Personal computing discussed

Moderators: farmpuma, just brew it!

 
matnath1
Gerbil Team Leader
Topic Author
Posts: 280
Joined: Thu Jan 27, 2005 3:21 am

Folding @ home PPD Explantion

Sun May 03, 2015 4:11 am

I am currently folding with two machines 24/7

i5 2400 gtx 970 45,000 ppd Alzheimers
Pentium G630 GTX 750 Cancer 14,000 ppd

Why am I seeing people posting astromical numbers such as 350,000 PPD for a Gtx 970 and 90,000 ppd for a GTX750?

http://www.overclock.net/t/475163/gpu-p ... d-database

I have simply downloaded and ran the "standard" folding @ home client and slid the Power Meter to FULL . Than selected alzheimers for the first one and cancer for the second rig.

Am I doing something wrong? Is there a way to dramtically increase my PPD on these sytems?

I know that link shows ppd from 2014 but I can not find anything current. Please send a link to more current PPD charts by GPU.
 
Kougar
Silver subscriber
Gerbil Jedi
Posts: 1955
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: Folding @ home PPD Explantion

Sun May 03, 2015 6:16 am

Yeah, that's really low. Sounds like you didn't get yourself a passkey. https://folding.stanford.edu/home/faq/faq-passkey/

The passkey confirms your identity and allows you to take advantage of the early return bonus system. The majority of current projects give a ERB bonus, and this comprises the majority of the PPD figures you are hearing about. You will need to read the entire FAQ for full details how to get and use it to get full PPD results for your systems.

Also, for newer GPU PPD figures you can find results in a several threads on the Stanford folding forums. EVGA's forum is also good for that, for the obvious reasons :wink:
 
farmpuma
Silver subscriber
Minister of Gerbil Affairs
Posts: 2538
Joined: Mon Mar 22, 2004 12:33 am
Location: Soybean field, IN, USA, Earth .. just a bit south of John .. err .... Fart Wayne, Indiana
Contact:

Re: Folding @ home PPD Explantion

Sun May 03, 2015 7:33 am

There are several reasons for ppd to vary by a considerable amount at any one time.

As Kougar mentioned the passkey early return bonus can boost ppd quite a bit depending the individual work unit and the speed of your hardware. You must complete ten work units with each instance of the client before the bonuses are awarded.

Any use of the computer for anything other than folding will throttle the client. In some cases the client production can be reduced to a crawl.

Each individual work unit will have a different ppd production. Within a given work unit core type the variation is usually fairly small. Variation between different core types (core 15 vrs core 17 for example) can vary considerably. Also GPU work units generally produce more ppd than SMP work units. My son folds with a 660ti and on 29 April he folded five work units for over 67,000 points, but on 30 April he folded five work units for just over 20,000 points.

I have yet to run any form of the version 7 client and was unaware that one could configure them by selecting a type of project. If maximum ppd is you ultimate goal (and who doesn't like lots of ppd) you need to be running a GPU client on both of your computers with an SMP client configured to run only three CPU cores on your i5-2400 system.

The folders who post pie-in-the-sky ppd are most likely over clocking their hardware to the ragged edge of stability and may quite possibly be using exotic cooling.

A list of the current active work units with lots of information including base points value -
http://fah-web.stanford.edu/new/psummary.html
Image Image
.* * M-51 * *. .The Whirlpool Galaxy.
 
matnath1
Gerbil Team Leader
Topic Author
Posts: 280
Joined: Thu Jan 27, 2005 3:21 am

Re: Folding @ home PPD Explantion

Sun May 03, 2015 8:05 pm

Thank you both for advising me on this. I ignored the passkey a while back as I thought it was just a way to "inflate numbers" for those who are competetive in the "sport" of folding @ home.

I did obtain and input passkeys into both of these folding machines and watched in shock as each one's PPD instantly rose by 3 to 5 times their prior PPD numbers:

The GTX 970 "rig" jumped from 45,000 ppd to 220,000
and
The GTX 750 "rig" jumped from 14,000 ppd to 45,000 ppd

Is this as obvious as it seems? Stanford is simply inflating the PPD numbers for those who Identify themselves with passkeys? (I guess they have their reasons).

I took some time off from folding. A few years back there were seperate clients for GPU and CPU but now the Folding @ home site says that their current "client" auto-detects and optimizes based on your hardware. All you have to do is move the slider between minimum and maximum settings. Is there a way to "max this out more"?

Thanks again for opening my eyes on this. I wonder if Stanford will retro my points?
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Folding @ home PPD Explantion

Mon May 04, 2015 6:03 am

Stanford wanted a secure method (in addition to the publicly visible, and therefore insecure, user name) to associate all systems under control of a single donor. This gives them leverage to punish cheaters (donors who install the client on unauthorized systems), since they can delete that donor's points, without worrying about people getting "framed" for cheating by someone installing clients under their user name without permission. To encourage people to use the passkey, they award a substantial points bonus for doing so; this makes cheating a much less attractive option for those who may have been so inclined.

From the user's perspective, IMO it makes more sense to view it as a penalty for *not* using a passkey...
Nostalgia isn't what it used to be.
 
farmpuma
Silver subscriber
Minister of Gerbil Affairs
Posts: 2538
Joined: Mon Mar 22, 2004 12:33 am
Location: Soybean field, IN, USA, Earth .. just a bit south of John .. err .... Fart Wayne, Indiana
Contact:

Re: Folding @ home PPD Explantion

Mon May 04, 2015 6:54 am

Also the passkey bonus points system is set up to encourage the fastest work unit returns possible, as Stanford can not perform a complete analysis on a given data set until all the work units of that data set are returned. Once the data set is analyzed they can then determine what simulation to test with the next data set.

Generally you only need one passkey per user name for all your clients, but it is possible that multiple passkeys for a given user name would also be OK.

Unless Stanford has changed the passkey early return bonus requirements they will not retro bonus points for your first ten work units.

It is possible that Stanford has refined the version 7 client to optimally configure a properly balanced GPU and SMP capable system. And it would certainly be less confusing than configuring said system with manual flag modifiers.
 
DPete27
Grand Gerbil Poohbah
Posts: 3079
Joined: Wed Jan 26, 2011 12:50 pm
Location: Wisconsin, USA

Re: Folding @ home PPD Explantion

Mon May 04, 2015 8:31 am

I only fold when TRFrankenbot fires up. Does Frankenbot have a passkey, or would a personal one still count toward Frankenbot?
Main: i5-3570K, ASRock Z77 Pro4-M, MSI RX480 8G, 500GB Crucial BX100, 2 TB Samsung EcoGreen F4, 16GB 1600MHz G.Skill @1.25V, EVGA 550-G2, Silverstone PS07B
HTPC: A8-5600K, MSI FM2-A75IA-E53, 4TB Seagate SSHD, 8GB 1866MHz G.Skill, Crosley D-25 Case Mod
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Folding @ home PPD Explantion

Mon May 04, 2015 9:57 am

DPete27 wrote:
I only fold when TRFrankenbot fires up. Does Frankenbot have a passkey, or would a personal one still count toward Frankenbot?

No, you need to use the TRFrankenbot one. DancinJack was historically the keeper of the passkey, but it looks like he didn't post on the 2014 TRFrankenbot thread so I'm not sure if he's participating any more. Any of the Distributed Computing forum "regulars" should be able to PM it to you when the time comes; given Stanford's motivations for using them in the first place, it should not be posted publicly.
Nostalgia isn't what it used to be.

Who is online

Users browsing this forum: No registered users and 2 guests