Look out everyone, p2501 is coming.

[Polizei]

Or at least rolling over me. Seemed like my points were fixing a flat tire as p2501 passed me in the fast lane.

:wtf: man... no hard feelings. ;)

...yeah... no real point to this thread. This forum died for a couple hours, so I figured I would spark some life into it.

In other news... um... hooray nVidia client?

[The_Barnacle]

All 9 of my clients are running p2653 or p2665. What are the stats on this new 2501?

[Polizei]

p2501 is a user here... he passed me in points as if I weren't folding any WUs.

[The_Barnacle]

:rofl:Hahaha, and I thought you were talking about a new project....

:ROTF::poke:

[p2501]

p2501 is a user here... he passed me in points as if I weren't folding any WUs.

Aww, come on, I didn't do anything! :D
No for real, I have just this one quad folding here 24/7, it's not like I have a farm (yet). But as you said, the new nv folding client seems interesting, question is when it'll be released. And what about those slide pictures, will g200 really be such a boost over 3870... I wonder. What ppd does one get from a gpu folding client anyway? Enough to kill the plan for a dedicated quad folder? :shrug:

:rofl:Hahaha, and I thought you were talking about a new project....

:ROTF::poke:

:doh::rofl:

[MikeB12]

Enough to kill the plan for a dedicated quad folder?

lately I've been wondering about the future of smp at fah too... the way they discontinued folding for the 19xx ati cards recently. little hints like in Pandes fah post post below kind of have me worried about how long I'll be dedicating my quad farm to fah... what is going to happen to smp points? and where is fah going with cpu crunching?..

http://foldingforum.org/viewtopic.php?f=16&t=2655
We don't normally post regarding projects that are still in beta testing, but we wanted to make an exception in this case:

The SMP client has been yielding a number of important results for us, but its speed has been limited by less-than-ideal scaling (and less-than-optimal CPU utilization) on 4-core and higher systems. We have spent a lot of time working on this and have a new version of the A2 core that does *much* better in this regard. The first versions of this new core are almost ready for release.

Running project 2662 on this new core, we see CPU utilization of >95% on both 4-core and 8-core machines.
In conjunction with the release of project 2662, we have decided to increase the points value of its sister project (using the A1 core), project 2665. New assignments of both projects will now be worth 1920 points. This make the 2665 points yield more similar to most A1 projects; it will also substantially increase the 2662 points yield.

Because the SMP portion of our project does not operate in isolation and because we value donors to all portions of the project, in the long term we will have to re-examine the normalization of SMP points yield relative to standard and GPU clients. We will also be introducing some additional bonus schemes. But these are things to come.
I should add that with the new high performance cores starting to mature, I am going to look to rebalance points project wide, including *all* clients. There are some disparities that need to be resolved in some uniform way, with the goal of pushing what's best for the science that's running right now.
Thanks for folding!

I wonder how this is going to affect smp's future. I'm not gonna be a happy camper if this ends up handicapping quad farms many folders have put thousands of dollars into... I may be reading in between the lines here, but to me it sounds like smp has a bleak future. anyway, hope I'm wrong... if not, I'll be a wcg user in the near future..

[p2501]

I wonder how this is going to affect smp's future. I'm not gonna be a happy camper if this ends up handicapping quad farms many folders have put thousands of dollars into... I may be reading in between the lines here, but to me it sounds like smp has a bleak future. anyway, hope I'm wrong... if not, I'll be a wcg user in the near future..

I don't think that smp will have a bleak future like you put it, but if they are capping points, couldn't we just go back to having one single instance per core (which is what he wanted to say I think, to level that difference)? Btw, why the hell don't they just award points determined by how fast a user returns a work unit? How are points determined over at wcg for instance? Should they really mess up the points system I think they will have an angry mob of all the committed folders at their doorstep, including me. I don't want to give this up just because they are to stupid to listen to their "fan base". :mad:

The question about what is achievable with gpu folding ppd wise still stands. :confused:

[MikeB12]

a2 core will utilize up to 8 cores, so there won't be a need to go back to running an instance for each core..

my comments above are just me being a pessimist...

but if they do "equalize" points and we already know from pande's comments that gpu or ps3 folding is more "geared toward the science" so to speak.. in my mind that either means increasing the ppd for the single core cpu client or decreasing points for the smp. so it looks like we're moving into an era where gpu type folding is more important than cpu folding. my pessimistic attitude comes from the fact that great points were awarded to smp when it came out and in the past year; and we're now in a situation where many users spent big money on high production cpu folders. now if fah comes along a year later and handicaps smp points, they're basicly saying to me "well guys, cpu folding is no longer as useful as it once was, now we want to take that point system away and gear it to the gpu" then that leaves a lot of farm quad users sitting on the fence thinking why the hell did I spend $4k on quads for cpu crunching over the last year... since this is not important as it once was, why not dedicate my crunchers to other projects like wcg or whatever project does optimize on cpu crunching.

and again, that's just me reading between the lines. not facts... if gpu type crunching is better for fah, then maybe it's a smart move on their part if that happens... it just means I'll put my cpu's on other projects and leave my gpu's on fah...

that's where I'm coming from.. and it's just .02, not facts. I'm reading a lot in between the lines here.
but from p2501's ques above about adding another quad folder, I think I'd hold off on that idea and wait til pande announces what this whole new points equalization plan will be. don't just assume they'll cater to the needs of the user population.. from their history of caring about what the user thinks, I don't think that plays into the picture too much.

I don't think any of use know what the ppd will be on the new nvidia client until it gets released and real wu's get turned...
but I would like to see how a 4870 w/480 stream processors does on the gpu2 client.
the thread I've been watching is http://foldingforum.org/viewtopic.php?f=42&t=2747
also some discussion here http://foldingforum.org/viewtopic.php?f=42&t=2588
for points/$$ though the 3870 is a good value. HIS at NE for $135 http://www.newegg.com/Product/Product.aspx?Item=N82E16814161218

[SocketMan]

This is my "best case" (ppd and efficiency)
2 cards: 3850 256mb each, on a dual core cpu, and a mobo with 2pciE slots, running 2 clients win X64.
Also I don't think it's time to start getting rid of Quad cores, but something to keep in mind when making a buying decision:

Hmmm.. should I get another quad (number 5:p:)
or a video card......;)

Fold on :hammer:

[p2501]

but from p2501's ques above about adding another quad folder, I think I'd hold off on that idea and wait til pande announces what this whole new points equalization plan will be. don't just assume they'll cater to the needs of the user population.. from their history of caring about what the user thinks, I don't think that plays into the picture too much.


Question here is how long will they take to decide, months? Folding gets done either way, so why don't they just take the users perspective into account? I don't get it. I mean the work done with smp and standard client should be about the same, with smp being a project to try to parallelize the cores? What they do to level the points taking gpu folding into account is a prolly different thing... :shrug:

This is my "best case" (ppd and efficiency)
2 cards: 3850 256mb each, on a dual core cpu, and a mobo with 2pciE slots, running 2 clients win X64.
Also I don't think it's time to start getting rid of Quad cores, but something to keep in mind when making a buying decision:

Hmmm.. should I get another quad (number 5:p:)
or a video card......;)

Fold on :hammer:

Your points do look yummy, that's roughly over 3500ppd on wus that seemingly get done quite a bit faster than on a cpu. 2 hours for a wu, getting up to 12 of 'em on a good day... :eek: Or is that FahMon estimate taking multiple wus into account for that ppd on your screenie?

[MikeB12]

Question here is how long will they take to decide, months? Folding gets done either way, so why don't they just take the users perspective into account? I don't get it. I mean the work done with smp and standard client should be about the same, with smp being a project to try to parallelize the cores? What they do to level the points taking gpu folding into account is a prolly different thing... :shrug:

I'm just theorizing, but equalizing the points either means 1-bringing up the single core client to a level that proportionate to what the smp turns on one core... or 2-bringing down the smp points to take it down to a level comparable to what the single core client turns at the single core ppd level. let's hope it's equalizing on the single core client by bringing it up. because if its the other way around and they bring the smp points in line with what the single core client runs, then that'll bring down the points way down on smp.
and from the 3850 example above, if that happens we would then have gpu's capable of turning more ppd than an oc'ed quad.

that's what I'm talking about. so let's hope the points equalization plan brings points up for the single core cpu clients, instead of bringing points down for the smp... otherwise quadcore folding is going to be a loss dollarwise for most of us. because a dual core machine with 2 gpu's now is the biggest producer for the dollar.. which means the investment in quad cpu tech this past year was just a misguided folding effort.

but all we can do is wait and see.. and hope that pande group has a good plan that's fair to everybody and keeps the majority interested in the projects..

[Kingcarcas]

I'm down with that, i'm tired of these 47, 89 or whatever points single core WUs i'm getting! :( My 3870 should be back online once i overclock it and that's 2000PPD :D

[sc00p]

Mike... I think maybe "they" are concerned about cpu-clients are dropping folding because of the much higher smp scores. Are they listening too eagerly single-core cpu folders? IIRC since the beginning of the smp client it was said "overly" credited smp-wus were just temporary... to draw people to make beta testing phase go by faster?
Have they found out the wisdom how to use newest gpus more efficiently with more complex atoms/calculations/wus?

[MikeB12]

I dunno.. any assumptions we make now are just speculation. we'll all find out what the deal is when the specifics of whatever equalization plan pande is referring to gets revealed... I just hope it doesn't mean severely slashing smp points, since I've built 3 quads in the last 6 months. we just took about a 1000ppd slash moving to 30xx and 2665 from 2653. if it gets slashed another 1000ppd or so with this equalization deal...that means $150 3870 would be turning the same points as an oc'ed quad... which would make me want to move to a different project that utilizes cpu power more. if it happpens, it happens... all speculation... I'm sure they'll do whatever is best for the project, and if that means cpu power is better served on other projects than fah for my farm, then so be it...

[p2501]

Dude, setting up 3 folders in the past 6 months, I somehow understand that you're really pessimistic about changes on the points system taking that last points slash into account. So I guess I'll just hold out on that folder.. bleh, I don't like that. :-/

[SparkyJJO]

I figure the SMP would drop in points at some point, I've been expecting it to at some point since the beginning. Question is, how much, I don't know :shrug:

[Kingcarcas]

I see you in the rear view mirror now :shocked: I think it's cool if they drop the points, they just need to make it fair for everyone and give you the right amount for the CPU power you've dedicated to the project. I freaked out when i saw WCG giving out hundreds of thousands of points daily to those people. I don't want it to be ridiculously high because then it seems it's just about e-peen. :shrug:

[SocketMan]

I dunno.. any assumptions we make now are just speculation. we'll all find out what the deal is when the specifics of whatever equalization plan pande is referring to gets revealed... I just hope it doesn't mean severely slashing smp points, since I've built 3 quads in the last 6 months. we just took about a 1000ppd slash moving to 30xx and 2665 from 2653. if it gets slashed another 1000ppd or so with this equalization deal...that means $150 3870 would be turning the same points as an oc'ed quad... which would make me want to move to a different project that utilizes cpu power more. if it happpens, it happens... all speculation... I'm sure they'll do whatever is best for the project, and if that means cpu power is better served on other projects than fah for my farm, then so be it...

Don't forget that these 3870s are $150 now and when they came out the price was about $230. Same thing with the Q6600: its $220 now much lower than it used to be. So if one decided to put a folding system together now
it would cost about the same wouldn't it?
I do think that "classic" ppd are way to low: I mean if one has a quad with
4 "classics" running that would give 800 ppd max ? if even?
Now running SMP on the same Quad,using the same amount of resources,
costing the same, the only difference being the client type and the ppd
amount which would be 4x the classics.;)
************************************************** *******
GPU's are still "beta" and so are the points, the main problem
I am having is the huge slowdown that takes place when running more than
one card per PC. Out of 3 motherboards with 2 or more PCIe slots only
one can (the "best case" scenario in the above post) fully "squeeze" 2 cards.
I've tried running 4 cards on a Phenom QC, the mobo has 4 PCIe slots
and ppd was about 2200:( Running just 1 card with 1 client on the same QC would give about 1700 ppd, right now I've got 2 cards in it and Fahmon is showing
2631 ppd or ~1315 each card)
Clock speed might be the factor (2500) and/or the software,all I can do is
hope for the best and prepare for the worst:)

[heater918]

I gave up on the points issued thing and crunch for the cause :)
(your cause>insert here<)

[sc00p]

When is it politically correct to say what I'm starting to feel... people don't care about the project - mainly points :(

[MikeB12]

It's always correct to say that scoop....

Crunching for the cause is the main reason everyone does it in the first place, and fah is not the only crunching project that has a worthy cause. SO crunching for the cause does not just mean fah only. As also stated, fah could do much better job managing the recognition system which they've neglected for so long now. maybe this equalization plan will remedy that. but the fact that they've neglected it for so long makes me wonder how effective and logical the decision making process is. There are other methods that have me doubting fah's management and it's been building for a while.. another example is SMP came out over a year ago and is still in beta. so we have software in main stream distribution as beta for over a year. there are other issues like when servers go down, there seems to be no overlapping responsibility of getting it back up, report it and you're just told that's so and so's responsibility... I'm not going to jump ship just yet, and if I do move it'll just be to another project that's dedicated to disease cure. I like the fah project, I'm just getting to a point where I don't trust the management philosophy or methods behind it that control the user portion of the distributed computing project. it's not really about the points, it's about quite a few things and maybe it's getting to a point with me where it's time for me to move to a different project. The little announcement by pande with no real details was what spawned these recent posts by me, but the underlying tone was there all along, I probably should have kept my mouth shut...

just disregard my malcontent insubordinate ramblings...
http://img58.imageshack.us/img58/6089/48325747rl3.jpg

[MikeB12]

and wow socket, you're only turning 2600 on 2 cards with the phenom at 2.5ghz, but the opty turned 3600 with 2 cards at 3.2ghz... a 1000ppd increase with 700mhz cpu core speed... guess that goes to show how dependent the gpu2 client is on cpu. I bet if you had those in a e8400 at 4ghz it would turn good results... when I tested gpu2 on my 2900pro it turned 1200ppd (slightly higher core priority) on a Q6600@3.4ghz while running smp... but my gpu load temps never reached anything even resembling full load, in fact it looked like gpu idle temps.. it was like nothing was even crunching on the gpu... it seemed it was all taking place on the cpu, because the smp took most of the hit.

my gpu1 client was much different on my old ati1950 card in an old sempy box, it at least hit some load temps so it appeared it was crunching... that's another thing that kinda bugs, gpu1 is now gone, so I gave my 1950 and sempy away to a friend recently. I purchased the 1950 in September new for specifically folding in the old single core machine. so much for that $120, a waste... if I would have known it was gonna be gone, woulda never bought it..

[Stijndp]

You're not alone in doubting the workings at Pande group Mike.

[p2501]

When is it politically correct to say what I'm starting to feel... people don't care about the project - mainly points :(

I wouldn't put it like you did, people still care about the cause. I don't know about you but I have several cases of cancer in my family so I DO care a lot, but the thing is, IF they are giving out points they should at least take the needs of people into account who put thousands of $/€ into machines to crunch, shouldn't they? I mean they're just points, folding gets done anyway!

[SocketMan]

and wow socket, you're only turning 2600 on 2 cards with the phenom at 2.5ghz, but the opty turned 3600 with 2 cards at 3.2ghz... a 1000ppd increase with 700mhz cpu core speed... guess that goes to show how dependent the gpu2 client is on cpu. I bet if you had those in a e8400 at 4ghz it would turn good results... when I tested gpu2 on my 2900pro it turned 1200ppd (slightly higher core priority) on a Q6600@3.4ghz while running smp... but my gpu load temps never reached anything even resembling full load, in fact it looked like gpu idle temps.. it was like nothing was even crunching on the gpu... it seemed it was all taking place on the cpu, because the smp took most of the hit.

my gpu1 client was much different on my old ati1950 card in an old sempy box, it at least hit some load temps so it appeared it was crunching... that's another thing that kinda bugs, gpu1 is now gone, so I gave my 1950 and sempy away to a friend recently. I purchased the 1950 in September new for specifically folding in the old single core machine. so much for that $120, a waste... if I would have known it was gonna be gone, woulda never bought it..
I think there was an issue with the early GPU2 client versions and the 29xx
cards.The cards kept running in 2D mode and not switching to the full power
3D mode.Also I used to run 4 clients on the Phenom with 2 Vga cards (2 clients per card) cause the early projects only used about 60 % of the card. That might
have been "helping" the 29xx cards to stay in 2D.
The client now has a viewer with 6 view modes,and (my favorite) iter/sec.
I guess if I was getting a quad (or E8400:explode2:.......snap out ....;
mobos with multi PCIe X16 slots would be on my short list.:)

[SparkyJJO]

When is it politically correct to say what I'm starting to feel... people don't care about the project - mainly points :(

Unfortunately that is what many people are in it for (not saying people here, just in general) but that is how they get more people involved. I'll admit that if it wasn't for the points system I wouldn't have as many systems running as I do now....