F@H old and new points calculations

[DarkJedi]

I've been away a while but I've started folding again for OcUK, I'm climbing back up the list . I was wondering what happened with the old stats for people...When I started F@H years ago, you got only 1 point for most WUs, now I see it's completely changed. My question I guess is - why weren't the old stats recalculated, because it's unfair, I'm sure we used to be higher up in the team rankings at one point, now all these new teams like CustomPC that have been around like barely a few years are killing us, I mean I was browsing through OcUK's list and come across old folders like 'foad' who has folded 17K WUs yet his points are only about 20k. If these hardcore folders had the stats recalculated we'd be much further up the Team rankings, I mean 20k points can be obtained in like a few dosen WUs with todays point calculated, basically the way I see it is it's unfair the original folders who dedicated so much.

thanks for anyone who can shed some light...

 

[SiriusB]

The new points are based on benchmarking carried out by F@H.

It would be massively time consuming to go through all the old WUs, bench them, give them a score, then trawl through EVERYONE'S stats and update them.

To be honest I don't think that many people care and it would be wasted effort since the science is far more important than making a few people happy.

 

[rich99million]

I can definitely see where you're coming from - when I started the project WUs were still worth less than a whole point (I remember a lot of 0.6 point WUs) and seemed to take pretty long too. I believe foad spent most of his time doing Genome@Home work which was always said to give lower points per unit CPU time than Folding@Home once the stats had been combined.

A lot has changed since then though, not only have new processing cores been produced and optimised with SSE/SSE2/Altivec but also computer hardware has greatly improved (as it always does) - Stanford's benchmark of 110PPD (or 220PPD if a much larger memory demand) on a P4 isn't really that crazy when you consider how much more scientific data is being processed on a modern machine with an optimised application.

I'll agree it does go more than a bit wrong when you get into the new clients that have been popping up over the last year - it doesn't seem fair that a C2D running the SMP client can get somewhere in the order of 10x that of running 2 standard clients even on the same machine, or that a GPU or PS3 can be run for a good 650-900 PPD.
These are new and exciting areas for the whole of DC and I can only guess that the points are there to thank users for pretty much not using their machine/gpu/PS3 for anything else in that time. It should also be noted these are the least stable clients and have the least amount of infrastructure server-side, beta testers have often had the benefit of more lucrative WUs maybe at the cost of losing some a fair few when something goes wrong.


As someone who's been with the project for a good 5-6 years now (losing count ) and who has mostly single-core machines without high-spec GPUs I definitely see your point.
Maybe Stanford should have revalued the existing points, or maybe even frozen the points and started again like SETI did when they switched to Boinc

In terms of the future the long awaited v6 has been in testing for a while now (not that I'd know any details) and it has been hinted that it will change the way points are awarded - I wish I had more to tell you on that, I can only hope they don't make a complete blunder of it


edit: of course you could always view the team stats sorted by number of units processed, though of course if someone had a bad run of EUEs at frame zero then that wouldn't help

 

[DarkJedi]

Thanks for the post rich99million, that's made things a little clearer. Also thanks for your time in making the team news thread every week - it's good to see how our individual team members are performing without having to spend ages looking at the EO stats.

SiriusB, I forgot about having to benchmark the old WUs, true it would take ages...my point was that most people do actually fold because of the stats, I'd like to see how many you would continue folding if the stats didn't exist anymore, I'm well aware of how important the science is...but if back in the day hardcore folders were told that soon points were not longer going to be about 1/WU, and we're now going to be much larger like 110/WU a lot would probably stop folding.