ![]() |
There seems to be much discussion in this thread about AP not completing in time on "somebody's" PC because of X or Y reasons and them not getting the credit. IMHO, it would actually help if the deadlines for AP gave more time for completion. For normal SETI units, the deadline is in proportion to the estimated crunching time, but AP deadlines don't seem to follow this rule. This means that some machines that are perfectly capable of finishing a SETI unit on time haven't a hope of finishing an AP unit before the deadline. If the deadlines for AP units were extended, then a WU might be completed within (say) 6 weeks, rather than failing to complete in a month, being sent to another machine, and possibly once again failing to complete after another month has gone by. (Hope this makes sense) ATM it seems that quite a few people are getting caught out not so much by the size of the AP WUs as by the fact that they need far more hours of crunching per day with the existing deadlines. ![]() |
Jim Marconnet ![]() Send message Joined: 7 Feb 08 Posts: 1 Credit: 90,636 RAC: 0 ![]() |
I say just vote against AP with your feet! I just got a whole bunch of AP jobs that say will take 934 hours each to complete. That's 39 days each! Simply ridiculous, IMHO to make work units so large! I'm veering away from AP just as fast as I can figure out how to do it. Jim |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 ![]() |
I say just vote against AP with your feet! Don't trust Boinc's predictions, if you take your last short AP WU and double it, Expect it to take 111 hours max (4½ days) or a lot quicker if you optimise it. Claggy |
Richard Haselgrove ![]() Send message Joined: 4 Jul 99 Posts: 14687 Credit: 200,643,578 RAC: 874 ![]() ![]() |
I say just vote against AP with your feet! No. Looking at the speed of the last Astropulse task you computed on your AMD host, they will actually be complete after about 4.3 days - and may even exit earlier than that (as the last one did). That 39 days is a mis-estimate by your BOINC client. We can explore that later if you want. But 4.3 days is still a long time, so you may want to visit your SETI@home preferences page, and de-select Astropulse tasks. |
Graham H Send message Joined: 28 May 99 Posts: 21 Credit: 84,774 RAC: 0 ![]() |
I agree with Rob Lilley's point about extending the deadline and about Astropulse not seeming to use the same goalposts....... A normal SETI unit that Boinc says will complete in about 9 hours I can normally do in less than 6. An Astropulse that it says I can complete in 89 hours before it starts gradually creeps up to over 120 hours by the time it gets close to finishing. As I said before, I only use my PC for about 6 hours a day during the week (it's switched off overnight and whilst I'm at work) with about 12-15 hours power on Saturdays and Sundays. This gives about 54 hours "on time" for a week and doesn't allow for actual "usage time", i.e. when I'm actually doing something else on the PC and Boinc is in "cycle steal" mode and only running on CPU cycles snatched between the other programs usage. Even if I left my PC idle all the time it's powered on, it would still take me at least 2 weeks to complete an AP unit, whereas it is taking about 3.5 weeks with my "normal" usage. |
archae86 Send message Joined: 31 Aug 99 Posts: 909 Credit: 1,582,816 RAC: 0 ![]() |
Even if I left my PC idle all the time it's powered on, it would still take me at least 2 weeks to complete an AP unit, whereas it is taking about 3.5 weeks with my "normal" usage.That seems an excellent reason to click the box to exclude future Astropulse downloads for that host. SETI used to have some extraordinarily long deadlines for part of the work distributions. That has a set of problems all its own. |
![]() Send message Joined: 9 Apr 04 Posts: 8797 Credit: 2,930,782 RAC: 1 ![]() |
My latest astropulse run with the 5.00 SSE3 optimized Linux app ended in 44.515 hours instead of the 47.787 of the preceding optimized app and gave me 778,11 credits instead of the former 749.76. I noticed the 59.36 radar blanking percentage. All this on my AMD Opteron 1210 at 1.8 GHz running SuSE Linux 10.3. Tullio |
Stacey Baird Send message Joined: 8 Nov 04 Posts: 41 Credit: 407,924 RAC: 0 ![]() |
My Astropulse assignment was over 447 hours to be completed by 2/24/09, I believe. I just aborted it because there was no way I could finish it within the time limit. I have also had at least five screenfuls of Seti enhanced, Docking, and cuda come down in the past two days along with huge batches of Rosetta. I am running boinc 6.4.5 and boinc manager 6.6.2. Should I upgrade to Manager 6.4.5? The welfare of the people is the highest law - Cicero If no one complains, the people must be satisfied. |
Stacey Baird Send message Joined: 8 Nov 04 Posts: 41 Credit: 407,924 RAC: 0 ![]() |
My Astropulse assignment was over 447 hours to be completed by 2/24/09, I believe. I just aborted it because there was no way I could finish it within the time limit. I have also had at least five screenfuls of Seti enhanced, Docking, and cuda come down in the past two days along with huge batches of Rosetta. Answered my own question by installing Boinc Manager 6.6.4. Things are going much smoother and faster - except for the massive download of work from Rosetta, Seti and Docking which may or may not ever finish by the allotted time which I received before upgrading to 6.6.4. But, we'll just keep-on crunchin' here on the sunny island paradise in the Sulu Sea. The welfare of the people is the highest law - Cicero If no one complains, the people must be satisfied. |
![]() Send message Joined: 27 Feb 08 Posts: 20 Credit: 265,401 RAC: 0 |
Is there an optimised AP app anywhere for 64-bit Windows (XP64) running on an early Athlon 64 (SSE2 only)? The current opt app ap_5.00r103_SSE.exe is compiled with the Intel compiler, which is known to be biased toward Intel compilers (ie if it's not an Intel CPU it defaults to unoptimised code). |
![]() ![]() Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 ![]() |
Incorrect, It is a Microsoft Visual Studio build, and No, Intel compiler doesn't do that unless you use certain dynamic dispatch features, which we don't. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
gejo Send message Joined: 28 May 99 Posts: 8 Credit: 1,606,948 RAC: 0 ![]() |
Matt wrote: Looks like "Astropulse V5" was finally released yesterday night. Question: Can I still use the opt. AP V5 r103 version or MUST I use the stock AP version until there is a new optimized AP version available? |
Richard Haselgrove ![]() Send message Joined: 4 Jul 99 Posts: 14687 Credit: 200,643,578 RAC: 874 ![]() ![]() |
Matt wrote: Looks like "Astropulse V5" was finally released yesterday night. I'm afraid you must NOT use the r103 version for the new work. You can leave r103 active with the existing app_info.xml file if you want, and you will get an occasional resend from the unfinished v5.00 work: or you can switch to the stock v5.03 application. Testing at Lunatics is proving frustratingly slow, because we have to compete for 'real life' WUs with everybody else: and it's proving hard to hit the scheduler at just the right moment. But we'll keep trying for you. |
gejo Send message Joined: 28 May 99 Posts: 8 Credit: 1,606,948 RAC: 0 ![]() |
Richard, thanks for the quick and clear answer. |
©2025 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.