Message boards :
Number crunching :
Work units processing performance has collapsed!!
Message board moderation
Author | Message |
---|---|
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
Problem stated The reported workunits average for completed computing tasks has reduced to almost next to nothing (currently less than 100/day) compared with the last couple of years when the same figure was almost 2,000/day every day. I am a long time user of Seti@home and have regularly got an average work done of around 2,000 ever since I have had this MacBook Pro. (No other BOINC projects). The Laptop computer is turned on 24x7 with a permanent connection to electricity and has total dedicated use assigned to SETI@home/BOINC when not in use, for continuous number crunching, usually 20+ hours each and every day. Configuration is otherwise solid and very stable. No other apps running just OS and SET@home usually.I removed all instances of SETI@home and also reinstalled completely but no difference? Any ideas/suggestions? Thanks! It might be helpful to share the details of my configuration. Nothing has changed for a long, long time.... MacBook Pro specification Early 2011 model 15.4 inch display (1680x1050) Storage 750Gb SATA Disk, (529Gb free) Superdrive Processor speed 2.3GHz Intel core i7 Single processor, 4 cores L2 cache 256Kb per core L3 cache 8Mb Memory 8Gb 1333MHz DDR3 (2x 4Gb chips) Graphics Intel HD Graphics 3000 512Mb GPU: AMD ATI Radeon HD 6750M (1024MB) OpenCL 1.2 OS: Darwin 14.4.0 Serial Number: C02G42H6DRJM OSX Yosemite version 10.10.4 Apple System Prefs Energy saver settings - Power Adapter Computer Sleep: Never Display sleep: +30 minutes Hard disc sleep: Never Automatic Graphics switching: Enabled BOINC Manager Version 7.4.42 wxWidgets version: 3.0.0 BOINC preferences Processor Usage Computing allowed always, while CPU in use. (no restrictions) GPU use allowed only when CPU not in use. and after 1 minute idle time Application switching every 60mins No restrictions on multi-processor (value =0) No restrictions on CPU time (value = 0) Network Usage Max download rate (no restriction, value = 0) Max Upload rate (no restriction, value = 0) Minimum work buffer 10.00 days Max additional work buffer 10.00 days Data transfers (no restriction, value = 0) NO confirmations when connecting to internet NO disconnects when done Disk and Memory Usage Disk Use at most (no GB restriction, value = 0) Leave at least (value = 0.10GB) Use at most (value = 60% of total disk space) Tasks checkpoint to disk every 36000 seconds Use at most (value = 75% of page file swap space) Memory value = 100% when computer is in use value = 100% when computer is idle No exclusive applications Disk space Used by BOINC 124.92Mb Disk space Free and available to BOINC 418.58Gb SETI@home Project properties (snapshot 12 Aug 2015 1600UTC) Master URL: http://setiathome.berkeley.edu/ User Name: John McIntyre Resource share: 100 Disk Usage: 101.76Mb Computer ID: 6123200 Suspended via GUI: No Don’t request more work: No Host location: home Tasks completed: 2377 Tasks failed: 19 Credit_User: 5097908.25 total ; 78.80 average Credit_Host: 3150692.44 total ; 78.80 average Seti@home User details UserID: 256329 Seti@home member since 13 May 2002 Computing and Credit Total Credit: 5,097,908 Recent Average credit: 78.80 Seti@home classic work units: 1,357 Seti@home classic CPU time: 24,537 hours |
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
It could be a cooling problem that is causing it to slow down so it won't burn itself up. |
woohoo Send message Joined: 30 Oct 13 Posts: 972 Credit: 165,671,404 RAC: 5 |
i don't know why it just shows a lot of aborts |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
I would abort all tasks and reset the seti Project. Or even disconnect and then reconnect the project. My guess is, the app became corrupted somehow. If you watch, do the tasks actually make any Progress, when they are running? |
rob smith Send message Joined: 7 Mar 03 Posts: 22220 Credit: 416,307,556 RAC: 380 |
Aborting will not help with the diagnosis of this problem as there will be no completion file to give hints. you currently have about 100 tasks available let some of them run to completion and then others can try and help. Please make sure you haven't inadvertently set the triggers for "other work running" - my daily driver PC decided to set this to stop BOINC when other work was consuming 1% of the resources - and it stopped processing! Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
Tasks checkpoint to disk every 36000 seconds This setting is way too high. This would mean, a task would never save a result for 10h run-time. I advise you to set this to something like 60 seconds or max. 300 sec.[/u] |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
parameter now set at 300 secs thanks! |
rob smith Send message Joined: 7 Mar 03 Posts: 22220 Credit: 416,307,556 RAC: 380 |
Wrong - Checkpointing is to capture data that has been calculated, but the task is still running - if a task completes before the checkpoint time then it gets reported in the normal way. How else do you think "crash and burn" tasks - those which complete in a few seconds - complete and get reported? That said 36000 seconds is considerably higher than the default of 60 seconds. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
Aborts were initiated by my own intervention as they wouldn't finish computing before their upload latest submission date back to SETI. Thanks! |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
Ok, the check-pointing interval may not be the cause, but it was way off. However, i would watch a tasks progress and see, if it finishes and is reported. |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
Several task will finish inside 10 hours. (0700UTC Friday 15 August) we shall see .... How do I make sure I haven't inadvertently set the triggers for "other work running" please? I am not conciously aware that I have? but will check parameters with direction.... thanks! |
rob smith Send message Joined: 7 Mar 03 Posts: 22220 Credit: 416,307,556 RAC: 380 |
...away from anything running BOINC just now, but from memory its in the BOINC manager, advance view, computing options (the words used mean what I said, not the exact words) Set it to 50% for starters and see what happens. A 10 hour deadline may be OK if you have changed your system in a "significant" way (or at least what BOINC considers to be significant) - it will take about a dozen or so tasks to be validated before the guesstimate becomes less unrealistic... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Oddbjornik Send message Joined: 15 May 99 Posts: 220 Credit: 349,610,548 RAC: 1,728 |
Have you tried turning it off, removing the battery, and then going over all ventilation openings with a vacuum cleaner set at "brutal" level? I do that to my laptops every now and then. It can have a remarkable effect. |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
It's not appreciably running hotter as far as I can see. The fan is ON and the app is running whilst number crunching is underway. thanks! |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
something other than this from BOINC Manager? Processor Usage Computing allowed always, while CPU in use. (no restrictions) GPU use allowed only when CPU not in use. and after 1 minute idle time Application switching every 60mins No restrictions on multi-processor (value =0) No restrictions on CPU time (value = 0) |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
no I haven't but I will consider it thanks! |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
Imprtant: Set CPU time to 100%.(not to zero) There is a bug in boinc. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Several task will finish inside 10 hours. (0700UTC Friday 15 August) I would disable the setting when using BOINC 7.4.42. I had the same problem on my Mac with 7.4.42. BOINC would suspend work claiming the computer was busy even though it wasn't. That appears to be what's going on here; http://setiathome.berkeley.edu/result.php?resultid=4217723312 BOINC is just suspending work over and over and with the checkpoint set so high nothing ever gets saved. The task starts from scratch after every suspension. I went back to BOINC 7.2.33 and the problem disappeared. Unfortunately I installed a CUDA card and now have to have 7.4.36 to use CUDA. So far I haven't had that problem. In BOINC Advanced View, go to Tools/Computing Preferences/processor usage. Then set 'While Processor usage is less than 0 percent...' Change whatever it is set to Zero. Or, install BOINC 7.2.33; http://boinc.berkeley.edu/dev/forum_thread.php?id=8378&postid=51551#51551 Also set 'On multiprocessor systems, us at most 99 %...' That will reserve a core for the ATI 6750 Set 'Use at most 100.00 % CPU time'. |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
Important: will do! thanks! |
John McIntyre Send message Joined: 13 May 02 Posts: 14 Credit: 18,937,378 RAC: 7 |
Thanks TBar for the guidance. |
©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.