Ideas for new point system

[perrosnk]

Hello everyone!
I am very new at the distributed computing world.
I was wondering if there is any way you can think of, to make a unified point system. For example, is there a way of real time measurement of the CPU or GPU work done, in flops or whatever else, and then get points based on flops or any other measure?
Is there a way of ensuring that nobody will be able to cheat the system?

 

[petrusbroder]

Hi perrosnk!
Welcome to the DC-forum.
This has been tried for "ages" ... but all the projects want to attract crunchers and thus give away points. If a project is very "efficient", i.e. if it - let us say - gives 100 points for a WU which takes 1 minute to crunch, then that project will attract more followers than a project which gives only 10 points for 1 minute crunching (using the same computer). Therefore - althought such a measure exists - that is not used by all projects.
Also, what would be a "golden standard"? A cpu or gpu crunching 1 megaflop/hour of doubble precision intergers or single precision integers?, using different ways to crunch them (sse - sse4 or some othjer optimization) and at what accuracy? Of course this can be agreed upon, but is it worth the effort and frustration and the bads will it may create? This is still volunteer crunching, each cruncher finds the project/projects he/she likes and which gives the best return (in their personal opinion), and the most important priority is to get the science crunched ... and to get the most possible effect, not only on science, but also in the stats.

So, let us get the discrussion going ... it has been some time ago we had a thread about the points! In the early years of crunching (1990-ies and 2000 - 2005) these discussions were heated and emotional, but then there were only a few projects and the "stats-wars", challenges, competions, races were extremely importanat ... but today? Not nearly as much as then.

Some oldtimers - such as Wiz, amd.borg, assimilator1, smoke and others remember this too.

And the elimination of cheating: that is quite impossible. If a man can construct some way to do things, an other one can think a way around it and earn more. The real cheaters will be found out, sooner or later, simply because they are too greedy and accumulate points in such a way that they will be discovered and then loose their points.

 

[Bubbleawsome]

Not to hijack this, but how would you cheat? don't you just process the WUs?

 

[somethingsketchy]

Not to hijack this, but how would you cheat? don't you just process the WUs?

http://forums.anandtech.com/showthread.php?t=2315415

Recently (very recently in fact), there was a cheating scandal, and several teams lost points over their processing of WUs. While it doesn't happen very often (thankfully), not everything is perfect with the code.

 

[Drsignguy]

http://forums.anandtech.com/showthread.php?t=2315415

Recently (very recently in fact), there was a cheating scandal, and several teams lost points over their processing of WUs. While it doesn't happen very often (thankfully), not everything is perfect with the code.

Yea, that was strange indeed. But they got caught and got reprimanded for their dirty deeds.

At one time, especially during my first couple of years of folding I watched the point contribution and it was slow going. You could tell by the amount of WU you have done compared to your points. Now, since the WU's produce more points and by the implementation of the GPU, the points can accumulate in a hurry, well depends on each set up.....

Now, I like watching our WU stats. That's what is truly impressive to me. Its incredible to see how many each of us have done and then send them back to Stanford for them to do their research. Massive amount of work really. Points are cool for the only reason to show you where you are in the standings but the true stats are the WU's.

 

[petrusbroder]

Yes, I agree fully!