I have my computing preferences in BOINC set so that the box is UNchecked for "Leave applications in memory while suspended," and also unchecked for "Compuing allowed while computer is in use." Nevertheless the WCG (World Community Grid) files remain in memory, to the tune of around 200MB!

I upgraded to the most recent BOINC version a couple of days ago (7.0.28) to be sure it wasn't a bug that had been fixed. I thought it was working okay after that, but now they're back again. Is there something else in the setup that's overriding my preferred settings?

Another thing that's weird is that right now I have 7 files in memory, whereas I believe only 4 should be running at a time (and only 4 are showing as suspended under the Tasks tab).

I'm running this on Vista (Service Pack 2) 32-bit with 4GB RAM. CPU is Intel Core2 Quad. I make sure everything is always upgraded.

Any ideas what could be wrong? I run a lot of programs and 200MG takes a chunk out of the available memory.

Thanks for the help!

Are you using the Web Preferences or the Local Preferences set in Boinc Manager? Local Preferences override Web preferences, you'll need to hit the Clear Button to revert to Web Preferences.

Claggy

I'm setting the preferences through the menu on the BOINC Manager window (that comes up when I double-click the system tray icon). I imagine that's at the local level, correct? If so, it should take precedence, from what you're saying.

Make sure that the tasks checkpointed at least once. When tasks have started, but never checkpointed, they'll stay in memory when BOINC suspends computation.

You can check when/if tasks checkpoint, by setting <checkpoint_debug/> in cc_config.xml

---

Starhugger,

A prof of mine always repeated that believe belongs in a house of faith ;>). WCG has several sciences that launch 2 or 3 processes, a controller/stager and a worker, only 1 for a task uses CPU time, so it's well possible that on 4 tasks of different sciences you have 7 processes showing in the task manager. Specifically CEP2, DSFL, GFAM, SN2S have show this. Do not think any other active science does this.

As for staying in memory, follow Ageless reply. If LAIM is off, no task should remain in memory after first checkpoint when paused. If they do, report detail, including what exact sciences on WCG forums else the techs will never know.

Notably, even when staying in memory, after a while the OS moves RAM occupying memory segments of a task off to virtual memory to make space for whatever else might need it.

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

Ageless: Thanks very much for the suggestion. I'll watch to see if this changes now.

SekeRob2: Thanks for the information, although I could do without the (unsolicited and unnecessary) grammar lesson. My writing is fine, thank you.

Sorry, my subconscious must have been typing the grammar lesson, which my conscious side is missing complete. There was no critique whatsoever on your writing grammar or syntax, that I know for sure.

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

SekeRob2, thank you for the apology: accepted. :-)

Hmm... It seemed to be working for a while, but this morning it's back to keeping the files in memory again. I keep my computer running all night, and the log confirms that several checkpoints were done througout the night, the most recent being almost 4 minutes before processing was suspended due to computer activity. (I have the checkpoint set to 60 seconds at the most.)

What's weird is that one of the jobs running (Clean Energy Project) hasn't been checkpointed. It has over 7 hours of processing having elapsed, with over 5 hours to go. It was last running on the 19th, which might have been before I added <checkpoint_debug>1</checkpoint_debug> to cc_config.xml. Perhaps I didn't do that correctly?

I just rebooted (for unrelated reasons...as far as I know!) and those files are no longer showing in memory, although that doesn't mean they won't get stuck in memory again at some point. I have no explanation for how/why they got stuck there, apparently last night.

But here's another weird thing: that one job that doesn't seem to want to checkpoint (Clean Energy Project) is now showing as only 36 minutes elapsed and 13 hours remaining. ??? I'm thinking maybe I should just abort that job; maybe it's a source of this problem. Does that sound like a good idea? I don't want to abort it if the majority of it really has been processed and if it's not the source of the problem.

I'm at a loss as to what is going on here. Any other insights? Thank you!

I added <checkpoint_debug>1</checkpoint_debug> to cc_config.xml. Perhaps I didn't do that correctly?

<cc_config>
<log_flags>
<checkpoint_debug>1</checkpoint_debug>
</log_flags>
</cc_config>

It's got to be cc_config.xml, check the extension as when you make it in something like Notepad, it'll get the .txt extension. And this file has to be written into the BOINC Data directory, where client_state.xml lives. It's not used when written into the BOINC Programs directory.

---

Everything checks out. The extension is xml (and I double-checked that this is the file I edited), and it's in the same directory as client_state.xml.

Everything checks out. The extension is xml (and I double-checked that this is the file I edited), and it's in the same directory as client_state.xml.

Can you post your Boinc startup messages from the Event log, the first 20 to 30 lines will do.

Claggy

Hi Claggy, thanks for the help. I'll put the startup messages below from when my 'puter was rebooted this morning, up to the point of restarting the various jobs and the one checkpoint that follows them.

FYI: I just came back after being out for a couple of hours (everything was still running) and files are back in memory again, even though they're suspended.

Edited: ...and in the few minutes since I put this post together, those files are suddenly gone from memory!


=====EVENT LOG=====
2013-03-21 11:22:06 AM | | Starting BOINC client version 7.0.28 for windows_intelx86
2013-03-21 11:22:06 AM | | log flags: file_xfer, sched_ops, task, checkpoint_debug
2013-03-21 11:22:06 AM | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
2013-03-21 11:22:06 AM | | Data directory: C:\ProgramData\BOINC
2013-03-21 11:22:06 AM | | Running under account Owner
2013-03-21 11:22:06 AM | | Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q8200 @ 2.33GHz [Family 6 Model 23 Stepping 7]
2013-03-21 11:22:06 AM | | Processor: 2.00 MB cache
2013-03-21 11:22:06 AM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 nx lm tm2 pbe
2013-03-21 11:22:06 AM | | OS: Microsoft Windows Vista: Home Premium x86 Edition, Service Pack 2, (06.00.6002.00)
2013-03-21 11:22:06 AM | | Memory: 3.00 GB physical, 8.76 GB virtual
2013-03-21 11:22:06 AM | | Disk: 465.76 GB total, 244.95 GB free
2013-03-21 11:22:06 AM | | Local time is UTC -4 hours
2013-03-21 11:22:06 AM | | No usable GPUs found
2013-03-21 11:22:06 AM | World Community Grid | URL http://www.worldcommunitygrid.org/; Computer ID 858508; resource share 200
2013-03-21 11:22:06 AM | World Community Grid | General prefs: from World Community Grid (last modified 11-Sep-2009 01:28:20)
2013-03-21 11:22:06 AM | World Community Grid | Host location: none
2013-03-21 11:22:06 AM | World Community Grid | General prefs: using your defaults
2013-03-21 11:22:06 AM | | Reading preferences override file
2013-03-21 11:22:06 AM | | Preferences:
2013-03-21 11:22:06 AM | | max memory usage when active: 767.16MB
2013-03-21 11:22:06 AM | | max memory usage when idle: 3068.64MB
2013-03-21 11:22:06 AM | | max disk usage: 10.00GB
2013-03-21 11:22:06 AM | | don't compute while active
2013-03-21 11:22:06 AM | | don't use GPU while active
2013-03-21 11:22:06 AM | | suspend work if non-BOINC CPU load exceeds 30 %
2013-03-21 11:22:06 AM | | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
2013-03-21 11:22:06 AM | | Not using a proxy
2013-03-21 11:22:23 AM | | Suspending computation - initial delay
2013-03-21 11:22:23 AM | | Suspending network activity - computer is in use
2013-03-21 11:23:13 AM | | Resuming network activity
2013-03-21 11:24:06 AM | World Community Grid | Restarting task E212263... using cep2 version 640 in slot 1
2013-03-21 11:24:30 AM | World Community Grid | Restarting task faah38516... using faah version 715 in slot 0
2013-03-21 11:24:30 AM | World Community Grid | Restarting task faah38528sing faah version 715 in slot 3
2013-03-21 11:24:30 AM | World Community Grid | Restarting task GFAM using gfam version 612 in slot 2
2013-03-21 11:24:47 AM | World Community Grid | [checkpoint] result GFAM... checkpointed
===============

I'm just taking a wild guess here, but could this have anything to do with how Vista does prefetching? Are the processes that you are seeing in the task manager actually using any CPU? or is it just sitting in memory taking up space? The other question to look at is...does Windows release that memory if another application is in demand?