81) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542994) Posted 17 Jul 2014 by Cavalary Post: While performance / watt is a crucial characteristic when I'm looking into CPUs and video cards in general, when it comes to crunching I'm largely just looking at total power use... And have to say, didn't so far consider GPU-only crunching as a possibility (is it? as in strictly GPU-only, not sort of "well, mostly GPU, but there are things it can't do so there will keep being times with both CPU and GPU whenever"), so I was just thinking it'd be added on top. Plus that I'd be worried about driver updates required, which as I said may cause issues with what I'd sometimes want to play, and with how priorities work for GPU and of any performance hit if crunching while the video card would be under high use otherwise. As for solar panels, heh, getting past the "little" issue of affording them, I'd put them where? 1st floor (or 2nd, if you count ground floor as 1st and not separately as we do here) of a 10 (or 11)-storey apartment building, on the north side, and with another building of the same height just on the right, perpendicularly. And if it'd be an update issue still don't see why'd it only affect any SETI@home WU that's the 2nd running at the time, not if it's first and not Astropulse units of any sort, and why it'd keep applying to all newly-downloaded ones too, even weeks after said update.

82) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542838) Posted 17 Jul 2014 by Cavalary Post: Yeah, that's what I was saying, so that driver should have no relevance either way. Plus that if I get around to playing something it's usually an old game, and I've learned quite a long time ago that avoiding a video driver update unless absolutely necessary is a good idea in this case. (Also, don't think I'm letting Windows Update do what it wants. It's set to check and notify, then I do the rest, ignoring video driver updates as a rule, and looking into any other drivers that may show up before ticking them.) Thing about GPU processing is that as an environmentalist there are times when I find it hard to justify keeping my computer to crunch on either way, so adding the GPU power use on top of the CPU one is not something that ever seemed right. When we'll get to 100% renewables I may reconsider, though I'd still be very edgy even then, for a number of different reasons. Otherwise, 4 more completed fine, 2 more going now... Though 2nd is slow and uneven in pace, should be one of the typical ~3h45m ones but 1.5h gone and 3.5h remaining estimated, nearly all CPU power going into those but seems to be going in fits, alternating the speed I'd expect with barely crawling, estimate either increasing by a second per second passed or dropping by 3-5 / each. But maybe just paying too much attention... I hope.

83) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542704) Posted 16 Jul 2014 by Cavalary Post: Actually, don't recall what I did with the WUs before updating. But either way nothing to do with the units already downloaded, much less possibly active, at the time, since the update was on Jun 27 and on Jul 15 I watched it instantly go through each in the list, download 100 more and instantly "finish" each of those too, and checking the tasks list here showed a whole other bunch like that on Jul 13 too. No NVIDIA update (and also not using GPU). Did the last Windows update on Jul 9. Sure, BOINC was running while the Windows updates were installed, always is, never caused any trouble so far. And obviously rebooted after said updates, how'd you finish installing them otherwise?

84) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542674) Posted 16 Jul 2014 by Cavalary Post: Hoping I won't jinx it, but worked fine since then. Just had it update, now checked task report in my account here, 17 reported tasks today, 11 valid, 5 validation pending (4 that ran normally plus the one that finished after 48 sec), 1 validation inconclusive (the one that stopped after some 3 min, which as you can see has the other result with little CPU time as well, so either we both glitched or the WU has issues). Also glanced at the reports a bit more yesterday and saw a whole slew of instant finishes reported on Jul 13 too, so guess the problem was older and I had just noticed it then. Granted, did notice that it was working on Astropulse WUs unusually often, as in one almost always, occasionally two, recently, not sure if since the update or just for the past week or so, so may be that it just kept skipping any S@h WU that ended up being 2nd for quite a while, but as long as it had AP ones, and considering how long those take, I didn't happen to see it. Not daring to reboot again just to test what will happen then though, will see when I'll next really need to I guess. Also, not updating BOINC again after this :/ Calculating average daily credit from the last 40 days report on BOINCStats, Jun 7 - 26 (before update), 1187.7. Jun 27 - Jul 15 (on latest version), 955.3, which is actually boosted by a 2357 on Jun 28, so just after update, possibly clearing away what was already finished before, so 877.4 if you exclude that day. For today, what was crunched after reverting, says 1188.

85) Message boards : Number crunching : Welcome to the 15 year Club (Message 1542383) Posted 16 Jul 2014 by Cavalary Post: Oh, could be worse. (Though I do still sometimes wonder at the unfair ratio between "could be worse" at bad things and "could be better" at good ones :p) Me, joined as soon as I could. Was behind a proxy at the time, so had to wait for the version (1.06, was it?) which supported those, though I first heard of S@h about as soon as it launched on the forum of (what used to be) a computer games magazine here (either already no longer in print or soon to poof at that moment). People jumped on the wagon, some gave up sooner or later, remember one who was too disappointed after completing the first WU and getting nothing from it, seemed to be expecting some readable results or at least some congratulations of some sort... Me, in for the long haul when I decide something. Initially PII 266 MHz, 32 Mb RAM, later 64 Mb. Then in 2002 P4 2 GHz, 512 Mb, down to 256 I believe in 2006 when one module failed. Since 2008, Core 2 Duo 3 GHz, 2 Gb. Currently wondering if it may be failing :/ Hope not... Poor things sure crunched though.

86) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542133) Posted 15 Jul 2014 by Cavalary Post: Ok, unistalled BOINC and cleared data, reinstalled 7.0.28 which was what I had before, only recently updated to latest version... Jun 27 says the file date, which incidentally marked a moment when my RAC started dropping, quickly from ~1200 to ~1000 where it was for the past week or so. At first, started 2 WUs fine. Reboot, one of them is instantly reported as complete, but starts another one fine. Suspend both to make it start new ones, starts them fine. Repeat this three times, still seems to work. Close BOINC Manager (as in just the window, not exit it from the context menu), open it again, one of those currently being worked on is reported as completed, after 48 seconds. Now continuing on another that was started, after I unsuspended them, so far going for 8.5 min. EDIT: Just realized something. Thought it just closed suspended tasks, starting them up again when it gets back to them, but I see several idle instances of S@h there in Task Manager, so guess suspend really is suspend. So new hypothesis: The error is with the second BOINC project process at any point, so if I suspend it and make it start another, it'll no longer be the 2nd process. Would indicate software issue and that my CPU's fine after all? But what issue then? Not like I did anything today in particular...

87) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542110) Posted 15 Jul 2014 by Cavalary Post: Further test, if I start it on the first core, it can continue on the 2nd. No idea if that means it only does the checks at start or the problem is only there or what.

88) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542061) Posted 15 Jul 2014 by Cavalary Post: Seems to be a core issue. Works on the first core, 2nd just instant finish, assuming with errors. Does this mean something just happened to my CPU? Just got on the computer around that time today... (But computer's on 24/7 otherwise.)

89) Message boards : Number crunching : Instant finishing of all WUs? (Message 1542054) Posted 15 Jul 2014 by Cavalary Post: Just noticed that SETI@home was instantly finishing all WUs, so checked the log and seems that it started doing it as of about 45 min ago (at the time I write this), after it finished an Astropulse WU and went back to S@h on one core (still has an Astropulse one on the other which seems to keep running). Snoozed it, restarted, had it do benchmarks, all seems fine (and Astropulse one restarted fine too), but S@h ones just keep doing this. What's going on? Temperatures normal, 66-67C each core as reported by OpenHardwareMonitor (before the snooze, 63-65 now shortly after restarting work), it's where it tends to sit at full use.

90) Message boards : News : AVG 2013 virus scanner false positive on SETI@home 7 for Windows (Message 1374205) Posted 31 May 2013 by Cavalary Post: No issues with BitDefender on either of those two files either, and this one's quite known for false positives too.

91) Questions and Answers : Windows : Hijack? (Message 522558) Posted 24 Feb 2007 by Cavalary Post: Seems like you have issues again. (Had 3 WUs waiting for download, did close/restart it, then rebooted, no dice, used /flushdns /registerdns as it says on site, one WU got downloaded but the other 2 still hanged, tried connection repair for all it's worth, but no change.) And it's doing it all over again. Same thing if I use just IE too... So, what does it do differently when it can't connect to SETI server?

92) Questions and Answers : Windows : Hijack? (Message 518366) Posted 16 Feb 2007 by Cavalary Post: Waited till I had 5 units completed, then tried connecting and at the same time loading sites. Still in NetCaptor too, planned to suspend while it was downloading new work and go on IE if it caused trouble again. It didn't... So appears to be linked to the times S@h has server issues. Probably the times it can't connect to scheduler and attempts a test connection to project site to see if the connection works. This makes me wonder what were the very brief hijacks that have happened weeks ago caused by though...

93) Questions and Answers : Windows : Hijack? (Message 517054) Posted 13 Feb 2007 by Cavalary Post: I'll try that next time S@h has something to transfer. Just managed to finally transfer this afternoon so nothing now. Would be weird to be it though, I mean if that's it, why now after all this time? Any S@h updates lately that changed the way it handles these transfers? Actually that transfer went a bit funny. It started with uploading results, which was over fast, then froze there, saying it can't find the work units. Browser going nowhere (timeouts) during that time. I suspended again when I knew my AV will check for updates. It did and managed to connect and update just fine, browser working too just then. But browser started working only after AV connected to update server, like it needed that connection from another program to separate itself from BOINC, or BOINC needed a connection from something other than the browser to stop rerouting, either way. Then when that was done I set BOINC to attempt to connect again and it finished downloads just fine too and once it was done with those everything was fine again. Well, this is weird. Clicked update now and it uploaded 1 result (didn't notice that there) and downloaded a new work unit. Not only did it manage both without hanging, but I specifically refreshed a page in NetCaptor just then to see what happens and it worked just fine. Quite confused (and feeling a bit silly, in case this will end up fixing itself)... Think I'll suspend networking for good again and let it finish several work units, so it'll have more to transfer next time I'll test, see what happens. Will be a few days...

94) Questions and Answers : Windows : Hijack? (Message 516978) Posted 13 Feb 2007 by Cavalary Post: Browser is NetCaptor 7.5.4, which uses the currently installed IE core, which is IE7. Been using NetCaptor for years, there haven't even been new versions for a long time, and the upgrade to IE7 was when it came out, months ago, so that can be ruled out since the problems started afterwards. Firewall is that from BitDefender 10 Internet Security. Just upgraded from BD9 Pro 2 days ago, so once again can be ruled out as problems started before. boinc.exe is set to have full permissions (both ways, all addresses, all ports) over TCP. No connection sharing. No proxy.

95) Questions and Answers : Windows : Hijack? (Message 516946) Posted 13 Feb 2007 by Cavalary Post: Noticed that every so often (with an annoying tendency to happen when I was sending mails or posting messages), as of a few weeks ago, my browser got hijacked. As in I was trying to go to a page but it was redirecting me through a server different than the one of the site it was on, so I was getting 404 errors. Till a few days ago it was just a momentary thing, just refresh and I was all good again, and since my system came up clean at all scans I blamed it on my ISP, who has this method of displaying announcements by rerouting users through their servers, so you see the announcement when you go directly to any site (then you can go back to regular browsing), but get 404 errors if you try to go directly to a specific page on a site. But as of a few days ago, it was staying hijacked for a long time, long enough for me to figure out where it was going (especially since sometimes it does go to the specific pages, nomatter where I'm actually tring to go to). Apparently SETI@home (I'm not running any other BOINC project) reroutes everything through the address it's trying to connect to, because I end up on setiathome.berkeley.edu or setiboincdata.ssl.berkeley.edu or www.google.com (apparently it tries to do Google searches when it gets a "not found" error, mainly I think because it's rerouting itself too, and when it tries to upload and download at the same time, or download from different sources, naturally this happens since it only routes to one site). Guess it's staying hijacked longer because of the problems with the S@h servers these days, keeps rerouting me till it completes whatever it's trying to do. And since it couldn't do that, I had to suspend BOINC networking so I'll be able to actually get online anymore. Needless to say, that's darn annoying! So, any fixes? (Oh, been running BOINC 5.2.8 for a long time now. Just upgraded to 5.8.8 a couple of days ago since I went to the BOINC site trying to figure out what's going on and saw what the latest version is. BOINC could sure use an auto-update option...)

Previous 20

 
©2020 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.