Message boards :
Number crunching :
Astropulse
Message board moderation
Author | Message |
---|---|
DaveLee Send message Joined: 9 Jan 01 Posts: 7 Credit: 3,894,728 RAC: 0 |
This may be a stupid question, but why am I getting only astropulse anymore? |
Odan Send message Joined: 8 May 03 Posts: 91 Credit: 15,331,177 RAC: 0 |
This may be a stupid question, but why am I getting only astropulse anymore? Not a stupid question :) I see you are running stock applications so I presume you do not have an app_info.xml? If you don't understand the question the answer is almost certainly "no". Excuse me if it sounds like I am treating you like an idiot I really am not. Best place to look is in "your account" and look at the preferences. I think you will have been in there to select AP V5. Make sure you still have got SETI@Home Enhanced selected. Otherwise it is probably just the scheduler playing one of its usual games. I have to turn AP on & off in my preferences to keep a flow of both AP & MB comming. There is no harm in just crunching AP if you are patient enough to wait for a wingman to complete. This can take a while :) |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
I know you are running stock applications, but that is interesting, as over in this thread, those of us using an app_info.xml file for optimized applications are noticing the same thing. The trick we have found is changing the venue for certain computers to a venue that only has multibeam (seti@home_enhanced) selected, and "allow work for other applications" deselected. That seems to be the only way (or the most reliable way) to get some tasks other than astropulse. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
DaveLee Send message Joined: 9 Jan 01 Posts: 7 Credit: 3,894,728 RAC: 0 |
Not a stupid question :) All are checked in my preferences. My concern with this is that my stats are falling through the floor while my system chugs along processing the APs. |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
Not a stupid question :) that happens for two reasons..the second is related to the first. 1) the task time is significantly longer than the MB tasks 2) because the task time is so long, you are waiting for wingmen to process and return their task before you can get your credit granted. You will get your credit in due time, and when the credit is finally granted, your RAC will make a decent upward jump, and so will the stats. If you feel comfortable with installing optimized applications, in general, that shows at least a 50% decrease in crunch time. On some systems, it is closer to 3-5x speedup. (mine is about 4.5x). All that does is gets the task crunched and returned, while I wait for the wingman to do the same. Credit and stats don't really bother me. Those are a nifty add-on to the whole point behind distributed computing, but the primary goal should be getting the scientific data analyzed. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
Bill Walker Send message Joined: 4 Sep 99 Posts: 3868 Credit: 2,697,267 RAC: 0 |
Forgive me for overlapping some other threads, but here are my comments on this topic. First, since installing ap_v5, I get almost (but not completely) only ap_v5 work units. I suspect that the Cuda crunchers are sucking up most of the regular multi-beam units. Also, because of the size of the ap_v5 work units, once I have 3 or 4 of these in my que then BOINC stops asking for more work for a while. Second, with the optimized aps, I'm getting more credit per CPU hour with ap and ap_v5 then I got with strictly multi-beam. Can't speak for non-optimized aps, I didn't run them very long. It make take longer, while you wait for wingmen to come through, but you do eventually get more credit. |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
I don't think the CUDA people are sucking up all the WUs. If I go to a MB-only venue, I have no problem getting 1-20 at a time. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
I don't think the CUDA people are sucking up all the WUs. If I go to a MB-only venue, I have no problem getting 1-20 at a time. Cuda person here. I have a few MBs left but the last couple of times it has asked for work all I've gotten is APs. PROUD MEMBER OF Team Starfire World BOINC |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
I don't think the CUDA people are sucking up all the WUs. If I go to a MB-only venue, I have no problem getting 1-20 at a time. I was down to a couple of hundred MB's (less than a day's worth with CUDA) earlier today, so I changed my preferences to MB only for a couple of hours and got over 1200 before I managed to change back. That should keep me going for a few days before I have to do it again. The trickle of AP downloads (just one for each request for work) seems to keep the CPU cores well supplied. Running 6.4.5 ATM. F. |
woodenboatguy Send message Joined: 10 Nov 00 Posts: 368 Credit: 3,969,364 RAC: 0 |
I don't think the CUDA people are sucking up all the WUs. If I go to a MB-only venue, I have no problem getting 1-20 at a time. Interesting - I have been struggling to keep the CPU busy with APs while getting an abundance of CUDA WUs whenever the machine requests more work. I resorted to asking for only APs in my preferences but typically get refused. When the cache looks like it will finally run completely empty of CUDA WUs as well I change my preferences back and get a flood of them at least. It's a surprize when an AP shows up. Regards, |
Bill Walker Send message Joined: 4 Sep 99 Posts: 3868 Credit: 2,697,267 RAC: 0 |
I don't think the CUDA people are sucking up all the WUs. If I go to a MB-only venue, I have no problem getting 1-20 at a time. I stand corrected. So, where are all the MB units going? From a quick glance at the server status page, there are still more MBs being created than astropulse. Just had a thought - the important indicator might not be number of WUs, but maybe user clock time for the two types of wu's. As a first assumption, if everybody does wu's as fast as me (~30 minutes for a MB, ~20 hours for ap), current server status says we are making .5*21 = 10.5 user hours/sec of MB, and 20*.4 = 8 user hours/sec of ap. The ratio of these 2 numbers should stay the same for other users, as long as ratio of their clock time for mb and ap remain the same as mine (does that make any sense?) So, why aren't we all getting about 10 mb hours for every 8 ap hours? Converting hours back to wu's using my numbers, thats 20 mb units for every .3 ap units. From rough memory, I have received about 1o or 12 ap units in the last week, and exactly 2 MB units (on my main machine only). We have a mb gap! Question for the Cuda Crowd: how many mb and ap units do you get per day, roughly? |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
from my last 90 ap 5.03 result duration: min. 48 739,52 sec max. 58 247,33 sec ---------------------- ap_5.03r112_SSE3.exe cpu Xeon 2GHz D5400XS V8-Xeon |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
I get no MB and, perhaps 7 or 8 AP (tends to be one / work request so Boinc has to send multiple work requests) unless I disable all AP and Other in which case I get a max day's worth of CUDA (1400) I assume - I managed to switch the AP preference back on today after 1200+ CUDA's had downloaded and I've had half a dozen AP's since then. Note that I am running a 2 day cache which currently has 45 AP 5.03's (about 10 hours each, 4 cores = over 4.5 days) and all those CUDA's (between 2.5 and 16 day's worth depending on how many are VLARs). I have found switching the preferences to be very successful over the past couple of weeks. F. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
I have allowed work 3 times this week and all have been AP 5.03 WU's thats 100 AP vs 0 MB In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
Yeah. I changed venues to an MB-only one and am starting to get MBs. Only one at a time because the request is usually less than 400 seconds, but after a few AP503's finish, I imagine the requests will be larger. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
If I don't touch my BOINC system, I'll have to wait about 15+ hours on my C2D system (E6700) before I get to my (first) AP_v5 workunit. It normally takes me 20+ hours to do a 5.00 AP (running opt. version), so from what I've read in the forum, I expect 30-40 hours for the new AP_v5). This could be interesting. Interesting as my E6550 takes about 10 hours with a AP unit and around 18 hours with AP_v5. It is only a 2.33 compared to your 2.66. |
Squyrrel Send message Joined: 8 Feb 00 Posts: 14 Credit: 24,882,903 RAC: 0 |
Could someone point me in the right direction to the preferences you are talking about please? I've looked at all the prefs options I can see in Seti, BOINC Client and BAM, but can't see the options for AP/MB. Thanks in advance. |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
Could someone point me in the right direction to the preferences you are talking about please? The dark-blue bar at the top and bottom of every page here at SETI > Account > SETI@Home preferences. You can edit the venues to only allow certain tasks, click update at the bottom, go back to Account, then "Computers on this account: View", go there, click the "details" link for the computer you want to change venue on, go down to the bottom of the page, and there's a drop-down box with the venues. Select the one you updated, and click the update button. The next time the BOINC client on that computer contacts SETI, the venue will be changed and it will only get the work you specified for the venue. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
Squyrrel Send message Joined: 8 Feb 00 Posts: 14 Credit: 24,882,903 RAC: 0 |
Many thanks. Seems so simple now! |
Matthew Love Send message Joined: 26 Sep 99 Posts: 7763 Credit: 879,151 RAC: 0 |
Question, The preferences are talking is this located in the BOINC program on the PC or on the SETI website where you manager passwords etc.. LETS BEGIN IN 2010 |
©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.