Message boards :
Number crunching :
Unmarked VLARs going to GPU
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
Fred's reschedule tool can handle that for you too Geek. It changes the RSC_fpops_bound so that you don't get those -177s. PROUD MEMBER OF Team Starfire World BOINC |
ded1o1 Send message Joined: 29 Sep 07 Posts: 68 Credit: 10,834,919 RAC: 0 |
In the last weeks i've got no .vlar on my GPU-only Rigs. But Today i got 287 Workunits marked as VLAR. Not old, unmarked, Workunits. I received a couple of hundred CUDA (GPU) VLARs today, they were not ghost re-sends because they ended in .vlar, seems like the new server side code that should only be sending them to CPU's has been disabled. Happly Fred's BOINC rescheduler spotted them and they are now waiting to be crunched by my CPU. Did they change anything? I remember that this happened once before when Matt recompiled the code, he said the new .vlar handling functionality is disabled by default. I wonder if that's the cause this time. I thought at first they had enabled the re-send feature to reduce the ghost armies haunting their servers, perhaps because the janitors threatened to form a picket line unless something was done about all the ectoplasm :p |
Helli_retiered Send message Joined: 15 Dec 99 Posts: 707 Credit: 108,785,585 RAC: 0 |
Oopsie - sounds like he forgot to reenable this Feature before he compiled the new Code with this Resend-Feature. ;-) Helli A loooong time ago: First Credits after SETI@home Restart |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
I noticed and wrote about it in a forum thread, but since then, I've yet seen very little WU's, marked as _vlar. I use BOINC 6.10.58 64BIT and when I hover over a filename, I see CUDA or 603 and no, I'm hallucinating ;) Anyone tryed this, but those FERMI, don't really have a problem with a vlar WU's, depending on the card, the amount of WU's ran together, can increase time a lot. Haven't used the [i]REScheduler[i], since I had to install, WIN XP64 on a new drive. |
JohnDK Send message Joined: 28 May 00 Posts: 1222 Credit: 451,243,443 RAC: 1,127 |
Well also just got VLARs _0 & _1 sent to the GPU... |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
|
SciManStev Send message Joined: 20 Jun 99 Posts: 6653 Credit: 121,090,076 RAC: 0 |
I quickly downloaded Fred's rescheduler and ran it on my XP machine, I will have quite a few more to move once they finish downloading. I have to say that Fred's rescheduler tool is excellent. The old 1.9 tool, was very good, and gave us an option when we needed it, but this tool goes much farther. The best part is that Fred can still adapt it if needed. This tool has saved me a lot of grief. Steve Warning, addicted to SETI crunching! Crunching as a member of GPU Users Group. GPUUG Website |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
I am sure this is just an oversight. The 'resend' feature code has not been used here for quite some time, and I am sure that when it was restarted, thought was not given to the recent changes to the scheduler code which stopped sending VLARs to GPU task requests. DA will probably pick up on that and make repairs at a later date. But for now, it's probably still a good thing to have the resends working to clear some of the ghosts out of the closet. It's not like it really breaks anything.....the GPUs will process the VLARs....although they are not good at it, they won't crash. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
S@NL - John van Gorsel Send message Joined: 5 Jul 99 Posts: 193 Credit: 139,673,078 RAC: 0 |
....the GPUs will process the VLARs....although they are not good at it, they won't crash. My experience is that under Linux, VLAR's mess up the GPU in such a way that only a reboot helps (frozen screen). Aborting the VLAR and restarting BOINC doesn't help so it's probably a memory issue. Since GPU on Linux is not a supported option and I wrote my own crude rescheduler it's not really a big deal. Seti@Netherlands website |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
....the GPUs will process the VLARs....although they are not good at it, they won't crash. Yes, on Windows it can depend very much on the card involved too. In general, higher performance cards will crunch them (though painfully slowly), while smaller more mainstream GPUs can and do experience poor machine usability, driver crashes or restarts under certain circumstances common with VLAR. So for many of us this hard road hasn't merely been a preference for high throughput, but for protecting the usability & stability of our machines for work & play. Not crunching on GPU while using the machine has always been an option, though that doesn't protect from possible negative consequences of frequent crashing to filesystem, OS and blood pressure ;) A curious side effect of all the difficulties, is that it has led down development roads for some of us that gradually open up a greater understanding of specific project algorthms, along with a more in depth understanding of GPU programming & architectures, than might otherwise have been needed or pursued... Never thought I'd look at it that way, but VLARs are good for develeopment :P Jason "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. |
S@NL - eFMer - efmer.com/boinc Send message Joined: 7 Jun 99 Posts: 512 Credit: 148,746,305 RAC: 0 |
Got close to 1000 VLAR's on one machine, they are really plentiful. Never seen so many of them on one machine. As the scheduler gone mad on that machine and send me a record number of WU, I still got plenty of GPU tasks. TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking. |
lupo Send message Joined: 29 Aug 10 Posts: 91 Credit: 4,736,407 RAC: 0 |
Is there a way to edit the seti configuration to ONLY process vlar's on the CPU's and leave everything else to the GPU's? Right now I have my CPU's totally disabled in my config and only process on my two GTX 470's. I noticed today my GPU's working on processing 2 vlar WU and they were taking about 2 hours per WU on the GPU's. I wondering if the VLARs will process any faster on my Q6600 CPU. Adam |
Blake Bonkofsky Send message Joined: 29 Dec 99 Posts: 617 Credit: 46,383,149 RAC: 0 |
Try the reschedular by the man that posted above you. I use it every time I get a fresh batch of WU's to move the VLAR's to the CPU. http://www.efmer.eu/forum_tt/index.php?topic=428.0 |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
Fred's newest rescheduler tool allows moving everything but VLAR's to the GPU. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
For those that missed it, this was posted yesterday by Matt... A scheduler that accidentally handed out VLARs to everybody was released to the wild briefly last week, but has long since been captured and detained. Jeff and I came up with a scheme (which Dave will implement) to better prevent this from happening in the future. PROUD MEMBER OF Team Starfire World BOINC |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65771 Credit: 55,293,173 RAC: 49 |
For those that missed it, this was posted yesterday by Matt... Yeah, I just had a vlar on My 295, It was easy to spot, As It said vlar, vlar, vlaaaar! So I impaled It, Now as soon as the hdd has some room I can upload and report once again. :D The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
Hmmmm, I wonder if the scheme they came up with was to not send anything out. That would stop it!! :-) PROUD MEMBER OF Team Starfire World BOINC |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65771 Credit: 55,293,173 RAC: 49 |
Hmmmm, I wonder if the scheme they came up with was to not send anything out. That would stop it!! :-) I have no idea Perry, Maybe if someone in the know were to pop in and tell US We'd know. Until then I guess We're playing the part of: VLAR the Impaler... ;) The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65771 Credit: 55,293,173 RAC: 49 |
Well I have 4 vlars, 3 on the gpu and 1 on the cpu on the PC that are hiding and I aborted 166 for the gpus that almost crash My PC(the 166 were marked vlar, I thought that was not supposed to happen anymore, As in no vlars on the gpu anymore), When they run the only thing that works for sure is the mouse cursor, everything else is close to locking up, and since they all look a like, 18 vlars for the cpu got axed too, Those vlars shouldn't be on the gpu, Hidden or Not, Yet there they were. Hopefully the staff can fix this asahp. The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Well I have 4 vlars, 3 on the gpu and 1 on the cpu on the PC that are hiding and I aborted 166 for the gpus that almost crash My PC(the 166 were marked vlar, I thought that was not supposed to happen anymore, As in no vlars on the gpu anymore), When they run the only thing that works for sure is the mouse cursor, everything else is close to locking up, and since they all look a like, 18 vlars for the cpu got axed too, Those vlars shouldn't be on the gpu, Hidden or Not, Yet there they were. Hopefully the staff can fix this asahp. Get either Rescheduler; you can set them so that any VLARs sent you are automatically moved to the CPU, regardless of whether they have "vlar" in their names or not. |
©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.