If you're thinking of buying hardware to fold with, an nVidia gpu tops the $/watt/PPD list. That assumes you have room, aren't already running an ATi card, are using some version of Windows, and don't want a PS3 to game with. :shrug:

New mobo = :goodjob:

I have an 8800GTX OC that will be in the e6850 build. This rig has a 5870, and I could not fold with it, yet. (It complained when I first tried. Have not tried since.)

News
Stanford is trying to prep all SMP users for the new SMP2, due out whenever, by updating the current client to be compatable with the new cores and WUs. If you're running an SMP client, please check this link. If you're still running a v5.9x SMP, this would be the time to join the 21st Century. The new client is v6.29.
http://foldingforum.org/viewtopic.php?f=24&t=12997

We are at #145. After another 6 weeks of bouncing around while some speed freaks pass us, we will resume our climb up the position ladder. I believe we can break 140 in a couple months. :celebration:

Milestones
Striken7 6,000 01-24-10
jaxawier 250,000 01-23-10
Milanius 25,000 01-22-10
glymfeather 70,000 01-22-10
Paceline 70,000 01-21-10
Mr.O 3,000 01-21-10
DragoonWraith 7,000 01-17-10

We have no new members, but there have been several "old" members getting their fold on again. :cookie:

Fold on! :goodjob:

Here's more news relating to SMP2.
http://foldingforum.org/viewtopic.php?f=24&t=13038
http://foldingforum.org/viewtopic.php?f=24&t=13039

After having my computer die on me a couple of weeks ago, I have gotten a new laptop which has at least twice the processing power than my old one, so I'm folding again... (Hence the inactivity)

Friends, Forumers, countrymen, lend me your ears. SMP2 is out of Beta testing and into -advmethods. It is still a Beta client, but running it no longer requires you to be part of the Beta test team, just add the -advmethods flag, either through configuration or to the target line of your shortcut. It now has it's own section at the Folding Forum.
http://foldingforum.org/viewforum.php?f=58&sid=387587fe6279f6af14211c2682e03b75

SMP2 WUs have a bonus system like the big advanced WUs for multiple quads. http://foldingforum.org/viewtopic.php?f=24&t=13039
I was able to get the bonus with a T6400@2.0GHz on a laptop, so I expect there to be little trouble for most hardware. The faster you can complete the WU, the bigger your bonus.
1- Fahmon will not take the bonus into account and show a very low PPD. http://code.google.com/p/hfm-net/ still has a few bugs, but it will calculate your projected bonus.
2- Like -bigadv, to get the bonus you need to have completed 10 A2 WUs on a client with a Passkey. Straight Windows folders will take a hit with no bonus until they finish 10 A3's. Linux users, or those who ran Linux WUs in a VM, with a passkey will get the bonus on the first A3 WU.
3- The demand is greater than the supply. Since hitting -advmethods, the number of people trying this keeps the servers pretty empty. 4 out of my 5 SMP clients are running A1 WUs at less than 1/2 the points of an A2 or A3. Linux users are better off, just switching between A2 and A3. There is one guy working on a modification of notfred's to allow you to easily configure for -advmethods in a VM. This would give adventurous Windows users a way to maximize their points with a VM. I'll look into that some more and report back.

This week in folding.
We are still at #145 and are due to bounce around for another month, but should continue moving up after that.

Milestones
Paceline 80,000 01-30-10
Floating_Jet 1,250 01-29-10
jaxawier 300,000 01-29-10
Striken7 6,000 01-24-10

We have no new members, but Floating Jet gets a "welcome back cookie". :cookie:

Fold on! :goodjob:

thanks for heads up.
added in -advmethods on a vm with suse,got an a3 at once,have to see how it works out.

Ok guys I'm willing to help
I got a core I7 920 and 4850. I need advice please
I have 1 GPU client client running with big wu's turned on I believe and slightly higher priority than other processes. I also have a DOS client which is driving me mad. I want to remove the DOS client but don't know how. I also would like to get and SMP and a few more GPU clients to help the cause

Nom nom nom nom...

Glad to be back Lets tear up these WU's! :toughninja:

I have a stable Radeon 5870 and Windows7 waiting, but it seems that this project does not support this card for some weird reason. Ok, I'll run Boinc on it instead at least until they decide to allow running this thing on the 5000 series.. For now my PS3 is the only thing silently doing work on this project in the corner.

tried -forcegpu ati_r700 ?
http://foldingforum.org/viewtopic.php?f=51&t=12935

edit: good it worked out chaul

Ohh, thanks. I'll try that tomorrow. I'll let Boinc exhaust the GPU work queue over night and then start folding with that option while waiting for ati_r800 support.

Edit: First work units have been completed. Each unit seems to take about 2-3 hours to complete. The client looks very stable too - the GPU core is under 97-98% load and temperatures are very reasonable at 57C. I've seen higher, maybe it means that they could optimize the client a little more.
[05:37:40] Project: 5743 (Run 2, Clone 59, Gen 331)
[05:37:47] Starting GUI Server
[05:40:14] Completed 1%
...
[07:50:32] Completed 100%
[07:50:33] Successful run

We are still sitting at #145, and will still bounce for another 4 weeks. The current trend is up 3 and down 3. The supply of A3 SMP2 WUs has improved and I haven't seen an A1 for several days. You will notice quite a PPD range on the various WUs.

Milestones
subparwelder 4,000 02-07-10
Dogsbody 1,750,000 02-06-10
Floating_Jet 2,500 02-05-10
jaxawier 350,000 02-03-10
Striken7 7,000 02-02-10
Chaul 450,000 02-02-10

2 new members :cookie:
Ragnalin_The_Thief 02-02-10
subparwelder 01-31-10


@Ragnalin_The_Thief, you made the team list with 0 points and 0 WUs. I'm guessing you had a WU crash early enough to not give you partial credit. Review the GPU install giude to see if you got all the steps.
http://www.stanford.edu/group/pandegroup/cgi-bin/edit/English/WinGPUGuide
I don't run ATi, but some of the guys can give you more info on the Environmental Variables that help the ATi GPU client run better. One thing, don't use the viewer. It's buggy and will adversely affect your folding experience.

By DOS client, do you mean the console client that runs from a command line window? If so, there is no uninstall. You just delete the folder you made for it. If you configured it to run as a service, re-run configuration to remove the service first. I run all console clients. The Systray client is neater, but, personally, I find the console more flexible.

Basic requirements for SMP2
1-a password protected Admin account
2- a passkey http://fah-web.stanford.edu/cgi-bin/getpasskey.py

Depending on what else you use your computer for, you might want to install a couple CPU clients to see haw they affect your system while doing those other things.

Fold on! :goodjob:

wonder if wine + ati works now ? saw depressing reports about it,whish me luck :read: could be fun to try though

enviromental variables (either in your user settings or system)
FLUSH_INTERVAL at 256 (128-256 with incremental of 32. 192 worked for me)

CAL_NO_FLUSH = 1

BROOK_YIELD = 2

CAL_PRE_FLUSH=1

I have just had the same WU twice. It's p2613_TETHERED VESICLES. Is is a common problem or not? I'm happy to keep folding, but I was just wondering

It's not a problem. A work unit is unique in its composition of project, run, clone and generation numbers. p2613 refers to the project number and as such completing work units of the same project is quite common. If you scan the logfiles or use a third party viewer you will be able to see the run, clone and generation numbers e.g Project 2613 (Run 23, Clone 2, Gen 100). I think it's also viewable in unitinfo.txt under the tag name.

If you start folding work units with the same combination of project, run, clone and gen numbers then that's usually indicative of something going wrong.

Ok, thanks. Will fold whatever they hand me =)

So I thought I'd add the GPU client seeing I found a way around my card not being supported, but have since noticed it really killed the SMP clients output.

The unit it had been working on went from <2 hours last night to >10 hours currently once the GPU client kicked in. (So it added 18+ hours.

So, I guess comparing PPDs the GPU is the way to go since the points awarded seem to be about the same. (400-500 so far say Fahmon.) The SMP client offered up <1100 PPD originally, 94.5 now, the 5800 was 5000 PPD, and is 4500 now. (New WU though.)

But then I read about the bonus points and that the new SMP core does things differently.

Basically is running both going to be worth it in the end? Does the points gained from one, condone the slow down in work, or does it all balance out in the wash so to speak?

I have killed the GPU client to finish off this SMP unit but would appreciate any advice on what would be best.

If you have quad core cpu etc, you can limit the smp to not use one of the cores. That last core is then used with the gpu, as they still have not implemented a way to offload all the calculation to the card with ati.. It always seems to take quite a lot of the cpu (for me 10%) for the gpu client. For me that also means that it reserves the whole core for the gpu on my core i7 860, one eigth of the cpu goes out the drain, while the SMP client is running with -smp 7 command line parameter. SMP seems to require synchronization with all the threads, but if one of the cores is congested, it'll slow down the whole process to a crawl. Same for GPU client, I think, vice versa. You should also set GPU client to run at higher priority from the settings, and let SMP run at the default lower setting.

Edit: FahSpy claims that I should be getting around 12000PPD, but I'm not. The GPU is running at 4000-5000PPD, PS3 gets about 1000PPD. That's without any bonuses. Now I'm running the SMP, but my total still only shows 6000-7000PPD at these stats sites, the last I checked, hmm. It doesn't add up.

I set the GPU client to use core 3, and the SMP is using the other 3.

I'm just waiting for Fahmon to update to see if the SMP drops from the current 26mnETA/1020PPD.

I have not changed priority.

Damn, the GPU client has already gone up 13% while the SMP is still at the 95% it was when I restarted the GPU. (Damn that is fast.)

The bottleneck seems to be with ATi GPU client and a dual core running SMP. Setting the gpu Environmental Variables helps a lot, but it can still take too many cycles away from the cpu. Quads still get into the issues Chaul mentioned. Dual core and ATi may just require GPU and a CPU client, for the time being. Some of the "better" cpu WUs will give you 400 PPD.

**I see you posted while I was writing. :shrug:

There is a very minor hit with nVidia. This laptop- C2D T6400@2.0, GT 120- will run the GPU and still finish SMP2 in time to get a bonus. They hope that GPU3 will remove the limits on the ATi cards, but they ran into more problems in the past months. The last news I saw said that ATi would still not be fully implemented if the client is released soon.

*******************

This is getting boring, but we're at #146. Projections show us making a nice move in the next month. We should break 140.

Milestones
Paceline 90,000 02-14-10
Chaul 500,000 02-13-10
jaxawier 500,000 02-13-10
gigabyte 250 02-12-10
Floating_Jet 4,000 02-12-10
subparwelder 7,000 02-12-10
(TES)_wolfpup 450,000 02-09-10


We have a new member!
gigabyte 02-11-10 :cookie:

Fold on! :goodjob:


edit- wolfpup, if you're running SMP2 A3 WUs, Fahmon will not show you the bonus points.

I'm running a q9550 @ 3.5 and a 5870 @ 900/1300. I had to use the following flags, -gpu 0 -forcegpu ati_r700 to get the 5870 to work though. It tricks it into thinking it's a 48XX series card.

Well, the SMP has not updated since changing affinity, so it still says 1020 PPD. Though I am sure it will change once Fahmon updates.

If, and that's a big if, it maintains these rates, I'd get just shy of 6000 PPD.

P.S.

I am using SMP2 A3 WUs.

The GPU was at 18% when I posted above. It is now at 50%.

P.P.S.

Get http://code.google.com/p/hfm-net/ instead of the http://fahmon.net/. I now see 2494 PPD on the SMP client. It also lists the correct core instead of "Unknown" like Fahmon.

But sadly it's still running slow even after the affinity changes. (Still 95%)

P.P.P.S.

Well, now it's saying the SMP is "Runningnoframetimes" and is indicated by a yellow bar. It was listed as "asynchronous" earlier.

Anyone know how to go back to both GPU and SMP being green? (Or synchronously?) I tired setting the date/time in windows, and using a different update location as suggested in a thread I found but all the did was make it go from blue (Asynchronous) to yellow (Runningnoframetimes).

It is progressing though, I just don't understand what it is trying to tell me, or how to fix it.

Ok, enough P.P.S's........


Running just the SMP, it turns back to green, but when I started the GPU client, it now says "Runningnoframetimes"

How do I get them to play nice together, what setting am I missing? (I know others get them to work, I just don't know how. )

P.S.

Bah! They are both green now. I'm going to assume it's not a major deal, (Bug?) but I'd still like any light shed in this now dark corner of my folding experience if anyone knows. (I'll obsess about it if it continues to happen.)

P.P.S.

Well, back to 1 yellow (CPU) 1 green. (GPU), but it's working on a new WU, and my PPD is now up to 10,000+.

Seems the GPU client uses 30-45% of a 4 core CPU. I was under the impression a "GPU" client, would not rely so much on CPU. Silly me.

And what does TPF stand for?

TPF is Time Per Frame. I have one machine that regularly shows one of the GPU and the SMP with an orange block. I could never find what it means. It comes and goes- usually I have it. It doesn't affect the program, and the times/points it gives are as acurate as a nonhighlighted entry on that or the other machines.

I've never seen "Runningnoframetimes".

ATi GPU can use up to a full core. People have gotten that down to 5% plating with EVs. nVidia uses 2-5% without playing games. All my SMPs are using all 4 cores and just being bumped by the GPU clients once in a while.

Mine seems to like 1.5+ cores used. I locked it to core 3, and the other 3 cores were all still showing 37-50% usage. It fluctuates between 0-2% before starting the GPU client.

I think I'll just ignore it for a day or 2 and see what comes of it. If the projected deadlines (Meaning I meet them.)/PPD (4500+ and 5300+)/credit (2431, and 358) are close to correct, I'm golden.

I do shut it all down while gaming, but I still think I'll make the early deadlines. (As long as I remember to restart them.)

Thanks for the help folks.

P.S.

Oh right, what EVs would you suggest? (Or a link to a guide or some such would be cool.)