dont get points from SMP or GPU

[CaqKa]

Hi

its a while that i have last folded, i recently saw a new client is there, that automatical detects the gpu.
since i have a nvidia ion on asus 1201n, i have said to me to give it a try once again.

after installing the cuda drivers it found and configured everything automaticly, but somehow it does make me any points from the GPU or SMP line... only uniprocessor. CPU is atom 330 (2 cores)
my stats name is same as here

i am sure that it folded a minimum of 4 WU's on the automaticaly sugested smp+gpu configuration.

[7im]

Hello CaqKa, welcome to the forum.

Please post the *** System *** section of the log so we can see your setup. Also post some of the log showing where SMP and GPU are not making any progress.

You may have to open the FAHClient data folder, under the Logs folder to find that previous log data.

[CaqKa]

system: https://dl.dropbox.com/u/1183736/fah_system.png

(hmm how can i copy paste this text?)


older log: https://dl.dropbox.com/u/1183736/log-20 ... 140854.txt

[Jesse_V]

system: https://dl.dropbox.com/u/1183736/fah_system.png

That works, but FYI the System section is at the top of the log. If you don't see it when you scroll up, hit the Refresh button. For example, here is my System section:

Code: Select all

18:42:34:******************************* System ********************************
18:42:34:        CPU: Intel(R) Core(TM)2 Quad CPU Q9000 @ 2.00GHz
18:42:34:     CPU ID: GenuineIntel Family 6 Model 23 Stepping 10
18:42:34:       CPUs: 4
18:42:34:     Memory: 3.86GiB
18:42:34:Free Memory: 3.57GiB
18:42:34:    Threads: POSIX_THREADS
18:42:34: On Battery: false
18:42:34: UTC offset: -6
18:42:34:        PID: 1194
18:42:34:        CWD: /var/lib/fahclient
18:42:34:         OS: Linux 3.2.0-23-generic x86_64
18:42:34:    OS Arch: AMD64
18:42:34:       GPUs: 1
18:42:34:      GPU 0: NVIDIA:1 GT216 [GeForce GT 240M]
18:42:34:       CUDA: 1.2
18:42:34:CUDA Driver: 4020
18:42:34:***********************************************************************

[CaqKa]

here is the old log with system on top
https://dl.dropbox.com/u/1183736/log-20 ... 140854.txt

[7im]

I can see where the GPU was making good progress (All those lines with 2%, 3%, 4%), but not so much with the SMP client. Do you have a lot of other programs running at the same time? That would really slow down the SMP client.

[Rel25917]

I'm not sure where to start on this one so I'll just list what i got from the log.

cpu-usage v='50'

The atom is already a very slow cpu for folding running it at 50% usage will make it useless for smp. It appears that the log covers roughly 3 and a half to 4 days and in the time you made only 12% progress on the smp unit. At that rate you will miss both the preferred deadline and the final deadline resulting in zero points. With all the pauses I see in your log I'm not sure if you would make any deadlines even if set for 100% usage.

15:25:14:WU00:FS00:0x11:Folding@home Core Shutdown: UNSTABLE_MACHINE
07:49:37:WU02:FS00:0x11:Folding@home Core Shutdown: UNSTABLE_MACHINE

These are both from your gpu slot, one at 38% one at 93%! Both these failure came when the client was restarting after a pause leading me to believe checkpoints aren't getting saved right from all the pausing going on. There are several possible causes for this but the end result is that data does not get saved right. With all the pauses going on in your log chances of a complication like that go up a lot.

I would recommend running 2 uni slots, one for each real core you have. If you can cut down on the constant pausing you might start getting some gpu work done. If not I would get rid of the gpu slot too.

[iceman1992]

I thought the cpu usage % doesn't work for SMP? The atom 330 is very slow, IIRC it was the first generation atom, 2C/4T.. And the project it's working on is 6098, with 1593 base points which is quite large. A unit with that kind of base points will possibly take around 14 hours even on a fast quad core..

[bruce]

No, I have an older (and less powerful) Atom: 1c2t.

[iceman1992]

No, I have an older (and less powerful) Atom: 1c2t.

I'm guessing an atom N270? They're both launched in 2008.. I remember the 330 was more for nettops, and n270 for netbooks.

[CaqKa]

I am running also other programs, thats my main computer, so a lot of it is chrome.exe.
after i managed to get the gpu running i see that it slowed down the computer to an unacceptable rate. thats when i set the cpu usage to 50%. thats also the cause for the pauses. i am still experimenting with what will give me the optimal results so i can use the computer in a normal manner and will also make some wu progress before the deadlines.
a few days ago i set the cpu usage back to 100%.

yes the atom 330 was more a nettop cpu.. but that does not make an difference. it's still slow :>

here is the current log: https://dl.dropbox.com/u/1183736/log-20120911.txt

So my question remains, why are the WU's from GPU/SMP not counting?

[Rel25917]

So my question remains, why are the WU's from GPU/SMP not counting?

Because from the logs I have seen so far you have not been able to complete any gpu or smp units.

14:24:51:WU00:FS00:0x11:Folding@home Core Shutdown: UNSTABLE_MACHINE
08:19:47:WU02:FS00:0x11:Folding@home Core Shutdown: UNSTABLE_MACHINE
19:59:01:WU00:FS00:0x11:Folding@home Core Shutdown: UNSTABLE_MACHINE
20:14:38:WU03:FS00:0x11:Folding@home Core Shutdown: UNSTABLE_MACHINE

4 more from your current log. You did start making better progress on the smp unit up to 28% before you started the uni slot. I also see a lot of connection issues in the logs, do you keep this computer disconnected from the internet at times or anything? Or any chance you are in China? There was a big problem for users there for a while but sounds like its fixed now.

PRCG from the 4 failed gpu units in the current log.
5768 (Run 11, Clone 12, Gen 3290)
10502 (Run 106, Clone 0, Gen 679)
10503 (Run 289, Clone 0, Gen 1126)
5769 (Run 3, Clone 126, Gen 4415)

From the older log.
10501 (Run 310, Clone 0, Gen 1057)
10502 (Run 170, Clone 0, Gen 628)

Thats 6 units your gpu has failed to process, all 6 seem to have failed after being paused.

[CaqKa]

so these WU havent been to the end if i understand that correclty?
whats PRCG?
No i am not from china, the computer online, whenever its ON, and the ISP has no problems :> (does not happen so much, so internet connection from m side should not be a problem).

what can i make that the computer will be more stable in FAH terms?

[Rel25917]

so these WU havent been to the end if i understand that correclty?

Correct

whats PRCG?

Project Run Clone Generation. http://folding.stanford.edu/English/FAQ ... tion#ntoc6

what can i make that the computer will be more stable in FAH terms?

As slow as the atom is I would run 2 uni slots instead of an smp slot. For the gpu my guess is something is preventing checkpoints from being written to the disk properly as it seems to always fail after a pause. My only idea for that is if your running antivirus is to keep it from scanning the work folder. Unless someone else has a better idea it may be best to just not run the gpu slot.

[CaqKa]

As slow as the atom is I would run 2 uni slots instead of an smp slot.

will try that. did not come to m mind :>

For the gpu my guess is something is preventing checkpoints from being written to the disk properly as it seems to always fail after a pause. My only idea for that is if your running antivirus is to keep it from scanning the work folder. Unless someone else has a better idea it may be best to just not run the gpu slot.

hmm no AV on my computer ...
other programs that could make some extensive work with the disk... thats utorrent, dropbox, chrome, launchy, googledrive sync. others are minor...

seems like i would need a reinstall. is the v7 client avaible also on linux (ubuntu), so i could test that also?