Why am I only getting 100 points per WU while this guy gets 1700?

Status
Not open for further replies.
What hardware are you using? If you have a dual core processor or multiple dual core processors, you should be using the SMP client.

The normal, console client is a bit more stable, however gets a much lower point value for the work units it computes. The SMP client will use 100% of your computer's processor while the console client will only work on one core at a time.

If you tell us what hardware you have we can help you get the most out of it.

 
Maybe you should ask on your team forums? And calling someone else on your team a douche on another teams site is just bad form man.

As for the point difference you are running normal clients like the 5.04 to newer v6 clients set for single cpu machines. The person you are referring to is running the SMP client which is meant for quad core cpus and is also ran on dual core cpus. So it the SMP client is using 2-4 cpus and turning out a WU every 24hours or so. Giving him 1760 or so points a day. Where you are running four single cpus using the single client turning 130-300 points per cpu per day approximately.

Hope this helps somewhat.





 
I agree with Majic, be a gentleman and don't call teammates a douche since everyone is working toward the same goal.

 
My last WU was worth 1760 too. I think that one came from my e2140 running @ 3.0ghz but I have the proc running at 100% and two instances. Moved from 12,521 to 14,281 today.

 
Xilikon said:
I agree with Majic, be a gentleman and don't call teammates a douche since everyone is working toward the same goal.

Click to expand...

QFT

OP,

We/I are all about competition, however name calling is, well unbecoming this project and it's true goal.

If you can run SMP.

For further support get a bar of soap insert in mouth and change team to 33.

Have a nice day.

 
Majic said:
Maybe you should ask on your team forums? And calling someone else on your team a douche on another teams site is just bad form man.

As for the point difference you are running normal clients like the 5.04 to newer v6 clients set for single cpu machines. The person you are referring to is running the SMP client which is meant for quad core cpus and is also ran on dual core cpus. So it the SMP client is using 2-4 cpus and turning out a WU every 24hours or so. Giving him 1760 or so points a day. Where you are running four single cpus using the single client turning 130-300 points per cpu per day approximately.

Hope this helps somewhat.


Click to expand...
But 4 X 300=1200, not 1700.
 
kazak said:
My last WU was worth 1760 too. I think that one came from my e2140 running @ 3.0ghz but I have the proc running at 100% and two instances. Moved from 12,521 to 14,281 today.

Click to expand...
But was that 1760 from ONE WU. Cause from what I'm understanding, if you're using the SMP client, instead of outputting one WU per core, you get one WU from 4 cores/processors. You just said you're running two instances, that would mean that you're getting two WU. If you're getting two WU (one for each core) and each are worth 1760, then I have every right to be pissed, if those two WU equal 1760, I still have a right to be pissed since 1760/2=880, much higher than I've ever gotten from one WU. Furthermore, if you're only getting one WU for two processors/core, that's still 880 points per processor/core.

Oh and here is an update:
03.10, 6pm 685 5
I only got 685 points yet I had 5 WU on that!!!! ARGHHHH!!!!!
 
My one Q6600 is producting 5000 points per day with 2 SMP clients. A single SMP can produce 1760 points wu in a bit less than 24 hours.

You can get this if you have a dual-core or better, running a SMP client. No need to be pissed off, the SMP client is your best choice for getting 1760 pointers (which is 98% of all the SMP WU).

 
If you have a machine with a single core processor, the WU's that get assigned are only worth between 15 and 300 points. However if you have a machine with either 2 processors, a dual core or quad core processor, you can run the SMP client, which a single WU can be worth 1400 to 1760 points. Combine that with the speed of the dual and quad core procs, and you could be pulling in the 1760 points in a single day. I have a dual core machine that routinely turns in 1760 pointers ever 32 hours. Does this answer your question?

What hardware are you running?



 
jljenkins28 said:
If you have a machine with a single core processor, the WU's that get assigned are only worth between 15 and 300 points. However if you have a machine with either 2 processors, a dual core or quad core processor, you can run the SMP client, which a single WU can be worth 1400 to 1760 points. Combine that with the speed of the dual and quad core procs, and you could be pulling in the 1760 points in a single day. I have a dual core machine that routinely turns in 1760 pointers ever 32 hours. Does this answer your question?

What hardware are you running?



Click to expand...
Yeah I'm running individual single core machines. That's total bullshit though, I have to putforth twice as much effort/electricity and I'm only get a fraction of the reward. Why should someone with multiple cores get more points than someone with multiple machines? It's a much more concerted effort to have multiple machines working on a WU than to have one machine with multiple cores.
 
Actually points are based on the amount of work the cores can do. A core2duo is exponentially more powerful then a pentium 4 so point values the core2. A small garden of powerful systems will always produce more then a large farm of single cores. Helps us to keep efficiency at a maximum.




 
Connundrum1 said:
Actually points are based on the amount of work the cores can do. A core2duo is exponentially more powerful then a pentium 4 so point values the core2.




Click to expand...
I was under the impression that the slower the system, the longer it takes to complete the same WU so it's all fair. Faster machines complete WU faster, slower machines slower.. Why should the faster machines get WU with more points? Also didn't they say their baseline machine is a P4 2.8C?
 
<sigh>

The SMP WU are more complex, worth more to Stanford. They give a higher point value to the so we will crunch the bigger and harder WU instead of running a bunch of normal clients crunching the small WU.

Stanford has a carrot (points) and they make it bigger and smaller depending on what they want us to focus on working for them.

If the SMP was not such a huge bonus for us we would all be running single core systems still (at least for our farms) and sanford would not be getting the work done that it needs.

 
All WU's are not created equal however. SMP (multi core units basically) are worth alot more since the data is worth more to Stanford. The WU's are also very time sensitive, the faster Stanford gets the results the better. So this score is also based off the speed in which it can be processed and returned.

I assure you that if you sell off a few single core boxen and consolidate to a few dual cores or quad cores you will save money (in the form of electricity, # of computers to maintain etc.) space and headaches.

edit: Kendrak got to it before I did. They're ninjas i tell you, the lot of them!

 
imzjustplayin said:
But was that 1760 from ONE WU. Cause from what I'm understanding, if you're using the SMP client, instead of outputting one WU per core, you get one WU from 4 cores/processors. You just said you're running two instances, that would mean that you're getting two WU. If you're getting two WU (one for each core) and each are worth 1760, then I have every right to be pissed, if those two WU equal 1760, I still have a right to be pissed since 1760/2=880, much higher than I've ever gotten from one WU. Furthermore, if you're only getting one WU for two processors/core, that's still 880 points per processor/core.

Oh and here is an update:
03.10, 6pm 685 5
I only got 685 points yet I had 5 WU on that!!!! ARGHHHH!!!!!
Click to expand...

Well that's the only time I got that much from a single WU. And I don't have SMP running on that box. On average, I only get about 200ish per WU and the two new WUs showing on my stats page only netted me 358.

I did install the SMP client and that Affinity app on another box running a Xeon 5140 but it's only at 58%. It is crunching at a briskly 12min per frame. The rest of my machines are probably crunching at 25min at best but probably around 30min-ish.

edit: it turns out that smp box did finish that one WU worth 1760! Awesome.

 
Kendrak said:
<sigh>

If the SMP was not such a huge bonus for us we would all be running single core systems still (at least for our farms) and sanford would not be getting the work done that it needs.

Click to expand...
That's not true, we'd instead be running multiple instances, one for each of our cores/processors. I just don't think it's fair that multicore systems get more points per processor than single core systems.
 
imzjustplayin said:
That's not true, we'd instead be running multiple instances, one for each of our cores/processors. I just don't think it's fair that multicore systems get more points per processor than single core systems.
Click to expand...

Life isn't fair, sorry some one had to say it.

Save your pennies and get a Q6600, or stop complaining.

Every bit helps the cause and your moaning cheapens it.

There are many here in the [H]orde that only have single core systems. They are valued with everyone else. There are some of us who have made it a hobby or have other reasons to spend our hard earned $$ on it. That makes us no better or worse.

I suggest you be happy with your system(s), or upgrade accordingly.

Have a nice day.
 
Just wondering since we are doing math here how long does it take to run a 300 pointer on a 3ghz+ cpu like saying a single core less then 24 hours? I am not sure cause I dont have any machines that fast to time it out. i know on my older 2ghz xp's it takes them just over 24hours or so not really sure of the number.
 
Let me chime on in what I have to say about this, I have highlighted some of the things that I think you need to concern yourself with. And if you cannot take the fact that somebody with only one computer whips your collective of ten or so, then you are not doing folding for the right reason but for your own damn e-penis. We all here support each other no matter of their level in the "game" of folding. As long we all are in the game of doing what we believe is right, finding cures, we welcome all into our circle and shun the others out who doesn't support the concept but just for the points and the e-penis (points are good, we welcome competitions but not in the way you brought it up.) Cut yourself some slack and get your ass in gear, and contribute away regardless of points or UPGRADE your little farm which you despite so much.

You spoke of the baseline for the regular cores but not for the SMP cores.

Excerpts from Stanford.
http://folding.stanford.edu/English/FAQ-Points
How do you decide how much credit a work unit is worth?

Points are determined by the performance of a given machine relative to a benchmark machine. Before putting out any new work unit, we benchmark it on a dedicated 2.8 GHz Pentium 4 computer with SSE2 disabled. (more specifically, as reported by /proc/cpuinfo on linux: vendor_id : GenuineIntel, cpu family : 15, model : 2, model name : Intel(R) Pentium(R) 4 CPU 2.80GHz, stepping : 9, cpu MHz : 2806.438, cache size : 512 KB). This machine runs linux, so all WUs are benchmarked with the linux core. Note: Currently the linux and Windows fah_cores run at the same speed, so this does not impact points performance.

We plug the results of this into the following formula:

points = 110 * (daysPerWU)

where daysPerWU is the number of days it took to complete the unit. This equation was chosen to match the points for previous Gromacs WUs from the previous point system. The upshot is that Tinker WUs will be worth more than before we set up the new point system (i.e. before April 2004).
Click to expand...

Why are some projects given significantly more points than others?

Certain projects require substantially more computer resources than others, either in terms of more disk space, more network transfer, or more RAM used. By default, these work units are given out to clients that opt in to request them. To reward those contributors for donating resources beyond the typical client, we currently give bonus points for these larger work units.
Click to expand...

How big are bonus points?

Currently the bonus points are a 50% increase over the standard benchmark point determination (described above). Please note that this value is subject to change.
Click to expand...

Excerpts from Stanford
http://folding.stanford.edu/English/FAQ-SMP#ntoc22
How do you decide the credit value of SMP work units?

Points are determined by the performance of a given machine relative to a benchmark machine, similar to the CPU client benchmark process. Before releasing any new project (series of work units), we benchmark it on a dedicated Macintosh Pro with 2 - 2.33 GHz Dual Core Xeon processors. (more specifically, 2 Woodcrest 5140 processors with 4 MB cache (each), 5 GB FBDIMM Memory (667 MHz DDR2), 1.33 GHz Bus)

We plug the results of this benchmark test into the following formula:

points = 1760 * (daysPerWU)

where daysPerWU is the number of days it took to complete the work unit.

Please note the very concept of a reference machine will mean that some WU performance will vary from the performance on your machine. Even between various Xeon processors, there are significant differences in architectures. Moreover, there are variations between WUs within a given project which can lead to speed differences.

Our goal is consistency within a given definition of a reference machine setup (described above), but beyond that, the natural variation from machine to machine and WU to WU will never allow any point system to perfectly predict what you get on your machine.
Click to expand...

Why is the SMP client important, and why is the benchmark set at that level?

The purpose of the SMP client is twofold: to take advantage of the high-performance capabilities of recent multiprocessor systems and to help develop a simulation architecture that will become one of the dominant FAH computing paradigms as multi-core chips become an industry standard over the next several years. High-performance clients enable us to run types of calculations that would be impractical on our standard architecture--calculations that enhance our scientific capabilities, and your scientific contributions, significantly.

High-performance clients often require more computing resources. SMP clients typically run on dedicated systems, 24 hours a day, and use more processing power, more disk space, more network resources, more system memory, etc. Also, a major part of the scientific benefit is dependent on rapid turnaround of work units; hence we assign short deadlines for SMP work units. To reward those contributors for donating resources beyond the typical CPU client, for completing these work units very quickly within the short deadlines, and for contributing to the development of our next-generation capabilities, we currently set a benchmark value (with included bonus*) proportional to these larger more demanding SMP work units. Without the SMP clients and your additional contributions, we would not be able to complete many important projects. *Please note the bonus value is subject to change.
Click to expand...

 
Majic said:
Just wondering since we are doing math here how long does it take to run a 300 pointer on a 3ghz+ cpu like saying a single core less then 24 hours? I am not sure cause I dont have any machines that fast to time it out. i know on my older 2ghz xp's it takes them just over 24hours or so not really sure of the number.
Click to expand...

This widely depends on the CPU you have, for instance, if you were comparing one running on a core of P4 at 3Ghz and one on a core of C2D at 3Ghz, the differences are vastly big. We have gone beyond the speed of the cpu itself and more into the design of the cpu itself. I don't know how I can answe you but I can say that on some of my borgs in the lab, they aren't even 3Ghz and they are capable of finishing regular clients in less than 24 hours if left on for an entire day.
 
Well that was really what i was looking for to see if reg/normal cpus are finishing single core w/u's in under 24 hours. Cause if they were only taking say 20hours then over a bit of time the points would really seem to even out not counting bonus units and the like.
 
Majic said:
Well that was really what i was looking for to see if reg/normal cpus are finishing single core w/u's in under 24 hours. Cause if they were only taking say 20hours then over a bit of time the points would really seem to even out not counting bonus units and the like.
Click to expand...

This all depends on the units for the regular clients, as seeing there are a whole lot more variety of units for the regular client rather than any other configurations.
 
Jon855 said:
Let me chime on in what I have to say about this, I have highlighted some of the things that I think you need to concern yourself with. And if you cannot take the fact that somebody with only one computer whips your collective of ten or so, then you are not doing folding for the right reason but for your own damn e-penis. We all here support each other no matter of their level in the "game" of folding. As long we all are in the game of doing what we believe is right, finding cures, we welcome all into our circle and shun the others out who doesn't support the concept but just for the points and the e-penis (points are good, we welcome competitions but not in the way you brought it up.) Cut yourself some slack and get your ass in gear, and contribute away regardless of points or UPGRADE your little farm which you despite so much.

You spoke of the baseline for the regular cores but not for the SMP cores.

Excerpts from Stanford.
http://folding.stanford.edu/English/FAQ-Points






Excerpts from Stanford
http://folding.stanford.edu/English/FAQ-SMP#ntoc22




Click to expand...
Yeah I was familiar with the formula on top but not the new one for the SMP client.
 
MrGuvernment said:
Did you choose to allow big WU or not...
Click to expand...
of course, not to mention I used "advmethods" setting every time. I was wondering, how do they determine which systems get what WU? I noticed some WU have long due dates, in a matter of months while others have due dates for a matter of days. Is this random or do they actually know which systems have a fast turn around and which ones don't? I've noticed that my laptop is constantly getting 3306 project which gives 137 points. This has a 5 day "desired" turn around, while my desktop has 2452 which has a 46 day "desired" turn around.

Also how come I can't access this http://fah-web.stanford.edu/psummary.html page from clicking links via the folding.stanford.edu website? Not to mention, do they have a points summary for all of the beta projects since that one does not include the beta ones anymore? I know this page exists but I can't find it through normal means which totally sucks.
 
Also, don't forget that one reason for the big bonus on SMP WU is the very very tight deadlines (3-4 days at most) compared to the regular WU, which usually get 2-3 months deadlines. If you don't let the computer run 24/7, you will always miss the deadlines.

imzjustplayin, you need to change the mentality and think of folding as your contribution toward the cure, not a way to measure your e-penis because I would kick anyone who think like that out of the team. We treat big producters and casual folders in the same respect because a WU is a WU and everything is a useful contribution. The regular WU is also valuable to them. If you just want to jump in the SMP bandwagon, just upgrade your computer.

 
imzjustplayin said:
That's not true, we'd instead be running multiple instances, one for each of our cores/processors. I just don't think it's fair that multicore systems get more points per processor than single core systems.
Click to expand...

HUH? Its 4 processors and if you compare a P4 core vs a C2D Core it is about 66% faster than a P4 even if the P4 is clocked at 3.9ghz vs a 3ghz C2D.

So if you logically look at the work being produced the Q6600 is producing 8-10 times the output that a P4 is and a E6600 is producing 4-5+ times the output that a P4 can do. Thus its almost 5 x 300 = 1500 plus the bonus for working on the harder/larger scientific projects which eat more resources... so 1760 points.

If you can't live with it upgrade and get [H]arder! :)

 
Xilikon said:
Also, don't forget that one reason for the big bonus on SMP WU is the very very tight deadlines (3-4 days at most) compared to the regular WU, which usually get 2-3 months deadlines. If you don't let the computer run 24/7, you will always miss the deadlines.

imzjustplayin, you need to change the mentality and think of folding as your contribution toward the cure, not a way to measure your e-penis because I would kick anyone who think like that out of the team. We treat big producters and casual folders in the same respect because a WU is a WU and everything is a useful contribution. The regular WU is also valuable to them. If you just want to jump in the SMP bandwagon, just upgrade your computer.

Click to expand...

QFT its all fun to be competitive but keep in mind whatever the contribution its for a good cause! :cool:

 
imzjustplayin said:
of course, not to mention I used "advmethods" setting every time. I was wondering, how do they determine which systems get what WU? I noticed some WU have long due dates, in a matter of months while others have due dates for a matter of days. Is this random or do they actually know which systems have a fast turn around and which ones don't? I've noticed that my laptop is constantly getting 3306 project which gives 137 points. This has a 5 day "desired" turn around, while my desktop has 2452 which has a 46 day "desired" turn around.

Also how come I can't access this http://fah-web.stanford.edu/psummary.html page from clicking links via the folding.stanford.edu website? Not to mention, do they have a points summary for all of the beta projects since that one does not include the beta ones anymore? I know this page exists but I can't find it through normal means which totally sucks.
Click to expand...

The first time you run the client it states something like Benchmarking....

That is when the SMP looks at your hardware output and estimates what WU to give you. (At least that is my understanding) Thus people with Q6600's always get the ginormous WU's to process, if they elected to get large WU and Scientific WU.

Hope you get your folding situation hammered out and get happy with the production or decide to upgrade if you can! :)

Its addictive, trust me, I've add two Q6600's in the last 2 weeks.


 
To get a analogy : You donate 100$ to the salvation army and you see a rich guy come beside you and donate 1 million $ to them. The right thing to say would be "Great donation, the Salvation Army will need it" but instead of this, your own way is "It's not fair, why can he donate 1 million and not me ??". This is just some selfish thinking.

Personnally, I do envy those who have a few Q6600 in the farm but I need to have my priorities straight and don't waste more money than you earn. Not everyone is that rich and not everyone will spend a lot of money just for folding ;)

 
My guess is he is a younger person, and he doesn't see it for the good of humanity he sees it as a competition with his friends... I could be wrong.

He should just be proud he is helping, as doing something is better than nothing!

As you get [H]arder you do look at folding slightly different. I mean if I add another Quad box I'm going to take my P4 and my PS3 out of the folding picture. Why? Cuz the production per watt isn't there and it is cheaper for me to run another Q6600 then those two machines and I will do more for the cause.

Also I can use them on occassion when I need a TURBO button :) When people like Mflop start getting on my threat list! :p

 
It is time like this I'm glad I fold for the [H].

Good cause, good team, great members!

Fold on!

 
Sunin said:
HUH? Its 4 processors and if you compare a P4 core vs a C2D Core it is about 66% faster than a P4 even if the P4 is clocked at 3.9ghz vs a 3ghz C2D.

So if you logically look at the work being produced the Q6600 is producing 8-10 times the output that a P4 is and a E6600 is producing 4-5+ times the output that a P4 can do. Thus its almost 5 x 300 = 1500 plus the bonus for working on the harder/larger scientific projects which eat more resources... so 1760 points.

If you can't live with it upgrade and get [H]arder! :)

Click to expand...
You realize you've said a whole lotta nothing, right? your "math" has no basis and is mostly unfounded. lol
 
Sunin said:
The first time you run the client it states something like Benchmarking....

That is when the SMP looks at your hardware output and estimates what WU to give you. (At least that is my understanding) Thus people with Q6600's always get the ginormous WU's to process, if they elected to get large WU and Scientific WU.

Hope you get your folding situation hammered out and get happy with the production or decide to upgrade if you can! :)

Its addictive, trust me, I've add two Q6600's in the last 2 weeks.


Click to expand...
Where is this benchmark information stored? Client.cfg?
 
Sunin said:
My guess is he is a younger person, and he doesn't see it for the good of humanity he sees it as a competition with his friends... I could be wrong.

He should just be proud he is helping, as doing something is better than nothing!

As you get [H]arder you do look at folding slightly different. I mean if I add another Quad box I'm going to take my P4 and my PS3 out of the folding picture. Why? Cuz the production per watt isn't there and it is cheaper for me to run another Q6600 then those two machines and I will do more for the cause.

Also I can use them on occassion when I need a TURBO button :) When people like Mflop start getting on my threat list! :p

Click to expand...
Oh shut up and get off your high horse. I'm not complaining about people with systems that have better performance, I'm complaining that there is an unequal disparity in the points being delivered. I don't like the idea that running single instances yields fewer points than one SMP client. If I were to run 4 single instances on a Core2Quad, it would give me fewer points than one SMP client, that's not how it should be, it should be relatively equal. Anyhow what stanford needs to do now is bring out a client that works on clusters of machines, instead of having to run an individual instance per machine. The current SMP client isn't really even SMP in the first place, it's based off of technology intended for clustering in the first place!
 
Status
Not open for further replies.
Back
Top