hey everyone, for those having issues with 6.22, roll back to 5.91/5.92

they have extended the deadline until they get 6.22 working

http://folding.stanford.edu/English/DownloadWinOther

Quote :

"Deprecated Windows SMP beta clients 'We have temporarily extended the deadline of the 5.91 and 5.92 beta clients for Windows SMP. These new binaries will expire on November 2, 2008. These are still beta software; they are also deprecated, so the PandeGroup does not actively support these clients. Furthermore, since new scientific cores and projects require features of the v6.x clients, we cannot guarantee the availability of work for the 5.9x Windows SMP clients at any time. That said, many donors are familiar with these clients and in recognition of that we have extended the availability through the early public beta testing period for the 6.x Windows SMP clients.' The following 5.9x clients are provided as drop-in executables for existing installations only. "

please also note as of this moment, they have the links backwards (5.92 is 5.91 download, vice-versa)

[Edited on August 28, 2008 at 1:27 PM. Reason : .]

8/28/2008 1:25:03 PM

got a PS3 today. First thing I did? start F@H. w00t

8/30/2008 9:05:48 PM

What the shit is going on?!?!

http://folding.extremeoverclocking.com/user_summary.php?s=&u=358071

[Edited on August 30, 2008 at 11:25 PM. Reason : lkhv]

8/30/2008 11:24:46 PM

^^^ ahh cool.

ive been getting ALL KINDS of issues with this crappy 6.22 client. EUEs, 07x, you name it.

8/30/2008 11:25:21 PM

ditto, i'm only running gpu right now because i can't seem to get 6.22 or even 5.91 working now.

8/30/2008 11:38:41 PM

I left the CPU clients running while I was away on vacation even though it wasn't counting them, but from the jump in my scores it looks like it finally started counting them a couple days ago!

How does Alexzogh score 40,000 per day?

8/31/2008 12:31:05 AM

from page 71
/message_topic.aspx?topic=118820&page=71#11879396

8/31/2008 2:04:22 AM

We're back up to 84th place.

8/31/2008 8:46:53 AM

^^ How far do you have to OC a 9800 to get 6000+ ppd? Mine's overclocked about 25% and it pulls a little over 5000 ppd. I have 3 clients running on my Q6600 (3GHz) which pull ~1500 ppd combined.

Am I doing something wrong?

8/31/2008 4:26:18 PM

What three clients are running?

each SMP client (should have 2 on an smp box) should produce ~1250 to ~1700PPD depending on the workunit.

8/31/2008 4:38:23 PM

I have 3 single-cpu clients because the SMP client just crashes (I've tried several versions of the SMP client).

8/31/2008 4:48:21 PM

Hey Limpy. You can have an overclock that will run everything except smp F@H. Its "THE" best way to test an overclock in my opinion. I have had an e6600 @ 3.9 that would run all games and 3d mark programs but would crash during F@H. So I backed it down to 3.6 and was good to go. That processor you have should be able to go alot higher then 3.0Ghz so I suspect heat, ram or mobo may be limiting you.

The best ppd you can do is run the 1 smp, 1 single cpu client and 1 gpu2 client. That should get you around 6000-8000. Running two smps will get you at most 4000ppd.

On a side note, in the next few weeks we should set a new record high for the team. #79

9/2/2008 8:22:21 AM

^^ so even when you run just one SMP client it constantly crashes? Do you run antivirus? If so which one?

and it looks like something is up with the individual stats. The team is getting it's updates, but they're not being attributed to the users yet...hopefully it'll catch up today

9/2/2008 8:46:35 AM

yea, i agree.

also, it's the start of a new school year, should someone post some new posters around campus?

9/2/2008 11:30:43 AM

^^ and ^^^

It's definitely not the overclock, because it crashes when I'm not overclocking either (and at full load the temps are all less than 50C, which is far below the 71C limit for the SLACR Q6600). It happens when I only run 1 SMP client. I get the error "Folding@Home has run into a serious error running the core. and will shutdown."

9/2/2008 3:24:20 PM

Do you run antivirus? If so which one?

Are you running Vista?

9/2/2008 4:57:31 PM

No antivirus software. Windows XP 64-bit edition.

When I run the install script it runs normally and successfully runs "foo.exe" in parallel "...you should see this message twice" appears twice.

Here's my output after a fresh install:

Launch directory: D:\Program Files (x86)\Folding@home\MPI
Executable: D:\Program Files (x86)\Folding@home\MPI\Folding@home-Win32-x86.exe
Arguments: -smp 4

[22:19:00] Configuring Folding@Home...

------ SNIP ------

[22:20:14] Work directory not found. Creating...
[22:20:14] Could not open work queue, generating new queue...
[22:20:14] - Preparing to get new work unit...
[22:20:14] + Attempting to get work packet
[22:20:14] - Connecting to assignment server
[22:20:14] - Successful: assigned to (171.64.65.64).
[22:20:14] + News From Folding@Home: Welcome to Folding@Home
[22:20:14] Loaded queue successfully.
[22:20:16] + Could not connect to Work Server
[22:20:16] - Attempt #1 to get work failed, and no other work to do.
Waiting before retry.
[22:20:24] + Attempting to get work packet
[22:20:24] - Connecting to assignment server
[22:20:25] - Successful: assigned to (171.64.65.64).
[22:20:25] + News From Folding@Home: Welcome to Folding@Home
[22:20:25] Loaded queue successfully.
[22:21:03] + Closed connections
[22:21:03]
[22:21:03] + Processing work unit
[22:21:03] Work type a1 not eligible for variable processors
[22:21:03] Core required: FahCore_a1.exe
[22:21:03] Core not found.
[22:21:03] - Core is not present or corrupted.
[22:21:03] - Attempting to download new core...
[22:21:03] + Downloading new core: FahCore_a1.exe

------------ SNIP --------------

[22:21:06] Verifying core Core_a1.fah...
[22:21:06] Signature is VALID
[22:21:06]
[22:21:06] Trying to unzip core FahCore_a1.exe
[22:21:06] Decompressed FahCore_a1.exe (2035712 bytes) successfully
[22:21:11] + Core successfully engaged
[22:21:16]
[22:21:16] + Processing work unit
[22:21:16] Work type a1 not eligible for variable processors
[22:21:16] Core required: FahCore_a1.exe
[22:21:16] Core found.
[22:21:16] Using generic mpiexec calls
[22:21:16] Working on queue slot 01 [September 2 22:21:16 UTC]
[22:21:16] + Working ...
[22:21:16]
[22:21:16] *------------------------------*
[22:21:16] Folding@Home Gromacs SMP Core
[22:21:16] Version 1.74 (March 10, 2007)
[22:21:16]
[22:21:16] Preparing to commence simulation
[22:21:16] - Looking at optimizations...
[22:21:16] - Created dyn
[22:21:16] - Files status OK
[22:21:27] - Expanded 4735055 -> 24426905 (decompressed 515.8 percent)
[22:21:28] - Starting from initial work packet
[22:21:28]
[22:21:28] Project: 2665 (Run 0, Clone 479, Gen 20)
[22:21:28]
[22:21:29] Assembly optimizations on if available.
[22:21:29] Entering M.D.
[22:21:35] Rejecting checkpoint
[22:21:37]
[22:21:37] rotein: HGG in water
[22:21:37] Writing local files
[22:21:43] Extra SSE boost OK.
[22:21:43] Writing local files
[22:21:44] send back what have done.
[22:21:44] logfile size: 9422Gromacs cannot continue further.
[22:21:44] Going to send back what h- Failed to delete work/wudata_01.xtc
[22:21:44] Warning: check for stray files
[22:21:44] ...
[22:21:44] ... Done.
[22:21:44] - Failed to delete work/wudata_01.sas
[22:21:44] Warning: check for stray files
[22:21:44] iled to delete work/wudata_01.goe
[22:21:44] Warning: check for stray files
[22:21:44] ck for stray files
[22:21:44]
[22:21:44] Folding@home Core Shutdown: EARLY_UNIT_END
[22:21:44] Finalizing output
[22:23:48] CoreStatus = 7B (123)
[22:23:48] Client-core communications error: ERROR 0x7b
[22:23:48] This is a sign of more serious problems, shutting down.

9/2/2008 7:02:31 PM

Here's a few suggestions: http://www.hardforum.com/showthread.php?p=1032930272

where are you adding your flags (-smp etc) from within the program, or via shortcut/registry?

are you using this one?
http://foldingforum.org/viewtopic.php?f=46&t=4913&st=0&sk=t&sd=a&hilit=Using+generic+mpiexec+calls

worst case you can always go back to the 5.91 until November 2nd (assuming they don't extend it again)

more info about 6.22r3: http://forum.uabit.com/showthread.php?t=141341<1--

[Edited on September 3, 2008 at 8:52 AM. Reason : -->

9/3/2008 8:47:55 AM

I created a shortcut and added "-smp 4" to the target. Some have said that it might be caused by an unstable overclock, but I have no signs of instability. Max load temps (running Prime95 max heat test) are less than 60C. Max temps running anything else at 100% load are under 50C. The Prime95 torture test doesn't detect any errors running for hours either.

Unlike others, the client doesn't even START folding on my computer. I'm just gonna stick to the GPU client for now.

9/3/2008 11:43:36 AM

One thing to try is putting the -smp (just do the -smp, nothing else except maybe -forceasm) in the arguments section when you're running the executable for the first time...instead of using the shortcut for it.

9/3/2008 3:11:49 PM

^tried that. no dice. It seems pretty intent on not working.

9/3/2008 4:32:46 PM

Just dropped in the 5.92 client and after crashing repeatedly trying to do the same WU, it gave up and downloaded a new WU, which appears to be working. We'll see how that goes.

9/3/2008 4:58:33 PM

Finished a SMP work unit in about 18-20 hrs. Looks like it was worth ~2000 pts
Any way to tell what the actual value of the work unit was?

9/4/2008 2:12:35 PM

installing FahMon to monitor your production is the easiest way

9/4/2008 2:26:33 PM

The GPU client is too CPU (single-thread performance) dependent! Playing around with fahMon I find that with my CPU clocked at 3GHz, I can pull about 5000 ppd from the GPU client. If I crank my CPU up to 3.8GHz (9x425), it'll pull over 6000 ppd, but there's no way I'd run it 24/7 with the CPU at almost 70C.

I'd bet that if they could split the CPU work from the GPU client into 2 balanced threads, my GPU could pull upwards of 8000ppd.

9/7/2008 1:29:02 PM

movin on up!!

9/7/2008 8:45:55 PM

Can you choose which projects you do and don't want to work for? I've noticed WUs from some projects offer less points, but require more computational work.

Projects 5506 and 5016 I get ~6000ppd rates from. Projects 5509, 5018, 5021 I get <4000 ppd rates from. It looks like only a few projects fuck you in the ass on points. Any way to refuse just those jobs?

9/9/2008 11:52:01 AM

http://folding.extremeoverclocking.com/team_summary.php?s=&t=46429

looks like Kentucky updated their servers.... at least we'll still be the #1 university supported by students and not by computer labs....

9/9/2008 11:54:29 AM

Just ran Windows Update on Server 2008 and none of the F@H clients work now.

9/11/2008 2:15:15 PM

Post the problem in the official forums so they can be notified about it. You may want to include what specific patch(s) it was.
http://foldingforum.org/

9/11/2008 3:16:52 PM

I installed 383 updates. A reinstall seems to have fixed it though. I only lost progress on the SMP unit.

9/11/2008 3:38:21 PM

PS, the GPU Client runs terribly under Server 2000ppd versus the 5-6000ppd under XP.

9/11/2008 4:44:25 PM

I'm starting to get 430pt GPU work units, and they're killing my production. My GPU client (8800GT) went from ~4700 PPD to 2500PPD. Here's a thread about the new GPU workunits: http://foldingforum.org/viewtopic.php?f=52&t=5452&st=0&sk=t&sd=a

Its just like with the SMP clients...he new client comes out and you get all kinds of points, then they scale it down. Used to get WAY more points with the 1760pt workunits on my SMP clients, now all I get is the 1920pt WUs which are only 50-60% of the PPD I got with the 1760pt workunits.

9/12/2008 10:21:29 AM

Apparently under Server 2008, the GPU client uses nearly no CPU time, so I can run 2 SMP clients. So the hit on the GPU client isn't so bad.

9/12/2008 10:45:18 AM

who's Jonathan?

9/13/2008 7:25:19 PM

man, my PS3 is pullin mad points right now

9/15/2008 6:30:43 PM

FUCK THE SMP CLIENT.

I've folded 4 units and I can't get the client to return the results, and now the client won't run either. It just hangs. I even did the whole "delete queue and work folder" (I actually renamed them), but it doesn't even download a new WU.

I tried using the "-send all" flag from several different versions of the client to see if I could just get it to send but no dice. Days of CPU time down the crapper. I've deleted the SMP client and don't plan on reinstalling.

9/15/2008 7:04:02 PM

joined w/ the PS3, woot.


btw, so when you run two clients once (PS3 + PC), you can use the same userid right? And it'll add it all up together?

[Edited on September 17, 2008 at 2:53 AM. Reason : fdf]

9/17/2008 2:43:51 AM

yep

9/17/2008 6:35:18 AM

i'm thinking about giving up on F@H for awhile.

a) version 6.22 is still shit, doesn't work on any of my computers, always get a missing work files or early unit end, 5.91 seems to conflict with my gpu client

b) gpu client is ok, but now with the newer work units, the old nvidia '2d slowdown' bug is back and normal usage with my computer is no longer possible as now the gpu client slows my computer down to old 486 speeds.

i'm sick of fixing this crap, once this stuff runs stable again, i'll come back, but it's too much work for so much frustration.

the only client that's stable these days is the PS3 client.

[Edited on September 17, 2008 at 11:36 AM. Reason : .]

9/17/2008 11:35:28 AM

aww don't give up the good fight!


a) i've posted on the official folding support board expressing out collective frustration about the smp client, and asked for the latest guide on how to get smp working. i'll let you know what they say.

b) I run the GPU on my home machine, and I just pause it when I wanna use it, and resule work when i leave. I think the points more than make up for the slight annoyance of having to pause/unpause it all the time.

9/17/2008 11:48:50 AM

^ Under Vista/Server 2008 the GPU client doesn't intrude at all. Sometimes I have to check to make sure it's still running.

9/17/2008 5:46:20 PM

it's not that it intrudes, it simply slows the computer down, this is only a known bug for nvidia-based gpu's and only on certain setups, the gpu client on my quadro at work works fine.

9/17/2008 6:02:29 PM

maybe just run the normal cpu client? slim down until its stable again.

I don't suppose there's a monitor utility that can somehow monitor the PS3 along w/my PC, eh?

9/17/2008 6:58:11 PM

i'm giving it one last try, this time i installed outside of /program files/ directory, hoping it was just a permissions based thing since i was on vista. i never tried it earlier because i'd never had problems until they updated it to 6.22, so whatever they did, they still added a bug imho.

9/17/2008 7:17:15 PM

^ I've re-installed the 6.22 SMP on two machines (MPICH) so far and all is well. This is what the guys over at the folding forum said to do:

Follow the guide here:
http://foldingforum.org/viewtopic.php?f=46&t=5057

Copy the latest executable into whatever folder you install into (and yeah, outside of program files is a good call). Heres the latest exe: http://foldingforum.org/viewtopic.php?f=46&t=4913

9/17/2008 9:20:14 PM

What kinda NS/Day are you PS3 guys getting? I'm reading up people saying things from 230 to 450NS/Day. I've done three units so far and they've all been about 30NS/Day...??

9/17/2008 9:25:14 PM

^no idea what an NS/day is, we all talk PPD here (points per day)

^^that's what i used... which they specifically say outside program files so that's what i did. we'll see. i wouldn't bet a dime on it though.

[Edited on September 17, 2008 at 9:29 PM. Reason : .]

9/17/2008 9:28:45 PM

Ok well from what i've done so far, two completed WUs, i got 20 points each and each took 8 hours... so thats like 60ppd. wtf? arent scores normally like in the hundreds?

NS/Day is shown on the PS3 client. its nanoseconds/day. so its showing the duration of protein folding simulation time per day of client processing.

http://www.overclock.net/overclock-net-folding-home-team/357943-ps3-s-ppd.html

600-1000 ppd here. wtf

[Edited on September 17, 2008 at 9:37 PM. Reason : dsd]

9/17/2008 9:35:05 PM

^^^^ Ok, reinstalling brought my second SMP client back to life. The first client deleted its work unit (which was completed days ago and I couldn't get it to send) and started a new one.

But it looks like it's running again.

9/17/2008 9:41:25 PM