Message boards :
Number crunching :
VLAR's to Kepler cards.
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next
Author | Message |
---|---|
S@NL - eFMer - efmer.com/boinc Send message Joined: 7 Jun 99 Posts: 512 Credit: 148,746,305 RAC: 0 |
I made a new BoincTasks thread so that more users will notice it. It is a feature going back to BoincTasks 1.40. Was mentioned in the change log and is logged in BoincTasks. TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking. |
Wiggo Send message Joined: 24 Jan 00 Posts: 34844 Credit: 261,360,520 RAC: 489 |
Either there arn't as many VLAR's around recently or the servers are taking the hint that I don't want them on my cards as I havn't had to abort that many today. Cheers. |
Fred E. Send message Joined: 22 Jul 99 Posts: 768 Credit: 24,140,697 RAC: 0 |
I also run BoincTasks and apparently it caused the problems I reported in this thread. Will watch for new version, thanks. P.S. Had another suspend last night and almost ran out of gpu work before I saw it this morning. Another Fred Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop. |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
I made a new BoincTasks thread so that more users will notice it. And I have posted my thoughts about it in the new thread. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Wiggo Send message Joined: 24 Jan 00 Posts: 34844 Credit: 261,360,520 RAC: 489 |
Either there arn't as many VLAR's around recently or the servers are taking the hint that I don't want them on my cards as I havn't had to abort that many today. I obviously spoke to soon as I've just aborted several more of the rotten things. :-( Cheers. |
mr.mac52 Send message Joined: 18 Mar 03 Posts: 67 Credit: 245,882,461 RAC: 0 |
Thanks Fred, I'll take your advice and not move the GPU VLAR's to the CPU any more. The rescheduler was a very handy tool back in the day! John |
Tom* Send message Joined: 12 Aug 11 Posts: 127 Credit: 20,769,223 RAC: 9 |
VLAR's on my GTX660 take 2-3 times longer than normal angle ranges. This would be fine if we received comesurate credit. Question - Is the low credit vs runtime due to creditnew? as the increased runtime does not seem to be due to cputime? Can we expect an increase down the road as more VLAR's on Keplers get validated? |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
perhaps yes. When slow devices will return more work credit should rise... at least peoples who somwhow understand that "CreditNew" say so... SETI apps news We're not gonna fight them. We're gonna transcend them. |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
at least peoples who somwhow understand that "CreditNew" say so... That people realy exist? Credit new is some new type of "black hole" |
William Send message Joined: 14 Feb 13 Posts: 2037 Credit: 17,689,662 RAC: 0 |
at least peoples who somwhow understand that "CreditNew" say so... IMO not even David understands CreditNew. A person who won't read has no advantage over one who can't read. (Mark Twain) |
W-K 666 Send message Joined: 18 May 99 Posts: 19074 Credit: 40,757,560 RAC: 67 |
at least peoples who somwhow understand that "CreditNew" say so... I'm not convinced he knows how BOINC works, it certainly doesn't work the way he thinks it does, so how he thought he could write a new credit system is beyond me. |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
As I post on beta, a simple <VlartoGPU>0|1<VlartoGPU> at the client side, could easely solve the problem, allowing we to choose if we want or no crunch VLARs on our GPUs. |
Tom* Send message Joined: 12 Aug 11 Posts: 127 Credit: 20,769,223 RAC: 9 |
Yeah my APR on cuda50's was just taking hold when I got a bunch of VLAR's now I'm back to crunching cuda42:-( |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
Any news on this area? If my memory don´t fail, some time ago, i see a vlar aborting script, does any one still have one or knows where to find? The video lag on the multiple GPUs when runing 2 or more Vlars is to high, even with changes in the configuration (mbconfig), so i need to find an automatic way to abort them or i will be forced to stop crunching SETI while use the host (loose about 40% of the processing time of the day) because you never know when the Vlars arrive and can´t babysiting on each host to manualy abort them each time they arrive. To de development people: Please give me a way to easely keep the Vlars away from my Nvidia GPUs! as was working on V6. |
Tom* Send message Joined: 12 Aug 11 Posts: 127 Credit: 20,769,223 RAC: 9 |
I know we have no experts here on the intricacies of creditnew, but with the knowledge that reassigning gpu tasks to cpu's and vice versa can have a disasterous affect on the resulting credit awarded, is it Possible that the schedular sic sic reassigning vlar'S to gpu's has had a major affect on the credit awarded since V7 was released?? All those long long running cpu jobs running vlars are now running much faster on GPU's. Or did they start over with v7 applications in figuring out credits from V6 apps? after all VLAR's are the same application as far as creditnew is concerned?? or is it? |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Just a bit of news. I contacted Eric to see if he could stop sending VLARs to nVidia GPUs. He answered yes, he is going to try to rewrite the code for the servers tomorrow and hopes to have it in place on Monday or Tuesday. Hopefully his coding efforts go well. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
ExchangeMan Send message Joined: 9 Jan 00 Posts: 115 Credit: 157,719,104 RAC: 0 |
Just a bit of news. Thank you for the information. Best news I've had in the last few days... +100 |
Wiggo Send message Joined: 24 Jan 00 Posts: 34844 Credit: 261,360,520 RAC: 489 |
Just a bit of news. That will cure my only problem with V7 then, I await with bated breath. Cheers. |
rob smith Send message Joined: 7 Mar 03 Posts: 22220 Credit: 416,307,556 RAC: 380 |
I wouldn't mind running VLARs if they were treated as a different "type" of WU, in the way that AP are different. I'm sure it is not impossible to have BOINC manage this situation better than the all or nothing switch. Its a real PITB to see a task identified as being a VLAR, with an estimated run time of say 15 minutes taking an hour or more and thus skewing the figures used to calculate the run time. Now there's a challenge for you Eric... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
W-K 666 Send message Joined: 18 May 99 Posts: 19074 Credit: 40,757,560 RAC: 67 |
I wouldn't mind running VLARs if they were treated as a different "type" of WU, in the way that AP are different. I'm sure it is not impossible to have BOINC manage this situation better than the all or nothing switch. Its a real PITB to see a task identified as being a VLAR, with an estimated run time of say 15 minutes taking an hour or more and thus skewing the figures used to calculate the run time. Now there's a challenge for you Eric... I was wondering if the VLAR's were actually been treated as outliers. I don't think this computer has done anything else on the GPU for over a day. I seem to have a lot of "Consecutive valid tasks" but significantly less "Number of tasks completed" and the APR has not decreased, or if it has not significantly. This might explain why the credits are not being adjusted as expected under CreditNew. The present state of play is; SETI@home v7 (anonymous platform, nvidia GPU) Number of tasks completed 424 Max tasks per day 670 Number of tasks today 14 Consecutive valid tasks 570 Average processing rate 186.00525336363 Average turnaround time 0.54 days |
©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.