PPD?
- Thread starter marsey99
- Start date
I don't have any personal experience, but according to this spreadsheet a 965 BE @4 GHz will rake in about 7-7.5K PPD
That is ofcourse with the Quick Return Bonus (which you will have to qualify for by completing 10 SMP WU's within the preferred deadline and keep a 80% succes rate in your WU Returns).
Edit: This type of spreadsheet is by no means accurate, as is does not supply any information about the rest of the system. GPU's/ if it's dedicated or not/ any other thing that might run in the background or even differences in WU's/Projects. But it will give you an approximate number, so your number should be somewhere around this score.
That is ofcourse with the Quick Return Bonus (which you will have to qualify for by completing 10 SMP WU's within the preferred deadline and keep a 80% succes rate in your WU Returns).
Edit: This type of spreadsheet is by no means accurate, as is does not supply any information about the rest of the system. GPU's/ if it's dedicated or not/ any other thing that might run in the background or even differences in WU's/Projects. But it will give you an approximate number, so your number should be somewhere around this score.
Hehe, alright then
for you and everybody out there:
WU
Work Unit, the thing you are actually processing
Project
Well quite easy this one, all WUs are part of a project, right now I am processing WU's for the projects 7131, 6986, 6051, 6530, 6526 and 10721
All project consist of WUs, these are categorized in a system, which is indicated by PRCG (Project, Run, Clone, Generation) in that order
Every project will have multiple Runs, Clones and generations, the amount of these will be specified by the owner of the project, he will decide when he has all the data he needs.
QRB
Quick Return Bonus, a bonus which is applied to the fast return of SMP and some Uniprocessor WUs . The bonus is on a Exponential scale, the faster you return you WU’s the higher your bonus, and thus your PPD.
The QRB has to be earned though. There are several requirements for getting the QRB.
1: you must have completed 10 SMP WUs (or QRB capable WUs)
2: you must have a success rate of 80% or more.
3: you must return the WU within the Preferred Deadline (Timeout in v7)
Ok now for some explanation, Requirement 1 should be fairly simple right?
Requirement 2 however could be a problem. It is however pretty simple, basically the ratio between successful WUs (those completed completely and within the preferred deadline) should always be higher than 80 %. Got it? Good.
Requirement 3 is another easy one, not on time? Too bad, no bonus just the base points.
Now for the Deadlines, deadlines you say? Yup, plural, we have got multiple.
Preferred deadline
The first deadline for the WU, if Stanford doesn’t have the WU back on their servers when this is reached (for whatever reason) the WU will be issued again to a different donor. Stanford does this to not delay the project even longer. A new WU will be made out of the result of the last one in the series, so if there is a very slow user somewhere or a lost WU (due to a crash/reinstall or whatever reason) the project can still move on. When a WU is re-issued both of the users can still get the full points (well the base points that is.. QRB will only count for those returned before the Preferred Deadline). This is called the Timeout in v7.
Final Deadline
The absolute deadline for the return of a WU. If you don’t meet this one you will not be getting any points and the WU will be dumped from your PC and a new project will be picked up. This is called the Expiration in v7.
Ok, about the clients.
Basically there are 3 flavors. Uniprocessor, SMP (Symmetric MultiProcessing) and GPU.
Uniprocessor
The Uniprocessor takes advantage of.. You guessed it, 1 core (and nope, not like the name suggest 1 processor
) this is the one which all the slow single core processor will run. Everything above a P3 500Mhz, should be able to finish these WU's on time. Average PPD over here is something in the order of 130-160 PPD
Onwards we rock, SMP is up next..
As the name suggests it will run on multiple cores. And will rake in a lot more points than the Uniprocessor client, due to the higher point WU’s and the higher QRB’s (K factor). The ‘downside’ to this type of work is the short deadline (3 days before the Preferred Deadline (called Timeout in v7), and 4 days Final Deadline (called Expiration in v7)). It goes without saying that the deadlines will change with the different Projects as there will be larger and smaller ones, but the 3 and 4 day deadlines seems to be somewhat of a average. This will average a lot higher in PPD, my e6400 @ 3.0GHz is averaging about 2100 PPD which is a considerable boost over the Uniprocessor but still nothing compared to the massive multiprocessor systems.
The GPU client
Will this runs on the GPU’s. Nvidia 8xxx series and above, ATI 2xxx series and above. Do note, the 8400 GS has some serious problems, which due to the low amount of points and work is does will not be solved anymore. Also the older ATI cards will be phased out during this year. The OpenCL instructions on these cards cannot be used to run the new Fahcore_16, and since the older Fahcore_11 is a horrible performer it will be removed from duty around September this year.
What does this mean for all the ATI card owners? Well basically any card under the 5xxx series will not be supported anymore. There will be no more WU’s made for the older Fahcore_11.
The Average PPD will differ enormously due to the large difference in GPU’s and their shaders (which do the work)
Ok this kinda only scratches the surface, but it should be some more info for you.
TL;DR? Ah well, your loss
Btw, if someone is interested in this, I could make an longer version of this, which will explain more. And go deeper into the different clients and everything else.
Ow and Marsey, call me Shon mate
WU
Work Unit, the thing you are actually processing
Project
Well quite easy this one, all WUs are part of a project, right now I am processing WU's for the projects 7131, 6986, 6051, 6530, 6526 and 10721
All project consist of WUs, these are categorized in a system, which is indicated by PRCG (Project, Run, Clone, Generation) in that order
Every project will have multiple Runs, Clones and generations, the amount of these will be specified by the owner of the project, he will decide when he has all the data he needs.
QRB
Quick Return Bonus, a bonus which is applied to the fast return of SMP and some Uniprocessor WUs . The bonus is on a Exponential scale, the faster you return you WU’s the higher your bonus, and thus your PPD.
The QRB has to be earned though. There are several requirements for getting the QRB.
1: you must have completed 10 SMP WUs (or QRB capable WUs)
2: you must have a success rate of 80% or more.
3: you must return the WU within the Preferred Deadline (Timeout in v7)
Ok now for some explanation, Requirement 1 should be fairly simple right?
Requirement 2 however could be a problem. It is however pretty simple, basically the ratio between successful WUs (those completed completely and within the preferred deadline) should always be higher than 80 %. Got it? Good.
Requirement 3 is another easy one, not on time? Too bad, no bonus just the base points.
Now for the Deadlines, deadlines you say? Yup, plural, we have got multiple.
Preferred deadline
The first deadline for the WU, if Stanford doesn’t have the WU back on their servers when this is reached (for whatever reason) the WU will be issued again to a different donor. Stanford does this to not delay the project even longer. A new WU will be made out of the result of the last one in the series, so if there is a very slow user somewhere or a lost WU (due to a crash/reinstall or whatever reason) the project can still move on. When a WU is re-issued both of the users can still get the full points (well the base points that is.. QRB will only count for those returned before the Preferred Deadline). This is called the Timeout in v7.
Final Deadline
The absolute deadline for the return of a WU. If you don’t meet this one you will not be getting any points and the WU will be dumped from your PC and a new project will be picked up. This is called the Expiration in v7.
Ok, about the clients.
Basically there are 3 flavors. Uniprocessor, SMP (Symmetric MultiProcessing) and GPU.
Uniprocessor
The Uniprocessor takes advantage of.. You guessed it, 1 core (and nope, not like the name suggest 1 processor
Onwards we rock, SMP is up next..
As the name suggests it will run on multiple cores. And will rake in a lot more points than the Uniprocessor client, due to the higher point WU’s and the higher QRB’s (K factor). The ‘downside’ to this type of work is the short deadline (3 days before the Preferred Deadline (called Timeout in v7), and 4 days Final Deadline (called Expiration in v7)). It goes without saying that the deadlines will change with the different Projects as there will be larger and smaller ones, but the 3 and 4 day deadlines seems to be somewhat of a average. This will average a lot higher in PPD, my e6400 @ 3.0GHz is averaging about 2100 PPD which is a considerable boost over the Uniprocessor but still nothing compared to the massive multiprocessor systems.
The GPU client
Will this runs on the GPU’s. Nvidia 8xxx series and above, ATI 2xxx series and above. Do note, the 8400 GS has some serious problems, which due to the low amount of points and work is does will not be solved anymore. Also the older ATI cards will be phased out during this year. The OpenCL instructions on these cards cannot be used to run the new Fahcore_16, and since the older Fahcore_11 is a horrible performer it will be removed from duty around September this year.
What does this mean for all the ATI card owners? Well basically any card under the 5xxx series will not be supported anymore. There will be no more WU’s made for the older Fahcore_11.
The Average PPD will differ enormously due to the large difference in GPU’s and their shaders (which do the work)
Ok this kinda only scratches the surface, but it should be some more info for you.
TL;DR? Ah well, your loss
Btw, if someone is interested in this, I could make an longer version of this, which will explain more. And go deeper into the different clients and everything else.
Ow and Marsey, call me Shon mate
Hehe, alright thenfor you and everybody out there:
WU
Work Unit, the thing you are actually processing
Project
Well quite easy this one, all WUs are part of a project, right now I am processing WU's for the projects 7131, 6986, 6051, 6530, 6526 and 10721
All project consist of WUs, these are categorized in a system, which is indicated by PRCG (Project, Run, Clone, Generation) in that order
-snip-
You should put this in it's own thread and expand on it more if you like. This would make a great sticky. It explains a lot in detail and it would be easier to point someone to the topic instead of explaining it all over again. Nice job.
Edit: Just keep it easy to read like this. So many F@H guides etc get really boring and complicated really fast.
dugdiamond
New member
it will be a few days before you get your bonus
yea i think this has crashed lol, nothing today dont sound right to me :|
edit
05.27, 9pm 0 23
05.27, 6pm 552 1
so it stopped running 3am last night?
http://folding.extremeoverclocking.com/user_summary.php?s=&u=559046
bugger
edit
05.27, 9pm 0 23
05.27, 6pm 552 1
so it stopped running 3am last night?
http://folding.extremeoverclocking.com/user_summary.php?s=&u=559046
bugger
ow boy, 23 WUs and 0 points.
Indeed sounds like either a bad WU, or an unstable overclock. To be sure though, post the log of your client.
Alright, I'll see what I can come up with. Are there any special requests for things that have to be in there?
Indeed sounds like either a bad WU, or an unstable overclock. To be sure though, post the log of your client.
You should put this in it's own thread and expand on it more if you like. This would make a great sticky. It explains a lot in detail and it would be easier to point someone to the topic instead of explaining it all over again. Nice job.
Edit: Just keep it easy to read like this. So many F@H guides etc get really boring and complicated really fast.
Alright, I'll see what I can come up with. Are there any special requests for things that have to be in there?
dugdiamond
New member
drop your OC down - and set an OC-profile just for folding (i have)
