Way wrong!

If you merge a computer with large(ish) amount of credit with one that has just opened its account then it looks like this new! computer is doing lots of work in very short time. Therefore the Recent Average Credit is very high.

As described in the BOINC Wiki the total credit is how far you've traveled, RAC is speed at which you heve done it.

Andy

If you look right now it's 477K whereas yesterday you reported almost 520K. Notice how fast it is dropping.

If you look at the results list for that computer, you will see:-

1. Current CPUID was created on 29 Sep 2005.
2. 11,450 credits but only 7 days since the ID was created
3. Crunch time for a successful result - 5.3 hours
4. Achievable score per day - probably around 320-340
5. Therefore days needed for 11450 credits - approx 34.

Can you see what must be happening.

6. The current results list actually goes back to around Sep 16 but there would have been more earlier.
7. The computer developed some sort of fault around Sep 27 - check the results list.
8. In trying to correct the fault, probably BOINC was reset/detached?? maybe a few times.
9. As a result new CPUIDs were issued along with fresh results. Take a look at how many :).
10. The user tried aborting many of these (click on some and look for "Aborted via GUI" - you'll find plenty.
11. The user eventually fixed whatever the problem was and normal crunching resumed on 29-30 Sep.
12. As a final cleanup stage the user must have merged all the phantom CPUIDs into the one we now see.

My guess is that this act of the sudden dumping of all credits into the latest CPUID causes the glitch in RAC that is now going to rapidly die away. The rocket took off with a sudden burst of high speed but didn't get very far to continue the analogy that I saw Winterknight say was in the Wiki.

Please realise that the above is guesswork on my part based on looking at the results list. Your persistence in asking the question finally prompted me into having a look at this CPUID. I can only see exactly what you can see and unless someone with greater knowledge chimes in, I don't know the exact nature of the problem or whether the answer I am giving you is totally correct. However it sounds plausible to me.

We are fortunate that CPUIDs are shown on the "Top Computers" list otherwise we wouldn't have been able to peek at this poor guy's problem. Notice how he is listed as "anonymous". That means his computers are hidden. I feel sorry for him because he probably thinks his problems had gone unnoticed and they probably would have if you hadn't come along and blown the whistle on him!! :). So it's all your fault if he comes along and accuses us of prying into his private business ;) :).

The great thing however is that he is back crunching away, returning validated results once again. He can enjoy his fame at the top of the top computers list for a little while longer yet :).

To me, there is probably a very important message here that I would like to mention. I'm assuming the problems were exacerbated by inappropriate resets/detaches that probably created a bunch of new/phantom CPUIDs. Therefore a "reset" type choice of action should only be used as a last resort when all other "gentler" options have been explored. Too many people are too used to the reset button on the front of a Windows computer, where lockups had to be cured with that button :). Maybe the "Reset" function should display as "Reset (last resort)" or something like that :).

Edit: Added some smilies to make the joke about "peeking" and "prying" a bit more obvious.

It was just a joke :). Notice the smilie on the end of that statement.

From what's left of his results list, I can't see if he had problems three weeks ago. Are you saying that prior to the start of the currently visible results list - say around Sep 10 - he still had a 500K RAC?

One small correction to the previous message. He restarted successful results around 28 Sep but he is not still going. Last valid result was returned on Oct 01. He might have stopped or it might just be that BOINC has decided to start running other projects for a while but it sure looks like he has stopped EAH.

Don't take too much notice of my queer sense of humour. It's perfectly fine to look at any information you wish and more power to you for doing so. Thanks for being persistent with your questioning. You forced me to do a bit of research so I could try to give you an answer.

OK, I've just had a look at the Client errors in the results I can see for your windows box. Have you tried looking up "There are no child processes to wait for. (0x80) - exit code 128 (0x80)" as an error message in the Wiki?

Tell us a bit more about it.