SETI MB CUDA for Linux

[sunu]

I don't know whether 270.30 supports the new GTX590, but even if it doesn't I'm sure the next driver release from nvidia will support it. What doesn't work currently is nvidia-smi.

[riofl]

cool so they finally got things going. actually i think i asked the wrong questin. i said driver didnt i.... im severely overworked.... i meant the seti software.  i can assume it now supports fermi. it has been long enough.

[riofl]

this is probably the wrong place to ask, but in boinc manager v 6.12.27 clicking the X in the upper right of the window does not cancel the app but minimizes it. this is not acceptable bahavior.. is there a way to change that? it is highly irritatintg.

[sunu]

I'm still with 6.10.58 and don't know what's happening with newer versions. Usually with boinc you can use managers with a different version than boinc. Maybe try an older boinc manager?

I've stopped using boinc manager. You can do many things with boinccmd. For example ./boinccmd --get_file_transfers shows you all pending uploads-downloads.

[riofl]

yeah im gonna go back to an earlier version. they also changed the order of the display columns and i found it lots more difficult than the older ones.. might just be because im used to the old style but it is not as quick to gather data i want to have like the old one was with important columns next to each other. in the new one they are spread across the window. they also removed the messages log and made it a menu option only and replaced the empty space with a notices tab which mirrors the messages on their website instead. yuk.

[Terror Australis]

Hi All,
As posted on the SAH board.

I've just put a new cruncher together. QX9650, Gigabyte X38 Mobo, Mandriva 2010.2 PWP Linux, 260.10 Drivers (the latest in the Mandriva repository) BOINC 6.10.58, Aarons Linux App and 2 GTX580's.

My problem is this. The system and BOINC both recognise there are 2 video cards and BOINC runs 2 GPU units. However they are both running on the one card. It's the card in the first PCIE socket that's doing all the work. The one in socket 2 stays dead cold. Using a cc_config file made no difference

Other things I've noticed.
In the xorg.conf files the Busses/Sockets are transposed, Socket 1 appears as socket 2 and visa versa. If I only use one card the system behaves perfectly.
Swapping the cards made no difference, it's still only slot 1 that does work

Any clues from the Linux fraternity ?

T.A.

[sunu]

My problem is this. The system and BOINC both recognise there are 2 video cards and BOINC runs 2 GPU units. However they are both running on the one card. It's the card in the first PCIE socket that's doing all the work. The one in socket 2 stays dead cold. Using a cc_config file made no difference

Your tasks say otherwise:
Device1: http://setiathome.berkeley.edu/result.php?resultid=1986746872
Device2: http://setiathome.berkeley.edu/result.php?resultid=1986700701

[Terror Australis]

Thats my point. As far as the system is concerned it IS using two cards, but it appears that the second card BOINC uses is a "virtual" card that running on the card in slot 1.

The second card is way too cold to be doing any work. The NVidia monitor reports a temp of 60deg for card one and a temp of 30 for card 2. I can't believe a GTX580 running at full blat would run that cool 

 The crunching times compare to those that are running 2 tasks on the one card as well. Time drops by half when only one card is fitted and only one task is running.

[sunu]

Now I get it.

Give me the output of :
ps -o cmd --no-headers -p $(pgrep setiathome)
and
nvidia-smi -q

Also post your xorg.conf and your xorg log.

I've never used a distribution provided nvidia driver. I always use nvidia's own driver. That latest official version is 275.09.07. Try that.

[Terror Australis]

Now I get it.

Give me the output of :
ps -o cmd --no-headers -p $(pgrep setiathome)
and
nvidia-smi -q

Why didn't you just say you wanted the Driver version and the CUDA app ?

V260.16 driver and setiathome-6.11.x86_64-pc-linux-gnu_cuda32 app

Also post your xorg.conf and your xorg log.

I've never used a distribution provided nvidia driver. I always use nvidia's own driver. That latest official version is 275.09.07. Try that.

[sunu]

Why didn't you just say you wanted the Driver version and the CUDA app ?

I don't. Please provide what I say in my previous post.

I have to go for a few hours. I'll check back later.

[Terror Australis]

Now I get it.

Give me the output of :
ps -o cmd --no-headers -p $(pgrep setiathome)
and
nvidia-smi -q

Why didn't you just say you wanted the Driver version and the CUDA app ?

Driver is 260.16 and CUDA App is: setiathome-6.11.x86_64-pc-linux-gnu__cuda32

I've never used a distribution provided nvidia driver. I always use nvidia's own driver. That latest official version is 275.09.07. Try that.

I tried installing that driver but the NVidia installer is a pain, the xorg.conf it creates doesn't work (X crashes on start up) and the problem is too subtle for me to pick up quickly.

Also post your xorg.conf and your xorg log.

Will do but I'll have to reinstall the 2nd card first to get meaningful files. will post them tomorrow.

It's only BOINC that has a problem with the 2nd card, as a video device it work perfectly, desktop and movies are fine. The system and BOINC know there is 2 cards, it's only that both instances of the app run on the same card without realising it. I'm suspecting a weird bug in the app.

T.A.

[sunu]

Also what version are your cuda libraries?

2.6.33 is almost a year and a half old. Why don't you use a newer kernel?

[Terror Australis]

Also what version are your cuda libraries?

2.6.33 is almost a year and a half old. Why don't you use a newer kernel?

Cos I work on the "if it ain't broke don' fix it" principal Plus I don't have the time to spend compiling kernels and drivers. Repository kernels may be slow updating but they work with a minimum of fuss (usually).

FYI
[root@Geekbox-Q9650 brodo]# ps -o cmd --no-headers -p $(pgrep setiathome)
../../projects/setiathome.berkeley.edu/setiathome-6.11.x86_64-pc-linux-gnu__cuda

[root@Geekbox-Q9650 brodo]# nvidia-smi -q

==============NVSMI LOG==============


Timestamp         : Sat Jul  9 13:05:33 2011

Driver Version         : 260.19.36


As I said Driver and app versions 



The Xorg conf and log files are attached. There are no glaring errors that I can see. The video system works perfectly for movies and games. The only app that has any trouble with it is BIONC. It's a weird problem. I suspect it's a bug in the Linux app and the way it handles multiple instances running on multiple devices.

Under Windows, the same hardware runs without problems and both cards are used. Compare these 2 machines Windows and Linux . It's exactly the same hardware, only the HDD with the operating system was changed.

T.A.

[sunu]

Is this with only one GPU? Some quick notes:

1) Why do you run as root?
2) There is no "--device x" flag on the running app. Is this because of small terminal window? Make it bigger and recheck.
3) This isn't what I would expect from nvidia-smi. Sample ouput:
==============NVSMI LOG==============

Timestamp                       : Sun Jul 10 17:19:00 2011

Driver Version                  : 275.09.07

Attached GPUs                   : 3

GPU 0:3:0
    Product Name                : GeForce GTX 295
    Display Mode                : N/A
    Persistence Mode            : Disabled
    Driver Model
        Current                 : N/A
        Pending                 : N/A
    Serial Number               : N/A
    GPU UUID                    : N/A
    Inforom Version
        OEM Object              : N/A
        ECC Object              : N/A
        Power Management Object : N/A
    PCI
        Bus                     : 3
        Device                  : 0
        Domain                  : 0
        Device Id               : 5EB10DE
        Bus Id                  : 0:3:0
    Fan Speed                   : N/A
    Memory Usage
        Total                   : 895 Mb
        Used                    : 265 Mb
        Free                    : 630 Mb
    Compute Mode                : Default
    Utilization
        Gpu                     : N/A
        Memory                  : N/A
    Ecc Mode
        Current                 : N/A
        Pending                 : N/A
    ECC Errors
        Volatile
            Single Bit           
                Device Memory   : N/A
                Register File   : N/A
                L1 Cache        : N/A
                L2 Cache        : N/A
                Total           : N/A
            Double Bit           
                Device Memory   : N/A
                Register File   : N/A
                L1 Cache        : N/A
                L2 Cache        : N/A
                Total           : N/A
        Aggregate
            Single Bit           
                Device Memory   : N/A
                Register File   : N/A
                L1 Cache        : N/A
                L2 Cache        : N/A
                Total           : N/A
            Double Bit           
                Device Memory   : N/A
                Register File   : N/A
                L1 Cache        : N/A
                L2 Cache        : N/A
                Total           : N/A
    Temperature
        Gpu                     : 74 C
    Power Readings
        Power State             : N/A
        Power Management        : N/A
        Power Draw              : N/A
        Power Limit             : N/A
    Clocks
        Graphics                : N/A
        SM                      : N/A
        Memory                  : N/A


I don't remember how nvidia-smi behaved with 260.xx. As I said try a newer driver.

I'll check the xorgs later.