Message boards :
Number crunching :
Lunatics Windows Installer v0.40 release notes
Message board moderation
Previous · 1 . . . 5 · 6 · 7 · 8 · 9 · 10 · 11 . . . 14 · Next
Author | Message |
---|---|
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
/HUG Jason Ive noticed this weird long running tasks some time ago but was only on some shorties... Ive thought it was an error on the ETS and I forgot about it. Now, Ive tried the Process Lasso on the host with the 560Ti's and Im crunching at twice the speed on WUs that have the same estimated task size... (may be there is some difference in the AR acting up, but definatelly the GPU usage went up...) Ill let it run a bit more and then Ill try the same on the other hosts... One thing to share about process lasso, Im system engineer and Ive felt a bit intimidated about touching those settings (may be, because I know all the things that could go wrong :D ) So if it were to me, I would preffer to have a command line option or a separated version selectable from the isntaller or whatever that dont rely on a priority rescheduler... Just a though... |
JLConawayII Send message Joined: 2 Apr 02 Posts: 188 Credit: 2,840,460 RAC: 0 |
If I remember correctly, Folding@home has such a setting for making task priority higher if you're having trouble with the GPU client. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14649 Credit: 200,643,578 RAC: 874 |
If I remember correctly, Folding@home has such a setting for making task priority higher if you're having trouble with the GPU client. So does GPUGrid, with their SWAN_SYNC environment variable - which, I gather, is supposed to switch the app from busy-wait to polling for synchronisation, or vice versa, or something. Never seemed to make much difference for me, but it's another way of getting control options from the user to the application. |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
A couple of comments on freeing one core, as a refinement but not as a recommendation. For any host with up to 100 CPU cores, setting 99% frees one core. BOINC rounds down to the nearest fraction. I wouldn't use that setting. Instead, when setting the <count> fields to control how many GPU tasks run at once I'd set the <avg_ncpus> fields such that when all GPUs had work there would be a CPU core freed, but if GPU work was running out all CPU cores would go back to doing pure CPU tasks. For a single GPU situation, <avg_ncpus> would be the same as or a tiny bit higher than <count>, for multiple GPUs it would scale down: 1 GPU 2 GPUs 3 GPUs <count> <avg_ncpus> <avg_ncpus> <avg_ncpus> 0.5 0.5 0.25 0.167 0.33 0.34 0.167 0.112 0.25 0.25 0.125 0.084 0.2 0.2 0.1 0.067 Joe |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13720 Credit: 208,696,464 RAC: 304 |
How many people have posted in this thread? Without counting, I'd guess at fewer than 20. What do the other 660 think? I only do MB work, so i haven't bothered with the latest release as it doesn't affect me. The advantage of the Lunatics installer is that allows pretty much anyone to install an optimised application without having to muck around manually copying & editing files. Adding yet more options, command line switches & other refinements detracts from the whole point of the installer IMHO. It's meant to make things simple, not more complicated. I think at most you might want to have a screen during setup that mentions that some people may have some issues, and point them to the readme file for more information. I'm happy with things the way they are at present- i ran the installer, edited the config file to run 2 WUs at a time on the GPU & then just let it do it's thing. If people want to reserve a Core or tweak other settings to get another .5% boost in throughput then they can go for it. But i wouldn't have those options as part of the installer. Just tell people at the start (and end for those with short memories) of the installation to read the readme file for known issues, and their possible work arounds. Grant Darwin NT |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
How many people have posted in this thread? Without counting, I'd guess at fewer than 20. What do the other 660 think? Thanks Grant, That's the original direction I was coming from. I'll remain avoiding making anything needlessly complex for standard running. Cheers, 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. |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
I forgot something.. A big 'Thank you!' to all who are in the Lunatics crew - and who are involved the we have the Lunatics Installers! Just a few ideas of one S@h member.. It's possible to extend the CUDA app with cmdline settings, so the members could in-/decrease the priority himself? It's possible to make a bench-test tool, a very easy for noobs like me, one click and the program say which app (CPU extension usage) is the best/fastest for the machine? For S@h Enhanced (MultiBeam) and Astropulse apps? This would be very helpful and nice.. Thanks a lot! I just wanted to make a bench-test for to see which AP 6.01 app (r555 vs. r557) is faster on my machine, but I failed.. - Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. - |
Mark Wyzenbeek Send message Joined: 28 Jun 99 Posts: 134 Credit: 6,203,079 RAC: 0 |
How many people have posted in this thread? Without counting, I'd guess at fewer than 20. What do the other 660 think? I ran the installer without trouble. I've finally downloaded a couple AP workunits but haven't started crunching them yet. I don't have a fancy GPU to use. Thanks for the update! The Universe is not only stranger than you imagine, it's stranger than you can imagine. SETI@home classic workunits 1,405 CPU time 57,318 hours |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
I'm happy with things the way they are at present- i ran the installer, edited the config file to run 2 WUs at a time on the GPU & then just let it do it's thing. If people want to reserve a Core or tweak other settings to get another .5% boost in throughput then they can go for it. But i wouldn't have those options as part of the installer. I agree with the part about not making the installation more complex, but I still think that relying on process lasso or any 3rd party tool to make one app to work almost twice faster dosnt sounds good... The optional command line to be edited manually (or an enviroment variable or whatever inside BOINC/SETI apps) sounds better from (my) user's point of view... About few people downloading this, In my case Ive dowloaded it for one new host devoted exclusively for Seti in which I dont care what it does as long as it is crunching but, Im not running the CPU AP apps in the other hosts cause is not efficient (for my taste, of course) so I might not have need of this new installation... I guess there is a lot of people in the same situation, not needing/wanting the CPU AP app... Again, Im just sharing my thought not saying what Lunatics should do... |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
I'm happy with things the way they are at present- i ran the installer, edited the config file to run 2 WUs at a time on the GPU & then just let it do it's thing. If people want to reserve a Core or tweak other settings to get another .5% boost in throughput then they can go for it. But i wouldn't have those options as part of the installer. Thanks, It'll all be considered for next release (Kepler GPU refresh + performance updates), taking into account that some users appear to need added functionality. "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. |
W-K 666 Send message Joined: 18 May 99 Posts: 19012 Credit: 40,757,560 RAC: 67 |
How many people have posted in this thread? Without counting, I'd guess at fewer than 20. What do the other 660 think? As one of the others, I report that on old h/ware, q6600 & gt250, the installation went without incident and the computer is crunching as expected. AP tasks have been hard to grab but now have a couple that should start later today, or early tomorrow. Think you should take the opinion that "no news, is good news". People don't usually hesitate to complain. |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
Being someone who just runs CPU, I love the installer. And since I run whatever is thrown my way on my 2 computers, I updated to this latest installer. I don't really have time or money to deal with GPU's right now, I'd like to, but just don't have the time. Maybe next year? LOL Thank you to the Lunatic's Crew for everything you guys and gals have done. Its outstanding work, has made installing these optimized app's so much easier even the non-techie folks can do it. I have always admired that we have some of the most dedicated folks here on this project that enjoy helping others, and not getting frustrated even when questions are posed more than once. I'm sure once things calm down for me, I'll be one of those who asks. There is a lot of information on these boards, and instead of trying to sift through it all, its easier to get it answered the right way without having to dig through so much info. |
LadyL Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 0 |
Seeing the thread exploded over night, if I don't get to your question, please don't hesitate to repeat it. @Micheal Miles - The NVidia OpenCL apps run on 270+ drivers but hog a CPU core. You'll either have overcommit or you can do a workaround and free one core - I think Joe posted how to do that. I'm not the Pope. I don't speak Ex Cathedra! |
LadyL Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 0 |
Will those work to fix this on the ATI client as well? I don't understand why half of my WUs run normally and half of them crap out. James - we just deliver the parcel. That's a question to put to Raistmer and ask him to look into and possibly fix. I'm not the Pope. I don't speak Ex Cathedra! |
LadyL Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 0 |
A couple of comments on freeing one core, as a refinement but not as a recommendation. Joe are you sure it's 1 and not >1? Besides the float representation may lead to values slightly smaller than 1 when adding up... I'd rather play safe and add another % of fraction. I'm not the Pope. I don't speak Ex Cathedra! |
LadyL Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 0 |
It's possible to extend the CUDA app with cmdline settings, so the members could in-/decrease the priority himself? Everything is possible. Whether it's practical is a differnt question and whether I sanction it... IMO the less knobs the better. There is such a thing as too many tuning options. PnP - not endless fumbling to find another half % of speed. It's possible to make a bench-test tool, a very easy for noobs like me, one click and the program say which app (CPU extension usage) is the best/fastest for the machine? Sutaru, you've been running benches for ages. If somebody has too much time, they are very welcome to take our test WUs and our benching scripts and write a nice colourful program, that does the thinking for you. Anyway, as far as I know, we are on our way to get rid of that bit. Would certainly make my life easier. I just wanted to make a bench-test for to see which AP 6.01 app (r555 vs. r557) is faster on my machine, but I failed.. http://lunatics.kwsn.net/index.php?module=Downloads;catd=44 looks like I haven't gotten around to put the AP bench online. Ok, now. unzip preserving folder strcuture, add apps, add WUs - available separately. I'm not the Pope. I don't speak Ex Cathedra! |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
In the last 24hrs I have noticed that the cpu scheduler has been preempting the GPU tasks and running CPU tasks at high-priority. The machine (A-SYS) is an i7/750/6Gb/ram/Win7 Ultimate/64-bit/EVGA GTX460SE/EVGA GTS250, BOINC 7.0.22, Lunatics 0.40. All app_info.xml settings are default except the GPU count. Machine is running at 85% CPU (6 cores), leaving 2 cores for GPU processing at 2 tasks each. 29-Mar-2012 00:17:37 [SETI@home] [cpu_sched] Preempting 07jn11ad.2890.1299.14.10.112_1 (removed from memory) 29-Mar-2012 00:17:37 [SETI@home] [cpu_sched] Preempting 15my11aa.14592.481.6.10.229_0 (removed from memory) 29-Mar-2012 00:17:37 [SETI@home] [cpu_sched] Preempting 15my11aa.7272.1708.9.10.247_1 (removed from memory) 29-Mar-2012 00:17:37 [SETI@home] [cpu_sched] Preempting 07jn11ad.2890.1299.14.10.113_0 (removed from memory) 29-Mar-2012 00:17:37 [SETI@home] [cpu_sched] Preempting 17jn11ac.25044.322838.4.10.148_2 (removed from memory) was the last time that preempting occurred and is still in effect. The CPU tasks started doing this with deadlines of 10 April and estimated run times for them are approx 30 minutes each. It is now working on ones for 11 April. I exited the BM, stopping all work and restarted Boinc with nothing changing. Also recycled the machine with the same results. I also have 8 AP6 6.01 (6 d/l'ed on 27 March, 2 on 28 March) with expected deadline of 21/22 April. When the first of these tasks came in they had a estimated running time of appro 120 hours each. During the last two days I have noticed the estimated run time go from 120 to 71 to 172 hours. Questions: 1) Why are the GPU tasks being preempted when this did not occur in Lunatics 0.39? 2) Is the scheduler just clearing out the machine as fast as possible to allow room for the AP6 tasks? 3) Do I have a major problem here? I don't want to regress to Lunatics 0.39 because of the new AP6 units. 4) Do I need to test the new _41x? I don't buy computers, I build them!! |
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 |
I get the same all the time when my DCF jumps after a slow GPU finishes. What is your current DCF? You may wish to update cc_config.xml to show the DCF changes. <cc_config> <log_flags> <dcf_debug>1</dcf_debug> |
tbret Send message Joined: 28 May 99 Posts: 3380 Credit: 296,162,071 RAC: 40 |
In the last 24hrs I have noticed that the cpu scheduler has been preempting the GPU tasks and running CPU tasks at high-priority. The machine (A-SYS) is an i7/750/6Gb/ram/Win7 Ultimate/64-bit/EVGA GTX460SE/EVGA GTS250, BOINC 7.0.22, Lunatics 0.40. All app_info.xml settings are default except the GPU count. Machine is running at 85% CPU (6 cores), leaving 2 cores for GPU processing at 2 tasks each. Must be something to do with your specific situation. My eight-banger AMD is running 4 cores only, but two nVidia cards two at a time with no interference. Like you, I'm looking forward to seeing what AVX does with AP v6 when I finally get to them. |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
I get the same all the time when my DCF jumps after a slow GPU finishes. What is your current DCF? You may wish to update cc_config.xml to show the DCF changes. Modified the cc_config.xml and reread it. 03/29/2012 09:06:07 | | Re-reading cc_config.xml 03/29/2012 09:06:07 | | Config: use all coprocessors 03/29/2012 09:06:07 | Milkyway@Home | Config: excluded GPU. Type: all. App: milkyway. Device: 1 03/29/2012 09:06:07 | | log flags: file_xfer, sched_ops, task, cpu_sched, dcf_debug, sched_op_debug 03/29/2012 09:06:11 | SETI@home | Computation for task 06jn11ab.31414.9883.7.10.59_0 finished 03/29/2012 09:06:11 | SETI@home | [dcf] DCF: 1.006398->1.006192, raw_ratio 1.004339, adj_ratio 0.997954 03/29/2012 09:06:11 | SETI@home | Starting task 20my11ad.5048.16018.13.10.110_0 using setiathome_enhanced version 603 in slot 2 03/29/2012 09:06:13 | SETI@home | Started upload of 06jn11ab.31414.9883.7.10.59_0_0 03/29/2012 09:06:23 | SETI@home | Computation for task 06jn11aa.24762.11110.3.10.232_0 finished 03/29/2012 09:06:23 | SETI@home | [dcf] DCF: 1.006192->1.005995, raw_ratio 1.004220, adj_ratio 0.998039 03/29/2012 09:06:23 | SETI@home | Starting task 24my11af.14050.16427.3.10.118_0 using setiathome_enhanced version 603 in slot 7 I don't buy computers, I build them!! |
©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.