Message boards :
Number crunching :
Preferences advice needed for Celeron 1.4 ghz.
Message board moderation
Author | Message |
---|---|
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
Hi All, Here is my messages log: 1/27/2006 2:32:38 AM||Starting BOINC client version 5.2.13 for windows_intelx86 1/27/2006 2:32:38 AM||libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3 1/27/2006 2:32:38 AM||Data directory: C:\\Program Files\\BOINC 1/27/2006 2:32:39 AM||Processor: 1 GenuineIntel Intel(R) Celeron(TM) CPU 1400MHz 1/27/2006 2:32:39 AM||Memory: 382.52 MB physical, 728.66 MB virtual 1/27/2006 2:32:39 AM||Disk: 93.15 GB total, 15.12 GB free 1/27/2006 2:32:39 AM|rosetta@home|Computer ID: 78725; location: home; project prefs: default 1/27/2006 2:32:39 AM|climateprediction.net|Computer ID: 273629; location: ; project prefs: default 1/27/2006 2:32:39 AM|Einstein@Home|Computer ID: 457198; location: home; project prefs: default 1/27/2006 2:32:39 AM|Predictor @ Home|Computer ID: 183422; location: home; project prefs: default 1/27/2006 2:32:39 AM|SETI@home|Computer ID: 1817497; location: home; project prefs: default 1/27/2006 2:32:39 AM||General prefs: from SETI@home (last modified 2006-01-25 02:48:58) 1/27/2006 2:32:39 AM||General prefs: no separate prefs for home; using your defaults 1/27/2006 2:32:41 AM||Remote control not allowed; using loopback address 1/27/2006 2:32:41 AM|climateprediction.net|Deferring computation for result sulphur_hz7b_000838775_0 1/27/2006 2:32:41 AM|SETI@home|Resuming computation for result 20se04aa.18132.19377.23586.1.2_0 using setiathome version 418 1/27/2006 2:32:41 AM||Suspending work fetch because computer is overcommitted. 1/27/2006 2:32:41 AM||Using earliest-deadline-first scheduling because computer is overcommitted. Is overcommitted bad ? Here are my BOINC preferences: Processor usage Do work while computer is running on batteries? (matters only for portable computers) no Do work while computer is in use? yes Do work only between the hours of (no restriction) Leave applications in memory while preempted? (suspended applications will consume swap space if 'yes') yes Switch between applications every (recommended: 60 minutes) 150 minutes On multiprocessors, use at most 1 processors Disk and memory usage Use no more than 10 GB disk space Leave at least (Values smaller than 0.001 are ignored) 0.3 GB disk space free Use no more than 75% of total disk space Write to disk at most every 60 seconds Use no more than 75% of total virtual memory Network usage Connect to network about every (determines size of work cache; maximum 10 days) 2 days Confirm before connecting to Internet? (matters only if you have a modem, ISDN or VPN connection) no Disconnect when done? (matters only if you have a modem, ISDN or VPN connection) no Maximum download rate: no limit Maximum upload rate: no limit Use network only between the hours of Enforced by versions 4.46 and greater (no restriction) Skip image file verification? Check this ONLY if your Internet provider modifies image files (UMTS does this, for example). Skipping verification reduces the security of BOINC. no I have 15 gigs free on my 100 gig hard drive. I am running 5 BOINC projects. SETI is currently using 6.3MB of space. All the other projects are larger. CP is the largest at 292.34MB. This is my first post. Are these settings optimized ? Jonathan |
web03 Send message Joined: 13 Feb 01 Posts: 355 Credit: 719,156 RAC: 0 |
EDF isn't necessarily a bad thing. Basically the BOINCMGR is letting you know that you have more work on hand than is potentially necessary. And based upon your connect time, your computer may not be able to complete everything before your next connect (which you have set at 2 days). What are your resource shares set to? I see that you have 5 projects attached on one machine. I personally have 6 projects attached to one machine at a time. Here's what my shares look like: Rosetta 100 (13.79%) Climate 200 (27.59%) Einstein 200 (27.59%) Pirates 100 (13.79%) SETI 25 (3.45%) SETI BETA 100 (13.79%) If you have recently just set up BOINC and/or changed your settings, it could take a few days for the scheduler to settle down and get you into a Round Robin mode. I wouldn't overly stress for now. Wendy Click Here for BOINC FAQ Service |
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
What are your resource shares set to? If you have recently just set up BOINC and/or changed your settings, it could take a few days for the scheduler to settle down and get you into a Round Robin mode. I wouldn't overly stress for now. Thank you for the reply. I actually did, thankfully, not overly stress about it, but I was sorta worried about causing undue server use on the various projects. (^: Here are my resource settings : rosetta - 50 climateprediction - 10 Einstein - 50 Predictor - 50 SETI - 50 I'm not sure why I chose those settings, but they seemed like what other people were doing with similar machines to mine. I don't want to timeout on getting WUs done as that would slightly stress out the projects as well. (^: The latest changes I have been looking at were the # of days I connect to servers. I tried 8 days and it seemed too much, and now I tried 2 days. I was hoping the messages would give me details on if my tweaking was working more effeciently, but "overcommitted" seemed slightly ominous, but I had the feeling that it wasn't really. I guess it is hard to tell between messages that are BOINC talking to itself and BOINC talking or being talked to by other project servers ? Jonathan |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
One thing that I have seen across the boards is this: If your just going to run 1 project, its ok to have a higher connection setting. However, the more projects you run, you want to lower that connection setting. Myself, I am connected to 7 projects, and have my connection setting set for .5 days It gives me roughly 2 WU's per project, except for SIMAP, which usually gives me about 4 WU's, since they crunch pretty quickly, and CPDN, just 1 WU, and SETI BETA only 1 WU. Keeps my computer plenty happy. And all my resource shares are the same. I just left them with the default 100 for each project. I have my "Switch project every xx minutes" set at 120 minutes, and I leave the application in memory, due to Rosetta. I got 1gb RAM, so this isn't much of an issue. I hope this helps you out a bit. Just remember, if your only gonna run 1 project, your ok to run a higher connect setting. I think a common number was 7 or 8 days. But if your gonna run more than one, I would lower it by half for each project you attach to. Take care and happy crunching! Jeremy |
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
One thing that I have seen across the boards is this: Thank you for the tips ! I am tweaking it back down to .5 and see how it affects things and I'll check on it again in a few days. peace, Jonathan |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
Your welcome. When coming back down from a high setting, it takes a while for it to kick in like its supposed to. I'd at least give it about a week or so to allow it to "Get into the grove" again. Good luck and let us know how that works for you. Jeremy |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
Here are my resource settings : Be aware that the numbers are not percentages. Yours works out to 23% for each of the 8 and about 5% for CPDN. I use actual numbers so that my shares would be 23, 23, 23, 23, 8 ... that way you don't have to do the math ... :) If you have more than one computer to do it "right" is a little more involved. But, with a spreadsheet is not that bad. My example is in Apple workds so will not do most people any good, and I just blew it up anyway with an upgrade to the slowest computer. But, the concept is pretty simple, figure out the rough capability of each machine and then calculate what a share works out to be over all. If anyone is interested I can post an example here or on the web somehow ... won't waste my time otherwise ... |
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
Here are my resource settings : Yep, the nice thing about BOINC 5.2.13 (and perhaps earlier versions) is that it automatically figures the percentages right on the projects tab. I only have one computer crunching at the moment, so I have no farm to calculate the percentages. Hmmm.... I guess I have a "house plant" instead of a "farm". (^: I have two other computers, but they refuse to network with each other. Someday someone smart will invent plug and play networking. (^: Jonathan |
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
Your welcome. When coming back down from a high setting, it takes a while for it to kick in like its supposed to. I'd at least give it about a week or so to allow it to "Get into the grove" again. Good luck and let us know how that works for you. Rodger that Jeremy. You make it sound like the algorithms are dancing and by tweaking I keep changing the beat. (^: Jonathan |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
Your welcome. When coming back down from a high setting, it takes a while for it to kick in like its supposed to. I'd at least give it about a week or so to allow it to "Get into the grove" again. Good luck and let us know how that works for you. Hehe yeah, thats a good way to explain it. Keep on crunching. Your preferences are more like a "Set It And Forget It" type of situation. Just Keep an eye on things. Don't continually hit the "Update" button, let it take its time to get into the swing of things. I'm sure you will work out just fine. Reading through all these boards have really helped. And reading stuff that I had questions about from the WIKI that Paul D Buck and associates have worked very hard on is a great source of info. You can find a link not only in my Sig, but in many many others. Check it out, browse through it. I think its the best "How To" to get things going that anyone should be able to get BOINC running on their machines. Take care. Jeremy |
MJKelleher Send message Joined: 1 Jul 99 Posts: 2048 Credit: 1,575,401 RAC: 0 |
If you have more than one computer to do it "right" is a little more involved. But, with a spreadsheet is not that bad. My example is in Apple workds so will not do most people any good, and I just blew it up anyway with an upgrade to the slowest computer. But, the concept is pretty simple, figure out the rough capability of each machine and then calculate what a share works out to be over all.I had to think a bit about this part, let me know if I'm getting you wrong... Your calculations are to figure out resourse share across your entire farm, so the percentages work out in general. The settings we enter work on a per host basis, so you want to make your adjustments based on the individual computers' capabilities. That about right? With my very different and only two computers, I just have the shares set for each host, and let the totals work out as they may. Two projects at 60/40 for the slow one, and a range of shares for the 8 projects (with two that don't always have work) on the main one. It's all good. 8-) MJ MJ |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
Your calculations are to figure out resourse share across your entire farm, so the percentages work out in general. The settings we enter work on a per host basis, so you want to make your adjustments based on the individual computers' capabilities. Yes... :) I start with the CS/Hr numbers as: System CS/Hour CPUs Percentage EQ-1 12.89 2 0.08088604418 EQ-2 10.46 2 0.0656375502 G5a 19.34 2 0.12136044177 P4a 12.96 2 0.0813253012 P4b 10.39 2 0.06519829317 T6522 12.44 1 0.078062249 A4400-x2a 25.79 2 0.16183483936 Xeon-32a 27.34 4 0.17156124498 Xeon-64a 27.75 4 0.17413403614 Total: 159.36 21 1.00 Which shows the theoretical production across each of the systems. Then, if we group the systems into 3 venues, in my case default, school, and work ... System Default School Work EQ-1 0.0808 EQ-2 0.0656 G5a 0.1214 P4a 0.0813 P4b 0.0651 T6522 0.0781 A4400-x2a 0.1618 Xeon-32a 0.1716 Xeon-64a 0.1741 Total: 0.532 0.3457 0.1214 Basically showing that the totality of the computers in the default group will do 53% of my total work load. The G5 in its lonesome (at the moment) will only do 12% ... the two Xeons do 34% With THAT ... we now can do our matrix table ... Projects Default School Work Effective Rs. Share CPDN 20 25 0 19.30126757 Einsten@Home 25 22 55 27.60372741 LHC@Home 0 0 0 0 Predictor@Home 0 0 0 0 Prime Grid 0 3 0 1.0370858434 Rosetta@Home 15 5 30 13.363453815 SETI@Home 5 5 10 5.6068022088 SIMAP 10 15 0 10.514871988 SZTAKI 5 5 5 5 WCG 20 20 0 17.572791165 100 100 100 100 So, in this example, by assigning these resource shares and placing the computers in the right venues, I can look at the effective column and see how the actual shares work out in practice. I did not add the formulas but they are straight forward math ... you can see for example, by putting the Xeons on PrimeGrid with 3% share, my effective is 1% of my total contribution. If I cahnge my Einstein@Home on the G5 to 100%, my effective Einstein@Home will become 33% of my totals ... This allows slightly finer control, mostly I use it to "balance" how the projects stand in my total contributions. So, Einstein@Home has had my focus for some time and it is getting a lot of attention. I had been also concentrating on WCG to move my "standing" in the project up. Same with SIMAP@Home which just moved into production status for both Windows and Linux ... LHC@Home alters this somewhat, but, I have not made modifications to account for this ... I just thought of a way I might be able to do that and may "fiddle" my calculations again. Anyway, the point is, for those of us with multiple platforms but who do want to emphasize some projects over others, this "system" allows better visibility ... for example, PrimeGrid is still in testing, and does not have much of a point, much like Pirates@Home, so, though I give them some time, I don't want it to be very much of my totals. 1 or 2% at most. Of course, with faster computers and more CPUs, 1% starts to become a fair amount of processing ... as you can see, my current production rate is 159 CS/hr, which is not fantastic, but is not bad. If I can complete my upgrades this year I hope to be looking at 200-240 CS/hr by the end of the year (and about $10,000) ... Hope this makes it clearer ... but, you did have it right ... |
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
When coming back down from a high setting, it takes a while for it to kick in like its supposed to. I'd at least give it about a week or so to allow it to "Get into the grove" again. Good luck and let us know how that works for you. Hi Jeremy, It has been over a week now with my new .5 days settings change for how often BOINC connects to send in WUs and I am getting odd results. I suppose BOINC is still getting in the groove, but anyway, I'll just post what I see and see if anyone has any suggestions or if I should keep it the same, then that is ok as well. (^: Since moving from 8 days to .5 days, Predictor, Rosetta, & Einstein have really loved the new settings. CP is still at 1.26% done so far. I guess this seems ok since it is due in December of 2006. But I have seen this due date change on me, so I am worried that CP will never get any action or at least it might hog my computer for all of November and December to get CP's WUs done on time. (^: SETI and Einstein have disappeared from my work tab. I heard SETI was down, so I guess that accounts for that. So putting two and two together I am assuming that Predictor, Rosetta, & Einstein are each using my unused cpus for half a day and then they sorta trade off with each other. At least that is what the statistics look like to me. A few odd things have happened lately. If I leave BOINC on for 2 days, then it seems to not release control of cpu power very easily or at all sometimes. Sometimes I will try to send out email with Netscape and it will timeout and it cannot send. So I basically put that email into a draft folder and exit Netscape Email and restart it, and then I can send email out. In other oddness, my computer sometimes gives 100% of cpus to BOINC and I can't seem to start any program without restarting the computer. Luckily it is not a soft or cold reboot. (^: I guess I am to assume that each program I have does indeed have a priority setting and a few of my programs are duking it out or perhaps two fat32 programs are trying to get through the same thin doorway. Slapstick comedy ! Anyway, to sum up, I don't run BOINC all the time. I run it when I am messing with photos and some programs, but shut it off while I do email, but only if BOINC has been running for two of more days straight. And I run BOINC before heading to sleep so it will run all night. (^: Jonathan |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
Hi Jon. When backing off on your settings, your computer is used to having the higher "cache" setting. By lowering this, you won't download as many WU's. What it sounds like to me is that your projects that are not downloading WU's, (Seti and Einstein), I'm guessing you have built up a pretty good sized "Long Term Debt". In the BOINC WIKI (found in my signature below) it talks in pretty good detail about how the LTD and STD works. In a nutshell, the projects are giving the time to the ones that have a lower LTD. As all the numbers even out more, you'll start getting WU's again for the other projects. Sometimes it takes a while to get things working right. Maybe another couple of days to a week or so. Also, your "Resource Share" plays a factor too. All of my projects get equal time. I suck with numbers, so I just left them all at the default 100, and I switch applications every 120 minutes. So each of the 7 projects gets 2 hours at a time, or there abouts. I got a CPDN WU as well, its currently at 5.95%, 133:39 hours done, 942:20 hours left to go. Check back on it in another couple of days and see how its working. |
Keck_Komputers Send message Joined: 4 Jul 99 Posts: 1575 Credit: 4,152,111 RAC: 1 |
When reducing the queue size expect it to take at least double the old queue size before the client gets into a groove. So in this case that is a minimum of 16 days for things to get to normal. BOINC WIKI BOINCing since 2002/12/8 |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
Thanks Keck. I couldn't remember how long it would take. I never had mine set so high. :-) @ Jon, Let us know in about another week and a half or so, and see how its working then. It may still need a bit of time after that to get into the swing of things. Let us know how it goes. Jeremy |
Jon C Melusky Send message Joined: 13 Dec 00 Posts: 13 Credit: 1,611,366 RAC: 0 |
@ Jon, Hi Jeremy, Things are back to normal and working well. No more over committed messages now. Predictor acts up from time to time, but I won't post here about it. Jon |
©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.