TR Forums

Ragnar Dan wrote:

david00214 wrote:

Good to have you back Pikapokey

I'm not sure you have that recruiting members to the team thing down, quite yet. You're not supposed to poke fun at people you just got to help out.

Pikyboy has been on/off member of UGN ... We keep it light with every one.. My favorites are Fuzzedup head and flybywirings ... It's fine some times

Have my 8800GT up and running with the GPU folder for almost a week now, so that should be about 1000 more PPD than I was previously providing with 1 core of my X2 4400+. Folding under SN26P_X2.

My 8800 GT gives me a good 5000 PPD running it 24/7. What's this 1000 PPD stuff about?

First off I would like to say I am sorry for not being able to help quickly with the call to arms. I have tried to contribute but there is really not much I can do since the only computer I have is an old Athalon 64 that is on for maybe 4 hours a day. (Financial considerations) Yea... its pretty sad I have managed to rack up an amazing 3.5k points in three months. But hey I try...

Now the good news is that I will finally be able to help more before Christmas, its taken me two years but I finally have enough money to get a nice new system. I'm planning on getting a q6600 and a 4870. Any idea how many ppd that will get?

Since I am a new and all, what would you recommend for clients to get the maximum ppd. I am also willing to change hardware if there is anything that will get more ppd for around a $1000-$1200.

I love this website and really hope to be able to contribute more soon. Please help all I want to do with my new system is a little bit of 3d computer graphics (Blender) a bit of gaming, and LOTS of folding. So please suggest the best way to max out ppd for my budget.

Thanks for helping me to help this great community!

nvidia video cards are a lot better for PPD, but if you game, ATI are better right now for price vs performance

HurgyMcGurgyGurg wrote:

First off I would like to say I am sorry for not being able to help quickly with the call to arms.

That's fine. It is after all a volunteer project and you don't have to feel bad if you can't contribute. You can always post something funny here to stir up the pot.

HurgyMcGurgyGurg wrote:

Now the good news is that I will finally be able to help more before Christmas, its taken me two years but I finally have enough money to get a nice new system. I'm planning on getting a q6600 and a 4870. Any idea how many ppd that will get?

Since I am a new and all, what would you recommend for clients to get the maximum ppd. I am also willing to change hardware if there is anything that will get more ppd for around a $1000-$1200.

By Christmas you may be looking at the Q9400. We can fine tune that build when the time comes. If you were to get such a system today, then at the very least you should run an SMP client + GPU2 client. The 4870 is not optimized yet (one would hope by Christmas Stanford should have a better GPU2 client for DAAMIT GPUs but you never know when it comes to writing software) so it is doing about <2000 ppd. 1 SMP + 1 GPU2 on DAAMIT should be about <3000 ppd. You can try to run 2xSMP with VMware to increase your points too.

After some part time folding I am now able to do it full time. I got a E6750 and a 4850 giving me about 2000 ppd.

Flying Fox wrote:

HurgyMcGurgyGurg wrote:

First off I would like to say I am sorry for not being able to help quickly with the call to arms.

That's fine. It is after all a volunteer project and you don't have to feel bad if you can't contribute. You can always post something funny here to stir up the pot.

HurgyMcGurgyGurg wrote:

Now the good news is that I will finally be able to help more before Christmas, its taken me two years but I finally have enough money to get a nice new system. I'm planning on getting a q6600 and a 4870. Any idea how many ppd that will get?

Since I am a new and all, what would you recommend for clients to get the maximum ppd. I am also willing to change hardware if there is anything that will get more ppd for around a $1000-$1200.

By Christmas you may be looking at the Q9400. We can fine tune that build when the time comes. If you were to get such a system today, then at the very least you should run an SMP client + GPU2 client. The 4870 is not optimized yet (one would hope by Christmas Stanford should have a better GPU2 client for DAAMIT GPUs but you never know when it comes to writing software) so it is doing about <2000 ppd. 1 SMP + 1 GPU2 on DAAMIT should be about <3000 ppd. You can try to run 2xSMP with VMware to increase your points too.

just chiming in with my experience, 1x q9300 @ 3.2 using one SMP client on three cores and a GPU2 4850 on the last one is about 3600 ppd. If you use a VM instead of the windows SMp, its around 4100. Not a bad guideline for how your performance may be if you OC a bit [and of course the 4870 being faster ]

Ragnar Dan wrote:

My 8800 GT gives me a good 5000 PPD running it 24/7. What's this 1000 PPD stuff about?

I'm turning in about three 480 point work units a day. I just noticed I only using about 50% of one core. Do I need to up CPU usage Percent. Its at 50% in advanced setting right now.

Great work everyone and special cudos to David!

I have been out of the loop for a bit dealing with aging parent issues. Not fun.

However, I will be adding more folding power shortly.

We have just three weeks to fend off Fuzzy-Wuzzy, so let's get to it! You .............. shall not ................... PASS!!

chasscF1 wrote:

I'm turning in about three 480 point work units a day. I just noticed I only using about 50% of one core. Do I need to up CPU usage Percent. Its at 50% in advanced setting right now.

Or GPU, or whatever. I don't recall having set mine, which assumedly means it took 100% of whatever was available.

At the beginning of the thread the screen shot of our production shows 473k ppd. But now we are up to 507k ppd.

Alliance Francophone (#5) is at 565k ppd. Think we can add on another 60k+?
We did hit 563k points for just one day (aug. 12th).

I have been out of the loop for a bit dealing with aging parent issues. Not fun.

Sorry to hear that. I know what it's like.

Only about a week and a half until i add another 3-6k ppd

[how much does the 8800 GT pull in, exactly?]

I've just added my A64 X2 3600+ to the mix, FAHSpy is projecting ~210 PPD. I'm running a GF7600GS, so no GPU-client for me (If I understood correctly).
I do feel the need to say, these clients are an utter disaster in usability.

pikaporeon wrote:

[how much does the 8800 GT pull in, exactly?]

Probably 4400-5000 depending on shader clock speed.
My 8800GT 256MB cards are doing 4700ppd each @ 650MHz core clock (shader clock linked). It seems with a couple different projects I can get 5200 though (rare).

Saribro wrote:

I've just added my A64 X2 3600+ to the mix, FAHSpy is projecting ~210 PPD. I'm running a GF7600GS, so no GPU-client for me (If I understood correctly).
I do feel the need to say, these clients are an utter disaster in usability.

You need to run the SMP client. 210ppd looks to be from a single core client.

Running the GPU2 client on an 8800 GTS 640 (sucking up 100% of Core 1) and the base 6.2 client on Core 0 of an E6600 OC'd to 3.24Ghz (360x9). Temps still in the low 40s on my P182/IP35Pro combo.

I went through the parts bin in the garage and was able to cobble togther three more platforms, although none of them are particularly awe-inspiring. It was kind of fun loading XP on a 10-year-old 4GB SCSI drive, but I forgot how loud that thing was! An old Deskstar spun up and has been running for a few weeks, but now I'm getting errors every 5 minutes so it's probably on it's way down.
Garagebox: Pentium 640 and an 8400GS
Grungebox: Celeron 356 and an 8400GS (4GB SCSI)
Deathbox: Opteron 165 and an 8600GT (30GB Deskstar)

I'll have to reload XP on all these when the 30-day trial expires. I wouldn't mind running Linux but the wireless devices I'm using are not well supported and I'm way too lazy to run a line out to the garage.

I'm impressed with how much extra power has been thrown to TR's folding effort since the CALL rang out. PPD is up over 20%! We're not out of the woods though. With the advent of this new GPU2 client there are teams out there now producing a lot more points than they were just a few months go.

So, if we wanna keep our top 10 status, and regain the two places we have fallen down the ladder, we need even MORE points. If you can bring about some more computing power, or get a friend to joint the TR folding team, well, ya know, that'd just be awesome

The GPU2 clients are actually nice clients to install & use. The Linux clients are pretty easy to setup & use, but no pretty pictures. The Windows SMP client however could do with a bit more user friendliness ... and that is exactly what they are working on now.
An 8800GT should pull in over 4000PPD, even with an old AMD 64 3200+. If it doesn't it usually means that using the wrong Nvidia driver or that another client is taking away all the free cpu. Check http://www.nvidia.com/content/forcewith ... wnload.asp

Pegasus wrote:

pikaporeon wrote:

[how much does the 8800 GT pull in, exactly?]

Probably 4400-5000 depending on shader clock speed.
My 8800GT 256MB cards are doing 4700ppd each @ 650MHz core clock (shader clock linked). It seems with a couple different projects I can get 5200 though (rare).

I de-linked my shader clock from the core clock from the start. I increased its speed 10% and am usually getting reported PPD of 5300, sometimes a bit less, now and then as much as a claimed 5670 or 5760, I forget which. At any rate, I think de-linking is probably the best way to go so the core doesn't limit the shader's speed.

I don't really know for sure, of course, because I don't know diddly about these modern video cards any more.

All I do know is that anyone who wants to help the team a lot and has a PCIe slot to spare should buy one, either new or even used from eBay. I'm considering a couple for my main machine.

Just tossed an Opteron 165 back into the mix this afternoon, Linux client with -smp and -advmethods. It grabbed a 2662 (one of the WUs which uses the new a2 core), and is now merrily crunching along at an estimated rate of ~1160 PPD.

just brew it! wrote:

Just tossed an Opteron 165 back into the mix this afternoon, Linux client with -smp and -advmethods. It grabbed a 2662 (one of the WUs which uses the new a2 core), and is now merrily crunching along at an estimated rate of ~1160 PPD.

You may want to be careful grabbing the -advmethod cores. I don't use them, but I've heard mention that they can cause problems, and the good ole regular SMP WUs are still worth plenty of points.

ok, my first post here at TR...
just wanted to say that i added my 9600GT to help with the cause... can't fold 24/7 but while i'm sleeping it's go time!!!!

p.s. WABBIT SEASON!!!

david00214 wrote:

You may want to be careful grabbing the -advmethod cores. I don't use them, but I've heard mention that they can cause problems, and the good ole regular SMP WUs are still worth plenty of points.

Yup, I'm aware that there are potential issues. The -advmethods SMP core (a2) sometimes fails to shut down properly after completing a WU, causing the client to fail to download the next WU. Since I've only got 3 systems running the Linux SMP core, it's easy enough to keep an eye on them and restart as needed.

For "fire and forget" SMP clients, you probably want to avoid -advmethods.

Unleashing the brother's 8800 GT on a part time basis. This friday it becomes full time, along with his Q6600 - I still need his permission to overclock it; he's sorta leery about that.

wabbit wrote:

ok, my first post here at TR...
just wanted to say that i added my 9600GT to help with the cause... can't fold 24/7 but while i'm sleeping it's go time!!!!

p.s. WABBIT SEASON!!!

Welcome wabbit!! Let us know when Elmer gets close and we will try to be vewy vewy quiet! Hu-hu-hu-hu-hu-hu-hu................

just brew it! wrote:

david00214 wrote:

You may want to be careful grabbing the -advmethod cores. I don't use them, but I've heard mention that they can cause problems, and the good ole regular SMP WUs are still worth plenty of points.

Yup, I'm aware that there are potential issues. The -advmethods SMP core (a2) sometimes fails to shut down properly after completing a WU, causing the client to fail to download the next WU. Since I've only got 3 systems running the Linux SMP core, it's easy enough to keep an eye on them and restart as needed.

For "fire and forget" SMP clients, you probably want to avoid -advmethods.

Since updating I've had zero problems on any of my SMP boxes, all of which use -advmethods. 2 diskless machines, and one my main machine's VM. The only problem I had was that the original update lost 2 WU's folded with the original core and client and Stanford wouldn't accept them. After I deleted them, everything has been trouble-free, and it is definitely helpful in point production.

I'd also point out that Stanford has said on the Folding forum that they're going to move everything over to the new cores at some point, after they eventually release a new A1 core.

But once more I'd like to remind everyone that the most efficient production possible for the time being is available by getting a G92 card. Mine uses less than 50 watts and produces nearly half my points.