I still think you should work with them rather than hold back. I don't quite understand your "incentive" for holding back... If you give them one or two lines and they don't try it, how is it different from giving them 100 lines and they don't try it? They have said that they basically inherited the code. As I pointed out before, if you are given code and are told that it works, most of the time you're not going to go over it with a fine-tooth comb looking for efficiency improvements right away...

I agree 100%, FULL credit should be given to valid results, no matter how fast it is getting done, by whomever and whatever. Period.

OK, so I ask that either you or zeitgeistmovie to define 'Full Credit'.

Alinator

The same credit I would get, if I had crunched that particular w/u, even if my box took 5 hours to crunch verses a couple of minutes.
That might sound crazy but, it's the same work isn't it?
Optimizations RULE.. :D
After all, even before this latest rubish, it still took twice as long for a Windows O.S. machine to crunch the same job as a Linux O.S. box. That's why my boxes dual boot Linux now.
That's a difference in optimization. Were do we draw the line?
Should I be credit punished because my Linux app from MW goes twice as fast as my Windows MW app?
NO.
Should others be punished because there app is faster than mine?
NO.

Ok, then isn't your last statement exactly what I said here?

The point is that the Credit Anarchists are saying the basis should be set to whatever anyone feels like setting it to.

The Kredit Kops are saying, no the basis should be set to what a plain vanilla math version of the app would result in using BM-T scoring calibrated to the specifications of the reference computer, since that is the way the Cobblestone is defined, period.

Alinator

I raised the amount of time WUs were being kept on the server to check out the results and make sure the WUs were correct.

And what have you found so far?

Again, if the result is valid, then so is the credit. And limiting the credit for a valid result is not being fair and equal.

How am i supposed to do this? i have no more access to his source code than you do.

It is your project is it not? Ask him for the code, and review it. But better yet, just do what you say you are already doing, review the returned result for validity. If it is valid, then again, the credit is valid. Period.

I agree 100%, FULL credit should be given to valid results, no matter how fast it is getting done, by whomever and whatever. Period.

OK, so I ask that either you or zeitgeistmovie to define 'Full Credit'.

Alinator

I raised the amount of time WUs were being kept on the server to check out the results and make sure the WUs were correct.

And what have you found so far?

Again, if the result is valid, then so is the credit. And limiting the credit for a valid result is not being fair and equal.

How am i supposed to do this? i have no more access to his source code than you do.

It is your project is it not? Ask him for the code, and review it. But better yet, just do what you say you are already doing, review the returned result for validity. If it is valid, then again, the credit is valid. Period.

Even for 4 cores it's not enough, assuming 5-6hrs per WU it needs at least 16 per day.

If something interrupts a disk write it will give you that error too Tutta, see the folowing > http://www.boinc-wiki.info/index.php?title=Unrecoverable_error_for_result_%27(result)%27_(process_got_signal_11) ... :)

Don't get me wrong now, I never said there was cheating. If I suspected there was cheating, I would have flat out said it was.
I need me a farm of those Pentium 4's! LOL

the farm of P4s won't help... you need the application to do this ;)

I did not say it was cheating, that has yet to be proved. I said it could be cheating, that has yet to be disproved. It could be other things as well (such as a win9X client that restarted causing the CPU time to go to 0). Until there is some sort of investigation, we won't know either way. I believe that it is probably only the project admins that can do the investigation, we can only bring the possibility of a problem to their attention.

There is no problem with this nor is it cheating. It's just ... doing the math a little bit faster that's all... and of course its valid.

... that when the purge starts, it's so expensive that it slows down the rest of the system and can't catch up, while pretty much makes everything slow and unresponsive. so to get the purge done we have to stop the assimilator and work generation.

Can't you automatically delete any WU that's been done for 24 hours or more? So rather than doing a full purge every 24 hours you'd be purging WUs continuously at roughly the same rate that they're coming in..

Perhaps keep a log of completed WUs (sorted by reported completion time by definition), and purge old (>24h) entries whenever you add a new one (i.e. a WU is reported)

if(thisWU was completed more than 24 hours ago) delete and examine next item;
else exit;

To put it in pseudo-code <.< (had a brain-fart, so that was the best I could do)

i'll try and give this a shot and see if the server gets overloaded again. a day of WUs (at ~6 WUs a second) is quite a lot... which is part of the problem