Tobit Wants YOU

T

Tobit

[H]ard|DCer of the Month - March 2010/May 2011
Joined
Dec 12, 2005
Messages
5,145
.. to consider running at least one classic client.

As some of you know, I've always been a proponent of running as much variety of clients as you can to best benefit science. I was talking with sirmonkey this evening on our IRC channel and he seemed surprised to know that there is one work server at Stanford with over 2 million classic client work units available. That is just one server and there are several classic client servers!

Now I know the points suck, believe me I do. However, the work still needs to be done at some point and I personally, along with Stanford I'm sure, would appreciate it if more people could devote at least one core to the classic client. If you have an older computer just sitting around, I beg you to have one less Latte a day and, instead, spend the money on electricity to turn it back on 24/7 to crunch a classic WU.

Here at teamtobit.net, I am running four classic clients at the moment and I will be adding a few more in the near future as I am a beta tester for Stanford and want to run a few more as there are always new classic client projects coming out. In the past two weeks, there has been over a hundred new ones released to public.

I am currently running a classic client on the following:

  • Xeon 3060 @ 2.4 GHz - Two cores running, averaging about 190 PpD per core.
  • 3.0 GHz Pentium - this is my mothers computer that I borged, she leaves it on 24x7 anyway so it might as well fold in the background. Right now it is doing 175 PpD but will sometimes do 250 depending on WU.
  • 2.2 GHz T4400 Pentium Dual Core - this is my dads laptop, always plugged in 24x7 sitting next to his recliner. It may as well fold on a single core as well, right? It is currently doing 250 PpD.

Anyway, if you have an older system available. PLEASE consider plugging it in, loading up a client, set it in the corner of the room and, like I do, forget about it so it can, slowly but surely, bust out some uni-processor work. If I could offer some type of incentive, I would... but I can't. However, lets just stop thinking of point mongering for a bit and get back to the science. Although we are all trying to maximize PPD, we are also here for the science in some capacity or we wouldn't be folding at all. Do not change any of you current systems but let's try and give back a bit to science by turning on some older cores if we have them.

Thanks to the best F@H Team there is, ever was, and ever will be.
 
Last edited:
hmmm

ill see what ive got laying around, maybe if i can fix the freezing issue on my Athlon X2 lappy ill set it to do some stuff......does the client run in linux ?
 
If I can find a good Q6600 to put in my dads computer, I'll give it a shot.
 
I've had my HTPC running the classic client for a week or so, since I didn't really want to put the load of the smp client on it (my room is hot enough already), plus I wasn't sure it would make the deadlines anyways. You weren't kidding when you said the points suck. Getting a whopping 139.8 PPD out of a 10019 on a 45nm Athlon X2 5000+.
 
team Tobit? next thing you know your gonna want me to use a different team number with your user id. ;) Ha i knew my old 18 pc 25ghz farm would come back in style someday woot. 2 gateway server's duel 1gig pIII's, 2 abit vp6's duel pIII's thats pentium not phenom. dozen athalons and xp's and lets not forget the world record Abit BX-133 raid! had the fastest gf2mx in 3dmark2000 ah memories. on a serious note i do have a couple running. now gotta check the parts bin to see what ealse i have.
 
Since I don't have my i7 doing any processor folding because my room gets too hot with just the video cards running I figured I would put it on. Its really not effecting the processor temperature and I'm getting a whopping 373ppd on this WU. Not too shabby compared to everyone elses numbers :p
 
Cypher- said:
Since I don't have my i7 doing any processor folding because my room gets too hot with just the video cards running I figured I would put it on. Its really not effecting the processor temperature and I'm getting a whopping 373ppd on this WU. Not too shabby compared to everyone elses numbers :p
Click to expand...

only 7 clients to go for that pc
 
I'll maybe get my dads PC going with the classic should he ever decide to fix it
 
W.Feather said:
hmmm

ill see what ive got laying around, maybe if i can fix the freezing issue on my Athlon X2 lappy ill set it to do some stuff......does the client run in linux ?
Click to expand...


RMclock utility.. best program ever.. the mobile processors will undervolt like beasts.. i got my turion x2 TL-60 from 1.175v to .985v running full load no problem.. dropped the temps 22C..

also got my athlon xp-m 2400+ from 1.1v to .900v
 
@ tobit

Does classic run on linux?
Would it run on like damn small linux or something extremely small?

What im thinking are some linux VMs, maybe run like 500 of them on one of my duallies.
 
I wonder why Stanford dosent weight them more? Do they consider the gpu and smp work units more valuble?
 
I don't understand this. Doesn't the point system exist to guide our decisions about what hardware to use and which clients to run? The Stanford page states that points are given out in proportion to the contribution you're making to the project.

Surely if they needed more people to run the uniprocessor client for a while, they would just increase the number of points you get for running it?
 
I have a pentium D that is running my 8800gts and I have one core with the classic client running. Getting about 146PPD per day, but I figured due to low points the work was backing up :(.
 
Maybe I'll have my dual-core laptop work on them again. I shut the clients down early in the year since I moved exclusively to SMP. It's a 45nm Core2 based processor @2.2GHz but it's on only a couple of hours each day, though.

sirmonkey1985 said:
also got my athlon xp-m 2400+ ...
Click to expand...
Got several old Socket-A Athlon processors but no working boards to use them in.

MrValentine said:
I wonder why Stanford dosent weight them more? Do they consider the gpu and smp work units more valuble?
Click to expand...
That's a long story and you'll likely hear conflicting answers from Stanford. The truth is God only knows... :confused:
 
thefreeaccount said:
I don't understand this. Doesn't the point system exist to guide our decisions about what hardware to use and which clients to run? The Stanford page states that points are given out in proportion to the contribution you're making to the project.

Surely if they needed more people to run the uniprocessor client for a while, they would just increase the number of points you get for running it?
Click to expand...


problem is to be fair they cant increase them.. if they did increase them all hell would break loose on the foldingforum.. when they were thinking about modifying the Kfactor(multiplier used for bonus system) on one of the SMP A3 WU's mid way through the WU run everyone went nuts because all those people that ran it before wouldnt of gotten those extra points from the Kfactor being changed so they decided not to do it..

while yes the point system is there to get people to run certain WU's more.. there are a shit ton more systems out there not capable of running the SMP client.. but the uniprocessor is an old client thats pretty much been forgotten over the last year and half.. it use to be the only client cpu client til SMP started.. while the points suck ass on them.. the WU's can pretty much run on anything.. throw a bunch of old pentium II's or pentium III's or hell even old athlon XP systems in a corner of a room naked and just let em fold the uniprocessor WU.. probably setup 5-6 systems run 24/7 and barely use the same amount of power as anyones SMP rigs.. at some point here in then next 3-4 months thats what ill be doing since i have a bunch of socket 754 and socket A boards laying around in boxes not being used..


APOLLO said:
Maybe I'll have my dual-core laptop work on them again. I shut the clients down early in the year since I moved exclusively to SMP. It's a 45nm Core2 based processor @2.2GHz but it's on only a couple of hours each day, though.

Got several old Socket-A Athlon processors but no working boards to use them in.

That's a long story and you'll likely hear conflicting answers from Stanford. The truth is God only knows... :confused:
Click to expand...


couple hours each day is fine since the uniprocessor WU's have a 92 day deadline or something like that..
 
I have 2 p4 2.6ghz machines with HT. I will put the classic client on these as they are already on 24x7 anyways, just need to put some as5 on the heatsink instead of what came on it from intel.
 
Last edited:
thefreeaccount said:
I don't understand this. Doesn't the point system exist to guide our decisions about what hardware to use and which clients to run? The Stanford page states that points are given out in proportion to the contribution you're making to the project.
Click to expand...
Conventional wisdom would agree with you but the wisdom guiding the decisions of TPB is not conventional...

Surely if they needed more people to run the uniprocessor client for a while, they would just increase the number of points you get for running it?
Click to expand...
Then they'd be forced to maintain a new paradigm. Something Stanford is wont not to do.
 
Last edited:
Dual PIII @ 1.2 GHz? It has been a long while since I fired that thing up (1U, 1 million dB.) I also have a Celly 666 overclocked to 766 (raises the question of ...why? :)) Is there a FreeBSD client that would run under FreeNAS? I think there is.
 
Stanford really needs to use some basics laws of economics in order to maximize science being done. Best of all, the currency they use is free!

It is all about supply and demand. With all of those smart people at Stanford, surely someone could solve this problem since it is really a simple optimization problem. If they need something getting done, then they need to give people an incentive to do it. Look at how popular -bigadv has become because all of us are trying to solve some maximization problem where we have a binding constraint. We are going to choose to do what maximizes the use of our limited resources.
 
musky said:
Dual PIII @ 1.2 GHz? It has been a long while since I fired that thing up.
Click to expand...
I also have a dual 1.1 GHz PIII @~1.2GHz. It's not the most stable machine out there unfortunately. It EUEs as often as not, otherwise I might have turned it on.
 
amdgamer said:
Pande Group really needs to use some basics laws of economics in order to maximize science being done. Best of all, the currency they use is free!

It is all about supply and demand. With all of those smart people at Pande Group, surely someone could solve this problem since it is really a simple optimization problem. If they need something getting done, then they need to give people an incentive to do it. Look at how popular -bigadv has become because all of us are trying to solve some maximization problem where we have a binding constraint. We are going to choose to do what maximizes the use of our limited resources.
Click to expand...


fixed it for you because Standford actually has nothing to do with F@H.. never have.. they just plastered the name on the project because the projects hosted there and when the project started it looked better to say that was part of a well known university..

the problem is while the current point system works its actually quite flawed..
 
sirmonkey1985 said:
problem is to be fair they cant increase them.. if they did increase them all hell would break loose on the foldingforum.. when they were thinking about modifying the Kfactor(multiplier used for bonus system) on one of the SMP A3 WU's mid way through the WU run everyone went nuts because all those people that ran it before wouldnt of gotten those extra points from the Kfactor being changed so they decided not to do it..
Click to expand...
PG created this problem because they didn't foresee future discrepancies that could arise with all the different clients they have available, or they did but dismissed them to be critical at the time. They are somewhat in a quandary with the point and bonus system as it stands. Too many discrepancies and I'm really beginning to dislike the bonus system itself (won't elaborate here). I am convinced these problems could have largely been avoided if methodically thought out but that's JMO.

.. at some point here in then next 3-4 months thats what ill be doing since i have a bunch of socket 754 and socket A boards laying around in boxes not being used..
Click to expand...
If you were local I'd give you all my Socket-A hardware that I have lying around and still working. I think in my whole time being an AMD customer, I may have purchased close to a dozen AMD processors... Some didn't make it through to our ramp for #1 whilst others died in the line of duty years later to retake the #1 spot. There are probably a few veterans who have some good fight left in them, LOL. :p
 
APOLLO said:
Stanford created this problem because they didn't foresee future discrepancies that could arise with all the different clients they have available, or they did but dismissed them to be critical at the time. They are somewhat in a quandary with the point and bonus system as it stands. Too many discrepancies and I'm really beginning to dislike the bonus system itself (won't elaborate here). I am convinced these problems could have largely been avoided if methodically thought out but that's JMO.
Click to expand...


the bonus points had its use.. it was to try and get people more involved in the CPU client.. the majority of folders had moved on to only running the gpu client because of the high points and lack of points with the uniprocessor client.. but that is where the problem begins.. the gpu client point system was flawed from the start.. the base line points for each WU was determined by... and ATI card.. thus the nvidia cards started showing numbers way higher then they should of been since the ATI client was completely flawed at the time.. but because thats where they started it ment they couldnt change it otherwise everyone would go ape shit about how they are ruining the point system.. blah blah blah.. the problem here isnt Pande Group.. the problem is that the majority of the people give a rats ass about the science and only care about the points.. people have become completely point based on their folding thats the underlying problem with the current point system.. instead of just a fixed number for each WU which is basically what the uni-processor client is.. but they have already gone this far to change it now its completely pointless.. Pande Group pushed themselves into a corner with the points system and they can never get out of it now. though the one thing i would expect to see soon is possibly a bonus system setup for the gpu client.. from the looks of it the client is setup to accept a bonus system unlike the gpu2 client. so all we can do is wait and see what happens from here on out..

If you were local I'd give you all my Socket-A hardware that I have lying around and still working. I think in my whole time being an AMD customer, I may have purchased close to a dozen AMD processors... Some didn't make it through to our ramp for #1 whilst others died in the line of duty years later to retake the #1 spot. There are probably a few veterans who have some good fight left in them, LOL.
Click to expand...

arg damn id definitely take those processors off your hands..
 
Well I just added another single smp client on my WHS box. So I have 2 classic clients running. I will see how this crappy heatsink on the WHS box holds up might have to shut it down.

It is funny I just gave away all my single core stuff. P4 3.0 couple athlon xp 1800+ 2500+ and 2600+ some p2 and p3 systems. Even gave away my old P2 xeon dual slot server with 2x 450 MHz Xeons and 1.5 gb ram lol. Those were the days.
 
I've had my parents computer running classic as a service for about a year now, mostly 24/7, just sucks some times cuz they get old sometimes and shut it off and forget to turn it on for a week or two, and I can't really monitor a missing 150 ppd on top of my scores so I dont catch it

Good to burn some of those older units up!!
 
Does anyone know if the linux classic client is faster than the windows classic client? Also I am assuming you can run multiple classic clients if you have each client in its own folder correct?
 
itsrickjames said:
Also I am assuming you can run multiple classic clients if you have each client in its own folder correct?
Click to expand...
Yes, and that would be the recommended setup if you have more than one core and for some reason cannot run the SMP client.
 
APOLLO said:
Yes, and that would be the recommended setup if you have more than one core and for some reason cannot run the SMP client.
Click to expand...

I can run the SMP client just figured I would help the cause for the classic work some.
 
itsrickjames said:
I can run the SMP client just figured I would help the cause for the classic work some.
Click to expand...
OK, create separate folders and separate shortcuts for each client. Multiple client windows will be opened with different WUs downloaded once executed. Just make certain you have different machine IDs for each instance and you'll be good to go.
 
Thanks for the support guys, to answer a couple questions.. YES the classic client can run on Linux. "Classic Client" is just a term to reference uniprocessor work units. Even the standard SMP client can run classic client work, you just don't use the -smp flag.
 
Tobit said:
Thanks for the support guys, to answer a couple questions.. YES the classic client can run on Linux. "Classic Client" is just a term to reference uniprocessor work units. Even the standard SMP client can run classic client work, you just don't use the -smp flag.
Click to expand...

Thats what I have been doing removing the -smp from the shortcut. I was just wondering if the units were faster in linux or in windows.
 
musky said:
Is there a FreeBSD client that would run under FreeNAS? I think there is.
Click to expand...
No FreeBSD native client but it does work well on *BSD under Linux or WINE emulation.
 
The return of the 48 core AMD boxen. 48 clients running on one computer lol.

Have to use a hypervisor, though.
 
running the 6.02 on my 2.6 celery smoothwall box .. is that the "classic" client?
 
I halted the SMP client and installed 4 classic clients. After having them all run for all night, I'm getting 1119 PDD! I got bold and cranked my clock way up to 4.0ghz, and was folding away stable. But my poor q6600 was getting too hot, even with water.

The PPD is better than I expected, but even so; SMP is about 7x more.

I'm going to leave the classic clients running for a while. After about a week, I'll start alternating the SMP client. It's actually kind of entertaining.

Screenshot for the sake of laughs. It's not running at that clock, that was just a little experiment.
fix.png
 
Last edited:
id honestly only recommend running the uni processor WU's if you have old systems laying around if you have a system that can run smp id run the smp WU's on it..

if you want to run uni processor WU's id say check your local craigslist page and see if you can find some old junk (working) dell or emachines, or athlon XP systems people are selling or giving away.. i found about 5 or 6 machines that are working for free on the craigslist where i am and will probably be picking them up tomorrow to run the uni-processor WU's..
 
sirmonkey1985 said:
id honestly only recommend running the uni processor WU's if you have old systems laying around if you have a system that can run smp id run the smp WU's on it..

if you want to run uni processor WU's id say check your local craigslist page and see if you can find some old junk (working) dell or emachines, or athlon XP systems people are selling or giving away.. i found about 5 or 6 machines that are working for free on the craigslist where i am and will probably be picking them up tomorrow to run the uni-processor WU's..
Click to expand...

I agree with you fully, I'm basically just playing around. I was sitting at my desk waiting on a laptop to reimage.

Which makes me consider something. I should start asking my clients if they would be willing to run Classic; I don't think many of them would go out of their way to leave the system on. But the deadline is so long, 2 hours a day on a decent proc would be enough.

I've got a friend running it on his iMac, and another one running on AMD 64 4000+, they just don't stay on all the time.

I wish Pande would implement a bonus structure for the Classic WU's.
 
You must log in or register to reply here.
Back
Top