Matroxusers team for CPDN BOINC

[VJ]

AAARGGGHHHH

Another power black out (1 hour 30 minutes, and counting). Guess I lost most of the crunching the system did today...


Jörg

[dshumake]

The crunching shouldn't have been lost, unless you have a very long wait to write to disk.

dshumake

[VJ]

Well, the wait is the default time, but I also have a write cache enabled on my drives...

Luckily, it didn't cause problems, and the loss to my crunched units seems very limited. Phew...


Jörg

[SitFlyer]

Suprised no one has mentioned,
we're 25th now

[DirtFarmer]

....with 25 members and over 25000 credits today. Impressive group if I must say so myself. Harnessing the computational power of the mighty MGA chips.

[dshumake]

Passed 200K.

[Umfriend]

congrats. Impressive indeed. I do have this notion that Maggi is slowly catching up with you though.

[dshumake]

I stopped running most of the machines all day. They were supposed to be running only during the nights. I had to go in and manually change the xml setting to make the selection stick.

dshumake

[VJ]

No monster update today, but I'm crunching away (so the credits will be added eventually). Hopefully next week...

(WOOHOO, a weekend without a powerblackout )

Jörg

[VJ]

This is depressing: for the second time, my boinc client seems to have forgotten all data (it just starts with the message "The host has not yet been assigned an ID", and the client lists no jobs (there should be 4 !). Nothing I could do brought back the jobs.
Consequently, I replaced it with my last backup, which is from last week... So I lost a week of crunching...

Any idea what is going on with the client, and why it keeps forgetting that?

(I have increased my number of backups: I will back up the boinc client every day)


Jörg

[Umfriend]

No clue. The model that restarted after 27 trickles had almost caught up (I was depsrately trying to get it to grant me credits again to keep ahead of RC Agent). After it did #24 it crashed....

I did consider removing that model entirely after the first restart so I'd get credits again right away, but I considered that cheating, so I did not. But now it loaded a new model by itself. The good thing is it'll get me credits again. The bad thing is I have yet another failed model and I would like to think I did contribute something to the scientific effort as well.....

[Umfriend]

BTW, there is a number of people that apparantly do not crunch anymore:
1. dbdg
2. Rattledagger
3. CaineTanathos
4. AZ

Whassup guys? We _need_ you!

And where is Paddy "The Perpetual Re-Wiring Job" when you need him?

[VJ]

Originally posted by Umfriend
I did consider removing that model entirely after the first restart so I'd get credits again right away, but I considered that cheating, so I did not. But now it loaded a new model by itself. The good thing is it'll get me credits again. The bad thing is I have yet another failed model and I would like to think I did contribute something to the scientific effort as well.....

I thought a failed model meant the predicted evolution was extremely unstable.
If I remember this correctly, that would mean that a model failing actually contributes something...
(showing that something goes wrong is a result)



Jörg

[Umfriend]

Well, there is another guy crucnhing the same model and he got quite a bit further already, close or in pahse 3. Must be me.

[GNEP]

Try underclocking your machine if you are desparate for it to complete a model.

Or try running memtest for a day - this could be indicative of a memory error that doesn't usually affect "normal" programs.