dumb? CUDA question
Questions and Answers :
GPU applications :
dumb? CUDA question
Message board moderation
| Author | Message |
|---|---|
|
Steve Robertson Send message Joined: 14 May 99 Posts: 38 Credit: 2,643,210 RAC: 0 
|
I've been running SETI since it was brand new but it's always been a background thing so forgive my stupid newbie question- I know nothing about how it operates. I just upgraded to the CUDA version and according to this, it is apparently working: (?) (scroll down for my question) 12/31/2008 10:19:11 AM||Starting BOINC client version 6.4.5 for windows_intelx86 12/31/2008 10:19:11 AM||log flags: task, file_xfer, sched_ops 12/31/2008 10:19:11 AM||Libraries: libcurl/7.19.0 OpenSSL/0.9.8i zlib/1.2.3 12/31/2008 10:19:11 AM||Data directory: C:\ProgramData\BOINC 12/31/2008 10:19:11 AM||Running under account SLAPMASTER2 12/31/2008 10:19:13 AM||Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4200+ [x86 Family 15 Model 107 Stepping 1] 12/31/2008 10:19:13 AM||Processor features: fpu tsc pae nx sse sse2 pni 3dnow mmx 12/31/2008 10:19:13 AM||OS: Microsoft Windows Vista: Home Premium x86 Editon, Service Pack 1, (06.00.6001.00) 12/31/2008 10:19:13 AM||Memory: 2.00 GB physical, 4.23 GB virtual 12/31/2008 10:19:13 AM||Disk: 325.87 GB total, 83.73 GB free 12/31/2008 10:19:13 AM||Local time is UTC -9 hours 12/31/2008 10:19:13 AM||Not using a proxy 12/31/2008 10:19:14 AM||CUDA devices found 12/31/2008 10:19:14 AM||Coprocessor: GeForce 8600 GT (1) 12/31/2008 10:19:14 AM||Version change (5.10.45 -> 6.4.5) 12/31/2008 10:19:14 AM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 4256635; location: home; project prefs: default 12/31/2008 10:19:14 AM||General prefs: from SETI@home (last modified 26-Feb-2005 10:44:48) 12/31/2008 10:19:14 AM||Computer location: home 12/31/2008 10:19:14 AM||General prefs: no separate prefs for home; using your defaults 12/31/2008 10:19:14 AM||Reading preferences override file 12/31/2008 10:19:14 AM||Preferences limit memory usage when active to 1022.88MB 12/31/2008 10:19:14 AM||Preferences limit memory usage when idle to 1841.19MB 12/31/2008 10:19:14 AM||Preferences limit disk usage to 10.00GB 12/31/2008 10:19:14 AM||Running CPU benchmarks 12/31/2008 10:19:14 AM||Suspending computation - running CPU benchmarks 12/31/2008 10:19:46 AM||Benchmark results: 12/31/2008 10:19:46 AM|| Number of CPUs: 2 12/31/2008 10:19:46 AM|| 2013 floating point MIPS (Whetstone) per CPU 12/31/2008 10:19:46 AM|| 3622 integer MIPS (Dhrystone) per CPU 12/31/2008 10:19:47 AM|SETI@home|Restarting task ap_03dc08ah_B2_P0_00192_20081229_11334.wu_1 using astropulse version 500 12/31/2008 10:19:49 AM|SETI@home|Restarting task 23no08ae.25650.2117.9.8.86_1 using setiathome_enhanced version 603 12/31/2008 10:53:16 AM||General prefs: from SETI@home (last modified 26-Feb-2005 10:44:48) 12/31/2008 10:53:16 AM||Computer location: home 12/31/2008 10:53:16 AM||General prefs: no separate prefs for home; using your defaults 12/31/2008 10:53:16 AM||Reading preferences override file 12/31/2008 10:53:16 AM||Preferences limit memory usage when active to 1022.88MB 12/31/2008 10:53:16 AM||Preferences limit memory usage when idle to 1841.19MB 12/31/2008 10:53:16 AM||Preferences limit disk usage to 10.00GB 12/31/2008 10:59:05 AM||Running CPU benchmarks 12/31/2008 10:59:05 AM||Suspending computation - running CPU benchmarks 12/31/2008 10:59:36 AM||Benchmark results: 12/31/2008 10:59:36 AM|| Number of CPUs: 2 12/31/2008 10:59:36 AM|| 2146 floating point MIPS (Whetstone) per CPU 12/31/2008 10:59:36 AM|| 3943 integer MIPS (Dhrystone) per CPU 12/31/2008 10:59:37 AM||Resuming computation 12/31/2008 11:23:09 AM||Suspending computation - user request 12/31/2008 11:32:42 AM||Resuming computation 12/31/2008 11:58:13 AM|SETI@home|Computation for task 23no08ae.25650.2117.9.8.86_1 finished 12/31/2008 11:58:13 AM|SETI@home|Starting ap_19no08ac_B4_P0_00059_20081231_07119.wu_0 12/31/2008 11:58:13 AM|SETI@home|Starting task ap_19no08ac_B4_P0_00059_20081231_07119.wu_0 using astropulse version 500 12/31/2008 11:58:15 AM|SETI@home|Started upload of 23no08ae.25650.2117.9.8.86_1_0 12/31/2008 11:58:18 AM|SETI@home|Finished upload of 23no08ae.25650.2117.9.8.86_1_0 12/31/2008 12:26:47 PM|SETI@home|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 1 completed tasks 12/31/2008 12:26:52 PM|SETI@home|Scheduler request completed: got 0 new tasks 12/31/2008 12:26:52 PM||General prefs: from SETI@home (last modified 31-Dec-2008 12:24:22) 12/31/2008 12:26:52 PM||Computer location: home 12/31/2008 12:26:52 PM||General prefs: no separate prefs for home; using your defaults 12/31/2008 12:26:52 PM||Reading preferences override file 12/31/2008 12:26:52 PM||Preferences limit memory usage when active to 1022.88MB 12/31/2008 12:26:52 PM||Preferences limit memory usage when idle to 1841.19MB 12/31/2008 12:26:52 PM||Preferences limit disk usage to 10.00GB 12/31/2008 12:27:07 PM|SETI@home|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 0 completed tasks 12/31/2008 12:27:12 PM|SETI@home|Scheduler request completed: got 0 new tasks I have a dual core system.... is it supposed to be running three things now that the CUDA is enabled? Or does the CUDA speed things up? I don't understand. Again, sorry for stupid newbie question. ID: 847467 ·  |
|
Maik Send message Joined: 15 May 99 Posts: 163 Credit: 9,208,555 RAC: 0 
|
The Cuda application is using your GPU (grafic prozessing unit) to calculate WorkUnits. It's additional to your two cores. If you don't want that the BoincManager is doing so, go to your Account-Prefs. and set it up there. ID: 847470 · |
 Dr. C.E.T.I.  Send message Joined: 29 Feb 00 Posts: 16019 Credit: 794,685 RAC: 0
|
. . . Steve - read what Paul D Buck has to say here: Message boards : Number crunching : CUDA and Resource Share <--- click me > have a great New year . . .  BOINC Wiki . . .Science Status Page . . . ID: 847493 · |
|
Steve Robertson Send message Joined: 14 May 99 Posts: 38 Credit: 2,643,210 RAC: 0
|
Resource share isn't an issue- SETI is the only thing I am running. ID: 847498 · |
MarkJ     Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 
|
I've been running SETI since it was brand new but it's always been a background thing so forgive my stupid newbie question- I know nothing about how it operates. From the above log: 1. You have just upgraded BOINC to 6.4.5 (from 5.10.45) 2. You are using the 6.03 app which is the standard (cpu) version of the seti multibeam application It should download the 6.06 app when it requests new work and then start running as 1+1 if it has some cuda work units. The gpu will process seti multibeam work units by running a "feeder" task and the other core can only do astropulse, because you are only attached to one project (seti). It won't by default run three tasks. There is something you can do to trick BOINC 6.4.5 to run three but it is a bit more advanced. BOINC 6.5.0 fixes that but introduced a bunch of other bugs. Have a read of the other cuda message threads and you'll see what I mean. BOINC blog ID: 847549 · |
|
Bambi Send message Joined: 15 May 99 Posts: 26 Credit: 5,704,701 RAC: 1 
|
You can create/edit a cc_config.xml file in your boinc data directory to make Boinc use all your cores AND your gpu, as it seems to default to all your cores minus 1. //Contents of cc_config.xml <cc_config> <options> <ncpus>5</ncpus> </options> </cc_config> //This uses all 4 of my cores plus the gpu. Once downside to this is, if your machine runs out of CUDA work units Boinc will still try to run 5 tasks, resulting in 1 core running 2 tasks. It does also seem to make my machine a little less responsive running 5 jobs. Bambi 
ID: 847745 · |
|
MeglaW Send message Joined: 21 Jun 00 Posts: 36 Credit: 479,460 RAC: 0 
|
i disagree with modding your client to run 2+1 (3) or 4+1 (5) on cc_config.xml, it makes it more then half as efficient at running the cuda workunits 
ID: 848230 · |
|
Maik Send message Joined: 15 May 99 Posts: 163 Credit: 9,208,555 RAC: 0
|
agree, i set my host back from 4+1 to 4 ID: 848235 · |
|
Steve Robertson Send message Joined: 14 May 99 Posts: 38 Credit: 2,643,210 RAC: 0
|
Maybe I didn't make myself appear dumb or "newbie" enough. I'm not going full retard here, just partial retard. Like Forest Gump. Sure he was slow, but what about the ping-pong bit and the olympics? Yur right, I just saw Tropic Thunder. But I digress... What is going to happen in 75 hours when the next work unit is done and SETI phones home? I'm currently running two astropulse units. Not to be an ass, but please phrase an answer as if I just told you that life was like a box of chocolates. Please don't mistake the fact that I've been running SETI longer than all of you (well, most of you anyway) for that I know what I'm doing. I really don't. Don't mistake age for wisdom. Notice the ET reference above. Does anybody even watch that movie anymore? Life = Chocolates. Well, actually Life > = chocolates, but that doesn't really take the shrimp into consideration. Can anybody help me? What is going to happen when the next work unit is done and SETI gathers the next? ID: 848450 · |
|
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0
|
... Assuming the Scheduler sends at least one setiathome_enhanced task, it will also tell your BOINC core client that it should download the 6.06 science application, the two CUDA DLLs which go with it, and various image and GPL related files. The core client will delete the 6.03 application if all the previously downloaded work for it has been completed. After the new WU and all required files are downloaded it will start the work with 6.06 when a CPU core is available. Joe ID: 848474 · |
|
Steve Robertson Send message Joined: 14 May 99 Posts: 38 Credit: 2,643,210 RAC: 0
|
That is the most awesome answer ever. ID: 848760 · |
OzzFan  Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28
 |
Does that mean it was a good answer, or was that sarcasm denoting that you still don't understand? ID: 848916 · |
|
Steve Robertson Send message Joined: 14 May 99 Posts: 38 Credit: 2,643,210 RAC: 0
|
My comment was sincerely sarcasm-free. I may have to relax by playing some ping-pong before I completely understand the answer. Does that answer your question? ID: 849136 · |
|
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28
|
Yes. lol ID: 849227 · |
|
Slow_Target Send message Joined: 5 Oct 02 Posts: 58 Credit: 6,704,641 RAC: 2
|
... I believe I understand, However.... I originally loaded cuda drivers and 6.4.5. I could not get cuda to run but eventually it did and downloaded 6.05. After many errors I decided to stop cuda. On the next server contact it downloaded 6.03 again. Installed newer (beta) drivers, enabled cuda and it downloaded 6.06. Things were better but still getting too many errors. I disabled cuda and it downloaded 6.03 again. Example after a restart: 12/29/2008 4:26:12 PM|SETI@home|Restarting task ap_31ja08aa_B1_P1_00376_20081211_01925.wu_2 using astropulse version 500 12/29/2008 4:26:14 PM|SETI@home|Restarting task 20no08aa.7238.20522.16.8.128_0 using setiathome_enhanced version 603 I enabled cuda however the server was down for maint. Decided to wait and disabled it. Next contact: 12/30/2008 7:10:20 PM|SETI@home|Started download of setiathome_6.03_windows_intelx86.exe 12/30/2008 9:56:03 PM|SETI@home|Starting task 21oc08ab.28179.20931.7.8.225_2 using setiathome_enhanced version 603 Had time to babysit and enabled cuda: 12/31/2008 8:09:53 AM|SETI@home|File setiathome_6.06_windows_intelx86__cuda.exe exists already, skipping download 12/31/2008 8:10:00 AM|SETI@home|Starting task 24no08aa.26307.20931.6.8.238_1 using setiathome_enhanced version 606 White working through the cuda it downloaded more work note it's 6.03: 12/31/2008 11:23:41 AM|SETI@home|Starting task 30oc08af.12538.4980.15.8.86_2 using setiathome_enhanced version 603 Now 6.03's all complete, 3 AP's in queue and started work on new project WCG. Next time Seti is contacted here we go again: 1/2/2009 6:24:38 PM|SETI@home|Started download of setiathome_6.03_windows_intelx86.exe It now has a single AP wu and several 6.03's. Why does it continue to get the 6.03 app and not just use the 6.06. ID: 849289 · |
|
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28
|
White working through the cuda it downloaded more work note it's 6.03: It states "starting", not "downloading", which means this must have been a task already downloaded and registered to the 6.03 app. Now 6.03's all complete, 3 AP's in queue and started work on new project WCG. Next time Seti is contacted here we go again: If any apps are registered to the 6.03 app, it will attempt to locate that file, and if it does not exist, it will attempt to download it from the servers because apps may not be backward compatible, so it always tries to crunch the WU with the registered app and nothing else. ID: 849300 · |
|
Slow_Target Send message Joined: 5 Oct 02 Posts: 58 Credit: 6,704,641 RAC: 2
|
Obviously I didn't give you all of my message list. There may have been 6.03 apps in the middle. I really can't remember. I guess your last line is the answer. I was assuming that 6.06 would be able to crunch all MB wus weather they are cuda or not. I say this because at one point all the work that I had was 3 AP's. This was during the server outage. When I did receive work, it also downloaded 6.03 even though I have 6.06 on my system. Other messages had stated that 6.03 would be deleted when you receive 6.06. Since it does that I assumed it would not be needed after that. I guess that was a bad assumtion huh. ID: 849316 · |
|
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28
|
Not having seen your situation or read the relevant messages, I can only assume that after the server outage, BOINC may have attempted to fill your work queue to ensure your CPUs remain busy, thus downloading workunits compatible with the only SETI app on your system; v6.03. Then, while downloading work, the servers notified BOINC that a new version was available, v6.06, and downloaded that, but it still had to crunch what it had registered to v6.03 with the 6.03 app. Once all v6.03 workunits have been crunched, it will be automatically deleted from your system. ID: 849327 · |
|
Slow_Target Send message Joined: 5 Oct 02 Posts: 58 Credit: 6,704,641 RAC: 2
|
I understand what you are saying. I was completely out of 6.03 tasks with only AP's to run. Everything coming in since has been marked 6.03. I set it for No New Tasks and it should run out in about 2 days, then I can get more work and see. Thanks ID: 849343 · |
|
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28
|
That got me thinking: has the v6.06 app been officially released yet? Science apps are auto-downloaded by BOINC, so if you downloaded v6.06 manually, BOINC won't know of its existence without the use of a custom built app_info.xml (known as the Anonymous platform mechanism, which breaks the auto-download capabilities of BOINC). ID: 849369 · |
©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.