hello all:
i want that BOINC just work when 5 minutes after my computer is idle, so i did that on preferences, but it\'s working at 100% cpu when im using the computer, why is that?
thank you all
p.s.: excuse my bad english.

There seems to be a problem making changes to Preferences at the moment. I\'m contacting Staff about it.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

and how do I know if this has been solved?

The discussion is here.

For some time since early October I had been having problems with my machine hanging for long periods when I interrupted the screen saver while climatechange was running.

I noted that although my setup was supposed to suspend processing when I used my machine, this clearly was not the case. I decided to change my profile so that application was removed from memory rather than just suspending work.

This seemes to have fixed the hang problem, but it took me a while to notice that I was never getting beyond a few hours of processing into my current work unit. I have not tested thoroughly, but it looks as though it might reset everything whenever I restart my PC. I note from the stats graph that I have not uploaded any data for about 10 days - which is probably about the time I made the profile change.

Any ideas?

---

Keeping the data in memory when something else is being processed, means that the modelling can continue from where it left off. Otherwise the program has to start the model from the last checkpoint. If the memory space is needed by the other programs, then the model data will be swapped to the hard disk \'virtual memory\', from where it can be retrieved.

If the option mentioned is set to NO, then the data isn\'t kept.
So what your program has been doing, is to recalculate the same area of time over and over.

If you MUST use the screensaver, (which can, depending on the hardware, use up to 50% of the processor\'s time to calculate), then at least wait until a checkpoint has been reached before you shut down the program.
There is a checkpoint counter on the globe display.

However, the main problem in your case, is that you\'ve crashed every model that you\'ve had since the 5th of October, as you can see from the Results page for your computer.

You may find it helpfull to look through the README files here, in particular \"Crashes and other problems \"

Thanks I will investigate.

I don\'t know whether I have an odd setup or am just unlucky, but I do seem to have a lot of trouble with Climate Change (SETI is my other project and that runs without obvious problems). Earlier in the year it was continually crashing and Mike Mars got me involved in the Beta site. That ran perfectly, but when I transferred back to the main site I had all the hanging problems and general annoyance that led to this.

---

Pressure of other work has stopped me investigating this problem further since Les Baylis\' note at the ned of November. However, over the weekend I tried to uninstall as much of the climate change material as possible in the hope it would re-install iteself and run properly. I also changed my settings so that it only suspended work rather than closing down. I could not find an uninstall process as such, so delteted the Climate change files from the BOINC folder and reset the project.

The problem with hanging for up to two or three minutes at a time has returned, although I have not yet followed the advice on removing it as a screen saver. However, of more concern is the problem of crashing.

It seems to run to about 8% and then I get an error in the log that reads:

11/12/2006 14:24:22|climateprediction.net|Unrecoverable error for result hadcm3ohc_0oqz_05580434_1 ( - exit code -1073741819 (0xc0000005))

and an on screen error window:

hadcm3trans_5.15_windows_intelx86.exe Application Error
The instruction as \"0x693a8a20\" referenced memory at \"0x00000044\". The memory could not be \"read\".

I\'m not sure if the addresses are identical with each crash, but the general message is the same each time. I am wasting both machine time and units and so need a sensible solution if I am to bother any more with this project. I have looked at the FAQs on crashes and none of them seem to be at the root of my problem.

Any ideas? BOINC seems to be OK as SETI runs OK.

---

-107xxxxxxx is pretty generic for usually hardware issue. Some of the fixes are:

Make sure video card drivers are up to date. Make sure DirectX is up to date. Do cleanup of dust and dirt from fans, heat sink(s), etc. Do memory testing (Memtest86+) and processor testing (Prime95).

Hope you find what the issue is, and can get yourself back crunching with a happy machine.

---

Hi Alan

Pooh Bear is quite right about the error codes. You\'ve had loads of models recently and they\'re all crashing within a day or so, mostly with -107 errors but also the odd -1, which can be caused by the same sort of problem.

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/results.php?hostid=197842

If you click on the result numbers in the left-hand column you\'ll get details of each model\'s fate.

Your computer isn\'t the problem - it more than meets the minimum specs. But each -107 or -1 is an unfortunate event on your computer. It\'s usually easy to stop this happening.

*Disable the screensaver. NOW. Right-click on the desktop - properties - screensaver - select none. View your globe only when you have time for it by using the button in the boinc manager, and don\'t maximise the globe window.

*In the Running the model README here, read the top tips
http://www.climateprediction.net/board/viewforum.php?f=36&sid=cecdb408e438821d177828cd1560cd0d

*In the Crashes README, look at

---item #5 by Mike about how to avoid model crashes in general and -107s in particular. All the advice is compulsory to keep models safe. Luckily it\'s all easy to implement.

---item #6 by Thyme Lawn about updating graphics cards, which is a free update from the web a bit like Windows updates. You really need to do this.

---When, after doing the above, you (I hope) have a model running stably, item #1 still in the Crashes README. It\'s instructions by Les on how to back the whole boinc folder up so that if by chance you have another model crash, you can restore the backup and continue crunching the same model.

That looks like a lot to take on board all at once. Take it step by step in the order I\'ve set it out. Let us know how you get on.

---

Cpdn news

Thank you both for your speedy feedback. Most things seem OK (see comments in italics below), but there is obviously a problem. One thing that puzzles me is that when I was running the Beta version during the summer all was going fine. I can only assume there is a signficant difference between the Beta and the production version that I am now struggling with. If anyone can pin that down it might prove useful.

-------------------

Pooh Bear is quite right about the error codes. You\'ve had loads of models recently and they\'re all crashing within a day or so, mostly with -107 errors but also the odd -1, which can be caused by the same sort of problem.

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/results.php?hostid=197842

If you click on the result numbers in the left-hand column you\'ll get details of each model\'s fate.

A quick look at some show a consistent occurrance of \"Not a JPEG file: starts with 0x01 0xda\". Not sure it that is cause or effect.

*Disable the screensaver. NOW. Right-click on the desktop - properties - screensaver - select none. View your globe only when you have time for it by using the button in the boinc manager, and don\'t maximise the globe window.

I did that yesterday, but download of new work packages only ran this morning, so will have to wait and see how it runs.

*In the Running the model README here, read the top tips
http://www.climateprediction.net/board/viewforum.php?f=36&sid=cecdb408e438821d177828cd1560cd0d

Probably the only relevant one I have not been doing is to manually disable boinc when I turn off my machine. Bit of a drag to remember this, but will see what happens.

*In the Crashes README, look at

---item #5 by Mike about how to avoid model crashes in general and -107s in particular. All the advice is compulsory to keep models safe. Luckily it\'s all easy to implement.

Sadly none seem relevant, with the possible exception of the stability test, which is a major undertaking. Temperature, etc. OK

---item #6 by Thyme Lawn about updating graphics cards, which is a free update from the web a bit like Windows updates. You really need to do this.

I have done the OpenGL test. Loads of (to me) unintelligable information that proves I am nothing like as computer literate as I thought! Machine passes tests 1.1 to 1.5 but fail 2.0 (Shading language and deferred rendering). Not sure how relevant this is to the problem. All video drivers appear to be up to date. The extensions verification have a number of reports along the lines of \"GL_ARB_imaging was not found, but has the entry point glColorSubTable \". Not sure if these relate to the 2.0 failures or some other problem, and I have no real idea how to remedy.

---When, after doing the above, you (I hope) have a model running stably, item #1 still in the Crashes README. It\'s instructions by Les on how to back the whole boinc folder up so that if by chance you have another model crash, you can restore the backup and continue crunching the same model.

That looks like a lot to take on board all at once. Take it step by step in the order I\'ve set it out. Let us know how you get on. [/quote]

---

I am unsure if this might be of help, but some ISPs do a on the fly compression, which seems to change jpeg files into lower resolution for faster download. If this is the issue (I usually do not think it causes a -107xxxxxx, but ???).

Go to Your Account, General Preferences and change the option:
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. to Yes. Then make sure you save and update.

Just a thought.

---