| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Chat Room
Thread: How would YOU like the points system to work? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 30
|
|
| Author |
|
|
OldChap
Veteran Cruncher UK Joined: Jun 5, 2009 Post Count: 978 Status: Offline Project Badges: 
|
I cannot fathom how the points system works.... One day my rig claims 25 points per hour the next it is 18 and these are not the extremes. It is also for a rig that only a small proportion of wu's seem to get validated. and that currently just runs SN2S.
----------------------------------------This all led me to wonder If I am Alone? surely not. Is there some algorithm that maybe takes many days to adjust a rig's points claim currently? So, boys and girls, how would you set things up? what priorities would you think fair to all participants? Runtime? Speed? Task size? etc. In short, what method of calculation is fair to all?  [Edit 1 times, last edit by OldChap at Aug 28, 2013 6:59:35 PM] |
||
|
|
OldChap
Veteran Cruncher UK Joined: Jun 5, 2009 Post Count: 978 Status: Offline Project Badges:
|
Let me pose a somewhat simplistic question that covers just one element of many...
----------------------------------------Rig 1 has 1 cpu running at 4.0G with 4 cores Rig 2 has 2 cpu's running at 2.0G each with 4 cores Both cpu's are from the same make and model range. Should they get the same points per day assuming they get the same amount of work done? What is the best measure of work done? [Edit 1 times, last edit by OldChap at Aug 28, 2013 6:58:13 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
What is the best measure of work done? A Cure |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
What is the best measure of work done? Ultimately, I'd say points. IIRC, boinc points are supposed to be standardized to measure actual work (the cobblestone unit). Point calculations and issuance aren't perfect by any means, but cpu time and results returned don't at all represent actual work done. Also, there are many factors that could explain the variance of points per hour.. multi-day validation probably being the biggest. The average of you and the wingman should, in theory, average out over time. I'd say points are a pretty fair estimate. Edit 2: My (more than likely naive) approach would be: Benchmarked cobblestones per thread per hour * cpu time = claimed points [Edit 2 times, last edit by Former Member at Aug 28, 2013 10:57:48 PM] |
||
|
|
astrolabe.
Senior Cruncher Joined: May 9, 2011 Post Count: 496 Status: Offline |
Sorry OldChap, but I have a life outside of WCG and I really cannot care. I don't babysit my machines and I can't tell you the points/hour that any of my machines have ever been granted.
Besides, IIRC you really don't have a say in how it works. |
||
|
|
cjslman
Master Cruncher Mexico Joined: Nov 23, 2004 Post Count: 2082 Status: Offline Project Badges: 
|
Does it really matter? You can't do anything with the points. I find the badges much more entertaining: they're like milestones or objectives. I really don't pay that much attention to the points. I've heard of people running one project vs another project (both inside and out of WCG) because it "gives" more points, which really doesn't make any sense since the points are useless. Just my way of thinking...
---------------------------------------- CJSL Gotta keep crunching... there's a world to save !!! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The quick answer to your question from me is "The points system seems pretty fair to me" (at least for CPU jobs; my GPUs are too slow to participate and I haven't researched how they're scored so I cannot comment on that). Different machine architectures have different strengths and weaknesses that we have no control over so if it worries you that much then you should only choose science apps that are suited to your machine!
I think the separate runtime measure is also a useful, though very different, indicator of contribution. I think when it was set up it was pretty fair. I think it's more debatable whether it still is now there are so many multi-threaded boxes out there. What's a good measure? Part of me says it's the number of hours the box is switched on and running science apps -- not the number of thread hours (or CPU hours, or processor chip hours) -- but I'm happy to accept things the way they are. Results returned is a pretty meaningless measurement in all respects as far as I can see as the length and complexity of the tasks varies so much from science app to science app, and some of our jobs have multiple steps or tasks within them too, so I ignore that one. Things are going to get more interesting as computing gets more truly heterogeneous. The ability to program task fragments to run on different engines (CPU, GPU, DSP etc.) as they're available and appropriate to the nature of the task will make the flop rather meaningless as a performance measure, but I'm sure some other measure will be derived that is as fair as it can be. And regarding your variable scores, I've certainly noticed that it takes the servers a while to get the measure of your machine when you switch to a new science app, and the averaging that goes on does seem to vary slowly over time (I assume that the techs have the ability to tweak scores to try to even things up across the different science apps) but that's all goodness as far as I can see. Sorry for the long post! |
||
|
|
johncmacalister2010@gmail.com
Veteran Cruncher Canada Joined: Nov 16, 2010 Post Count: 799 Status: Offline Project Badges: 
|
What is the best measure of work done? A Cure Oh, I think this is priceless!!!!!!! I agree completely- these sterile points discussions add precisely nothing. I don't care about points - let's get cures for all these horrible diseases being researched as well as the clean water and clean energy applications!!! Well said!!  crunching, crunching, crunching. AMD Ryzen 5 2600 6-core Processor with Windows 11 64 Pro. AMD Ryzen 7 3700X 8-Core Processor with Windows 11 64 Pro (part time)  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Oh, I think this is priceless!!!!!!! I agree completely- these sterile points discussions add precisely nothing. Then why respond? I don't care about the points either (does anyone?), but I do look at them to measure "work done". The OP question sounded reasonable.. |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: 
|
Does it really matter? You can't do anything with the points. I find the badges much more entertaining: they're like milestones or objectives. I really don't pay that much attention to the points. I've heard of people running one project vs another project (both inside and out of WCG) because it "gives" more points, which really doesn't make any sense since the points are useless. Just my way of thinking... CJSL Gotta keep crunching... there's a world to save !!! Maybe they think they are doing 'more'? |
||
|
|
|