SMP folding keeps hanging?

MrTRiotMrTRiot Northern Ontario Icrontian
edited July 2011 in Folding@Home
I'm currently running the folding@home smp console client and it keeps hanging. I'm not too sure why exactly. It'll go for maybe an hour or two and then just stall and stop working. All 4 cores are still in use but the progress just halts...


Any suggestions as to what to do?


CPU is an AMD 965 BE @ 3.4Ghz
«1

Comments

  • csimoncsimon Acadiana Icrontian
    edited May 2011
    Mr TRiot wrote:
    I'm currently running the folding@home smp console client and it keeps hanging. I'm not too sure why exactly. It'll go for maybe an hour or two and then just stall and stop working. All 4 cores are still in use but the progress just halts...


    Any suggestions as to what to do?


    CPU is an AMD 965 BE @ 3.4Ghz

    Please give more details ...like show your config and such. Also, what indication do you have to show that it is hanging? ...and is your CPU stable?
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited May 2011
    Standard config according to the website with "-smp" enabled and big workloads. Indication that it's hanging, it stops updating in the console yet the CPU load is still at 100% in the task manager. Happened last night when I went to sleep. Woke up and it started hanging an hour and a half after I looked at it...

    CPU is stable. Hasn't been overclocked at all. It's not over heating either, stays at 40c even under a full load
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    It stops updating in the console? What do you mean?
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited May 2011
    Stops updating the % points. It usually goes, "Completed 0 out of 100000 steps (0%)" then to 1%, then 2% and so on. When it was hanging it stopped updating in the console...

    Also, I figured with a quad core I'd be getting a lot more PPD but I haven't really seen it :-/
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    How are you checking the console? What program are you using to open it and see it? Are you closing it before checking it again? For instance ...if I open FAHlog.txt in notepad it can't update unless I close it ...because it is open. If I leave it open all night notepad won't be updated again until I reopen it, although everything is working fine. Chances are if your CPU is 100% then it is folding ok.
    Also, how do you know how much PPD you are putting out? You only start getting bonus points after you fold about 10 successful units. Best thing to do is leave everything alone until you have done so.
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited May 2011
    I'm checking the console by opening up the program. I'm using a command line folding@home program from the stanford website...
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    Ah ...so you're running it manually and just opening the console that way. I gotcha.
    Is there anything else going on that could take over from the client? Like another software? Maybe a screensaver? Like something else robbing the CPU cycles.
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited May 2011
    Only thing I can think of is low power/stand by mode. Would that cause folding to hang??

    Also, as state previously, is there any possible reason why my PPD is still extremely low?? :(
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    PPD could be low if you haven't turned in 10 wu's yet to get bonus points as stated previously.
    However and also, any program that runs in the background will interupt F@h with "standard" settings. I wouldn't modify the settings though I would look to see if anything else is running first.

    Here is a sample of my client.cfg to show my configuration:

    [settings]
    username=csimon
    team=93
    passkey=255eed34454bef506b40e2170805688d
    asknet=no
    machineid=1
    bigpackets=big
    extra_parms=-smp -forceasm
    local=190

    [http]
    active=no
    host=localhost
    port=8080
    usereg=no

    [clienttype]
    type=3

    [core]
    checkpoint=30
    addr=

    Here is a sample of my taskmanager to show how much CPU usage is going to F@h.

    Sorry about the delay. I needed a Sunday afternoon afterbeernap. :cheers2:
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited May 2011
    [settings]
    username=MrTRiot
    team=93
    passkey=
    asknet=no
    machineid=1
    bigpackets=big
    extra_parms=-smp
    local=2

    [http]
    active=no
    host=localhost
    port=8080
    usereg=no

    [core]
    priority=0
    cpuusage=100
    disableassembly=no
    checkpoint=15
    ignoredeadlines=no
    nocpulock=0
    addr=

    [power]
    battery=no

    [clienttype]
    memory=4000
    type=0

    ...What kind of beer were you drinking? Something good I hope!
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    Check your taskmanager as well.
    You have no passkey generated? Well that's how you get the big points son!
    Check out the Quick Return Bonus Plan Qualifications.

    Today I was drinking Blue Moon Summer Honey Wheat with a twist of orange, chilling in the cool pool, and enjoying it quite immensly! http://www.bluemoonbrewingcompany.com/

    Also, here is the passkey generator here http://fah-web.stanford.edu/cgi-bin/getpasskey.py

    I would think you may want to figure out what's causing the "hanging" first however.
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited May 2011
    I tried to add a pass key to my folding but it won't recognize smp anymore...

    Do I have to wait for a new WU before using my passkey?
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    I didn't. The passkey is ver tricky most of the time. I just keep trying until I get it. Also, copy and paste it somewhere so that you can have it in the future.
  • _k_k P-Town, Texas Icrontian
    edited May 2011
    Make sure you have the most up to date exe for your client.

    The simple thing to do when you start having a cmd line client screwing up in any way is find the working directory of the install, where the work folder is along with the config file and the actual executable, and delete all unnecessary files. This means deleting the work folder, FAHlog, FAHlog2, fahcore, etc. The only thing that should be left is the folding@home-win32-x86.exe and client.cfg. Launch the SMP client again and it will download everything that it needs.

    The other thing that helps a lot in trouble shooting, I do this for all of my f@h installs now, is to leave a shortcut to launch the SMP client in the program directory with the configonly flag on it so you can easily and quickly change the clients configuration. How to do this is create a shortcut icon and open the properties for the shortcut. In the target line put -configonly at the very end of the target line with a space after the last "("C:\Program Files (x86)\Folding@Home Windows SMP Client V1.01\Folding@home-Win32-x86.exe" -configonly).

    That was simply covering all of the basic stuff, now to work on the problem.

    Make sure that your cool 'n quiet junk is turned off for the processor in BIOS and any software AMD that allows you to OC or manipulate proc settings. Double check the power profile you are using and that nothing ever gets turned off except the monitor.

    Make sure to put a passkey on the smp client and add -verbosity 9 to your extra parameters, this increases the amount of information printed in the output.
  • TushonTushon I'm scared, Coach Alexandria, VA Icrontian
    edited May 2011
    EDIT: started this earlier then finished cleaning, hence why a lot is repeated from k's post
    Mr TRiot wrote:
    I tried to add a pass key to my folding but it won't recognize smp anymore...

    Do I have to wait for a new WU before using my passkey?
    No.

    Try the following:
    Close out any existing clients (either close window or end task in taskmgr)
    Open cmd prompt and navigate to folding folder
    Launch with "fah6.exe -smp -configonly" or whatever your exe is named

    Make sure your settings include the following:
    username, passkey, either leave priority at default (idle) or change it to low, big for WU type, yes to the advmethods question.

    advanced options:
    the second to last line where you launch it should have -smp -verbosity 9 on it.
  • csimoncsimon Acadiana Icrontian
    edited May 2011
    Thanks for the backup _k_ & Tushon ...I knew I could count on you guys!
    Tushon I was going to PM you next ...hehe :fold:
  • _k_k P-Town, Texas Icrontian
    edited June 2011
    Any word on this? I was looking at your production numbers and saw some 481s which should mean you are turning in SMP2 WUs, projoect 6050s series. Also the 9800 GT you have listed on your profile will net you about 5k in ppd.
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited June 2011
    It seems to be working and not hanging anymore. I think the problem was my computer going into stand by/power saving mode. My PPD still isn't what it should be. Fahmon says I should be getting 6k or more but that isn't happening. Hopefully it'll turn around once I finish 10 WU...
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited June 2011
    Oh, and any recommendations on a good GPU client?
  • _k_k P-Town, Texas Icrontian
    edited June 2011
    What do you mean a good GPU client, you can run either GPU2 or GPU3 clients? However if you ever upgrade and stick with NVidia and buy a fermi based card you will have to change to GPU3.
  • MrTRiotMrTRiot Northern Ontario Icrontian
    edited June 2011
    Oh ok. I wasn't sure if I could run GPU3 with my video card
  • TushonTushon I'm scared, Coach Alexandria, VA Icrontian
    edited June 2011
    Run GPU3, see what your PPD is like, then run GPU2 and see if it is better
    besides the P11177 and 11179 the PPD is roughly the same. on those 2 WU's you should see mid 6k range on the 9800GT if you can get the shader clock to 1728 or 1782.

    as far as the benefits of GPU3, theres very little vram usage, temps are far lower, and you can run hardware accel applications while folding at the same time(something i couldnt do with the gpu2 client).

    disadvantages to gpu3 is the cpu usage is slightly higher. 1-3% but for me the advantages out way the excess cpu usage. i lose roughly 1000 PPD on my standard SMP client when running both of my 8800GS's.
  • _vid__vid_ alabama
    edited June 2011
    finally picked up a quad amd 955be and almost had first wu completed with it(91%) and it threw an error. went into bios and lowered cpu voltage from what it had on auto a bit and trying again. anyone else have any helpful pointers? im just running smp with big, passkey and -smp flag. tia!
  • TushonTushon I'm scared, Coach Alexandria, VA Icrontian
    edited June 2011
    There will be (a very limited) some that are bad and error, but lowering the CPU voltage from the "normal" will not help. Set it back to auto (or w/e the default is for your proc) and you can try running a Prime95 stress test overnight or OCCT for a similar period of time.
  • _vid__vid_ alabama
    edited June 2011
    well, auto had it set at like 1.5041 or something like that with no o/c whatsoever and it was getting rather hot which i thought might have causes the error. at any rate, ill try setting it back to auto next time i reboot. oh, first smp unit results and in the process of being returned as i type this! woohoo!
  • TushonTushon I'm scared, Coach Alexandria, VA Icrontian
    edited June 2011
    Good deal. If you aren't overclocking, just leave it on auto so the proc will dictate what it should be getting. If you list your proc, we should easily be able to find out what it should be, and if it is getting hot at it's "normal" level of voltage, then there is an air circulation problem.
  • _vid__vid_ alabama
    edited June 2011
    Tushon wrote:
    Good deal. If you aren't overclocking, just leave it on auto so the proc will dictate what it should be getting. If you list your proc, we should easily be able to find out what it should be, and if it is getting hot at it's "normal" level of voltage, then there is an air circulation problem.


    its an amd phenom ii 955 black edition. have an asus xfire iii board and was reading that asus mobos sometimes set slightly higher voltages on auto than suggested default for some reason. upon inspection, it appears i had put a tad too much as5 on the hsf so i cleaned it all off and off cpu itself and temps are a bit lower. i even took the opportunity to clean out the dust thats collected in case fans as well. im thinking i would still benefit from an aftermarket cooler as opposed to the stock one that this came with. its okay at best but i seriously dont think i could even attempt an oc with it. do you or anyone else have any good suggestions for a solid air based hsf for somewhere ~$30ish dollars or so? id like to get this bad boy running nice and cool so i can attempt even a small o/c to get those smp units turned in even faster! (sorry if i sound excited, this is my first quad)

    *edit* i apologize but i neglected to mention that i am running latest bios as well since a friend asked me that earlier.
  • _k_k P-Town, Texas Icrontian
    edited June 2011
    955 BE
    Voltages 0.850-1.40V
    Max Temps (C) 62'C

    What is the vcore range from idle to load?
  • TushonTushon I'm scared, Coach Alexandria, VA Icrontian
    edited June 2011
    _vid_ wrote:
    do you or anyone else have any good suggestions for a solid air based hsf for somewhere ~$30ish dollars or so? id like to get this bad boy running nice and cool so i can attempt even a small o/c to get those smp units turned in even faster!

    Nice list
  • _vid__vid_ alabama
    edited June 2011
    @_k_ i have voltages set to 1.350 in bios and it seems to stay constant from what i can tell.

    @tushon thanks for the list, ive read up some people talking up the coolermaster hyper 212 saying great things about it, so i might give that one a try.

    many thanks for all the help!
Sign In or Register to comment.