Message boards :
Number crunching :
Anything relating to AstroPulse tasks
Message board moderation
Previous · 1 . . . 19 · 20 · 21 · 22 · 23 · 24 · 25 . . . 120 · Next
Author | Message |
---|---|
Speedy Send message Joined: 26 Jun 04 Posts: 1643 Credit: 12,921,799 RAC: 89 |
Damn, and my main cruncher is missing the party. Sitting here in the dark, typing this on my BlackBerry. We had a wicked wind storm today, 200,000 homes without power. Been out for 9 hours now, and the real bummer is 3 doors down has power. :( That's no good to hear at all. I hope everyone is okay. I'm sure you have already thought of this have you checked the fuse box? Hope you get power back on soon |
betreger Send message Joined: 29 Jun 99 Posts: 11366 Credit: 29,581,041 RAC: 66 |
Aps have been flowing along quite nicely but it look as though it's going to end soon. |
betreger Send message Joined: 29 Jun 99 Posts: 11366 Credit: 29,581,041 RAC: 66 |
14 channels of APs to split and no splitters running = Panic? |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1855 Credit: 268,616,081 RAC: 1,349 |
28 channels now... All just money in the bank!!! |
betreger Send message Joined: 29 Jun 99 Posts: 11366 Credit: 29,581,041 RAC: 66 |
14 channels of APs to split and no splitters running = Panic? Properly stated 28 channels now... Total panic !!! |
wiesel111 Send message Joined: 5 Jan 08 Posts: 9 Credit: 1,227,675 RAC: 1 |
New statement from Matt: http://setiathome.berkeley.edu/forum_thread.php?id=78080#1720691 So all the (ap) assimilators and splitters will be offline about a week (possibly longer than a week). |
betreger Send message Joined: 29 Jun 99 Posts: 11366 Credit: 29,581,041 RAC: 66 |
The SSP shows lots of APs waiting to be split, next week looks very promising. |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
The SSP shows lots of APs waiting to be split, next week looks very promising. Well.........maybe. Don't forget that these are still 2011 datasets, mostly for reprocessing the MB work with v7. A lot of those 2011 AP splits do not produce any new AP work. A few do. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
So tasty. I set No New Tasks a few days ago just to keep from pestering the scheduler for APs that it won't have, and I saw that they are being splat again, so.. om nom nom nom. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
JaundicedEye Send message Joined: 14 Mar 12 Posts: 5375 Credit: 30,870,693 RAC: 1 |
!!!!!! JUST when I filled the hoppers with MBs, I'm sure the APs will be long gone before I start getting any. That's what I get for running on auto. "Sour Grapes make a bitter Whine." <(0)> |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
Oh, my.... The kitties are noshing on APs....and the crunchers are lovin' it. Meow! "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1855 Credit: 268,616,081 RAC: 1,349 |
Enjoying the current AP stuff flowing, and pleased that I just managed to get one of my crunchers (WinXP Pentium D, 3.40g) upgraded before it hit. Swapped out a GT610 with a GTX750TI SC (new driver 355.82), ran Lunatics installer (0.43b) and updated BOINC to 7.6.9. All pretty routine, by now. BUT Looking at one of the results of good work (it validated, at least) there's a lot of errorish looking stuff in the results file I'm not used to seeing. Wondering if one of you more knowledgeable folks could take a peek and see if there's anything that should require action on my part? https://setiathome.berkeley.edu/result.php?resultid=4362310672 What I'm talking about is all the scratch_space messages. Makes me wonder if either something went sideways with the Lunatics installer, or perhaps the NVidia 355.82 driver is a problem? I know there were some issues with some newer NVidia drivers. Would appreciate any observations! Thanks! Edited: to include Lunatics & BOINC version numbers; I should know better ... |
WezH Send message Joined: 19 Aug 99 Posts: 576 Credit: 67,033,957 RAC: 95 |
"New nVidia drivers for Windows with OpenCL 1.2 support (350.x) declare bool2 as reserved type that leads to compilation error if bool2 redefined as structure." Open AstroPulse_Kernels_xxxxx.cl with notepad and change every bool2 to bool_2 EDIT: If I remember correctly, that is fixed in newest Lunatics v0.43b setup. |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1855 Credit: 268,616,081 RAC: 1,349 |
"New nVidia drivers for Windows with OpenCL 1.2 support (350.x) declare bool2 as reserved type that leads to compilation error if bool2 redefined as structure." Apparently not fixed in 0.43b, or at least not in the version I downloaded from Mike's World this afternoon. But I am amazed and impressed that I didn't even have time to go back and edit my original message for typos before you had sent me the solution, which I have implemented. :) Assuming it was the straight .cl file that needed changed rather than the .cl_board name type file, which did not seem to have any bool satements. Also unclear, but I assumed that a restart of BOINC was also advisable, so did so just in case. In case it wasn't clear, _many thanks_, and can you pass on that apparently this is _not_ fixed in 0.43b? Thanks, Jim ... |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14655 Credit: 200,643,578 RAC: 874 |
Apparently not fixed in 0.43b ... Could anyone familiar with the construction and use of the Installer please read this report and check that I haven't made a silly mistake? Looking at the original source files for AstroPulse_Kernels_r2887.cl that are packed into the installers (32-bit and 64-bit), both have the expected #if 1 [unconditional] ... bool_2 #else [unused] ... bool2 #endif legacy path, but that worked fine on my machines, and I can't see any other problems. Maybe a second pair of eyes...? Edit: looking at the example https://setiathome.berkeley.edu/result.php?resultid=4362310672, I see AstroPulse v7 Windows x86 rev 2721 From installer 0.43b, that should be rev 2887, and those numbers should be in the name of the .cl file you edited (Jim, could you check and let us know what file number you have, please). I can't quickly locate where r2721 comes into the scheme of things, but it's certainly not the right version to use for NVidia cards. |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1855 Credit: 268,616,081 RAC: 1,349 |
From installer 0.43b, that should be rev 2887, and those numbers should be in the name of the .cl file you edited (Jim, could you check and let us know what file number you have, please). I can't quickly locate where r2721 comes into the scheme of things, but it's certainly not the right version to use for NVidia cards. Can't explain why it was using 2721, as that was old NVidia, from before the upgrade and still using the GT610. Maybe that's what it ran because the compile failed? Anyway, after editing the file Astropulse_Kernels_r2887.CL to fix the Bool issue as instructed above, all is happy. Here's a link to a recently completed job. It is using 2887. https://setiathome.berkeley.edu/result.php?resultid=4362440930 I can pass on the copy of 0.43b that I got from Mike's, if anyone wants to look at anything in the actual file. Or just grab the x86 I did here: http://mikesworldnet.de/download.html That machine is X86, so that's the one I downloaded. Perhaps the fix made it into the X64 but not the X86 file? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14655 Credit: 200,643,578 RAC: 874 |
I've checked the x86 v0.43b download from Mike's mirror, and it has the same fixed 2887 .cl file as my original sources. I've found r2721: that was in the original 0.43 installer, quickly replaced by r2737 in 0.43a when excessive memory usage problems surfaced. Those date back to the original launch of AP_v7 in October last year. Edit - looking again at result 4362310672, it has three visible (failed) attempts with rev 2721, but finally - right at the bottom - a successful run with rev 2887. It must already have been trying to run with the old files before you ran the v0.43b installer, and been finished off afterwards. I think we can say that v0.43b has done its job! |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1855 Credit: 268,616,081 RAC: 1,349 |
looking again at result 4362310672, it has three visible (failed) attempts with rev 2721, but finally - right at the bottom - a successful run with rev 2887. It must already have been trying to run with the old files before you ran the v0.43b installer, and been finished off afterwards. I had about 10 APs on the machine when I commenced the software portion of the upgrade. Usually I try to take upgrades step by step, with some burn time in between steps, but in this case I needed to press forward with the software portion because the APs were failing to run. Sorry I didn't connect the dots on that piece of it all in my mind. I had forgotten that the APs had tried to run. I think we can say that v0.43b has done its job! Not sure that follows, as I still don't get why if 0.43b did it's job I had to manually edit the .cl file to change the bool statement format, if that is a file .043b created. Previous state of existing any AP WUs on the machine shouldn't effect that, should it? Pardon the pun, but at least to me that doesn't compute ... But then I don't have an in-depth understanding of all the files and structures. Thanks for all the help! |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14655 Credit: 200,643,578 RAC: 874 |
Don't think that follows from the above. Original issue was that after running 0.43b, I had to manually edit the .cl file to change the bool statement. That's not related to the fact that the APs had previously run and failed, logging a 2721 attempt each pass. Separate issue, imho. Did you actually see tasks restarting, failing, backing off, trying again... after you'd run the v0.43b installer and allowed BOINC to restart? I'd have expected them to start running cleanly straight away with no further ado, though of course there would have been other tasks still 'waiting to run' from the failed rev 2721 attempts. As I mentioned in my first answer, there are still bool2 references in the rev 2887 .cl file, but they are in the inactive legacy path. So your search/replace would have found things to change, but I honestly don't think you 'had' to go through that additional step - we've had something over 3,000 downloads of v0.43b since release, and nobody else has reported having to edit the current .cl file. |
Mike Send message Joined: 17 Feb 01 Posts: 34272 Credit: 79,922,639 RAC: 80 |
Don't think that follows from the above. Original issue was that after running 0.43b, I had to manually edit the .cl file to change the bool statement. That's not related to the fact that the APs had previously run and failed, logging a 2721 attempt each pass. Separate issue, imho. That`s what i was thinking of as well. With each crime and every kindness we birth our future. |
©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.