I noticed I had several invalids in a row when attempting to test my new laptop a few days ago. It seemed like it was running fine, and even managed to complete and return several valid units.

Just a guess, but I seemed like it started when I started hitting the suspend button to pause the unit mid-processing.

Here's my data in case it might be useful...I've stopped running it on my laptop as I don't want to overheat it, but it would be interesting to see if there was something actually causing this...



Core 2 Duo Mobile, 2 gigs of ram, Windows MCE, BOINC 5.4.11
Computer ID: 85462


B05560_ 0168_ CTMA4A-37-23-2-c2 Invalid 11/22/2006 10:21:38 11/24/2006 18:41:59 3.41 33 / 9
B05560_ 0049_ CTMA4A-37-14-10-c2 Invalid 11/22/2006 10:16:26 11/24/2006 18:41:59 3.85 37 / 6
B05560_ 0126_ CTMA4A-37-2-18-c2 Invalid 11/22/2006 08:31:02 11/23/2006 18:00:56 3.10 30 / 11
B05559_ 0195_ CTMA4A-37-25-21-c1 Invalid 11/22/2006 07:18:15 11/23/2006 18:24:07 5.18 50 / 12


Interesting to note, the units that were invalid all took over 3 hours, while the units that were valid were about 2 hours or less. Also, I noticed with the last unit I attempted to complete on this machine, it was over 50% complete, and then I rebooted, and the unit started over. When it started over, it seemed to take even longer to reach the 50% mark, and so I simply aborted the unit, figuring it would probably end up invalid like the others anyhow.

Mousie, since it is a laptop then there is a reasonable chance it was caused by overheating. I can't offer a good explanation for why it would affect some WUs but not others except to say your notebook may be just on the borderline, sometimes it overheats while other times it does not. Did you notice the CPU temperature while crunching?

Many motherboard manufacturers have free temperature monitoring utilities you can download from their website. There is also the most exellent Speedfan, 3rd party software, which monitors temperature of almost all motherboard brands. In addition to monitoring temperature, such utilities usually allow you to speed up the cooling fan(s) sooner. In fact your notebook may already have such a utility installed by the manufacturer or available from their website.

You might also try throttling the CPU with Threadmaster. If you search these fora for Threadmaster you'll find links to advice on setting it up. From what I've heard it works well but can be a bit tricky to setup. Also, the 5.6.x alpha versions of BOINC have a built in throttle but it's reported to not work very well. They claim 5.8 will see the throttle fixed but perhaps the current alpha, 5.7.x already has a proper throttle.

Yet another alternative is to use UD which has an easy to use throttle that is known to work well.

---

Mousie,

experiment with switching off power saving features when u r hanging on the 110 / 220 juice. Best place is likely in the BIOS, but there is often software like Speedstep that controls it....depends on the brand.

Just reading some lappies running at 80 / 100c continuously...57c can't fry an egg. For a solid temp check, try out Speedfan.... www.amico.com. The latest is now also able to deal with the Core 2 Duo's.

cheers

I, too, have been getting a large amount of "Invalids". All clients are the standard client.

Right now I have 298 Invalids.



I thought this was being worked on.
I can't afford to pour $$ into electricity and equipment to have the results "tossed aside" and denigrated.

Mousie,

I've had this problem too on HDC WU's, and Sekerob has commented on it in the HDC thread.

The HDC checkpointing doesn't work too well, so if you suspend when you are well into a task, it can start over. Unfortunately, your CPU time does not start over, so when you put in your (valid) claim, you get penalised for being a statistical outlier.

I think.

Jonathan.