SETI MB CUDA for Linux

[sunu]

Your app_info xml seem ok. If I'm not mistaken 6.4.5 doesn't support running two different versions of the same app simultaneously (e.g. 6.03 on the cpu and 6.08 on the gpu). If you already have 6.03s it will run them, but all new wus will be 6.08 (gpu).

[s52d]

Thanks!

Your app_info xml seem ok. If I'm not mistaken 6.4.5 doesn't support running two different versions of the same app simultaneously (e.g. 6.03 on the cpu and 6.08 on the gpu). If you already have 6.03s it will run them, but all new wus will be 6.08 (gpu).

Uh, ugly.... Maybe perl CPU_GPU_rebrand_V5.pl should be run daily?
One day of 6.4.5: seems to handle X and CPUs better.
- X is more responsive (still stops a bit once in a while, price for using GPU)
- one day with no errors
I guess overwriting libcuda.so helped a bit.

ls -lrt boinc/lib*
-rwxr-xr-x 1 iztok users 1150912 2009-01-16 23:30 libcufft.so.2*
-rwxr-xr-x 1 iztok users  252680 2009-01-16 23:30 libcudart.so.2*
-rwxr-xr-x 1 iztok users  254912 2009-06-15 03:10 libcudart.so*

a bit ugly to have two libcudart at the same time (and outside /usr/lib).



6.3.20 is then next to try, after SETI starts sending new job again.

BR
Iztok





Sligltly later: boinc 6.3.20, library copied to ~/boinc directory, smells ok.
validated WU:
http://setiweb.ssl.berkeley.edu/result.php?resultid=1263122254

X works, CPU load OK (shared among X/CPU-seti/GPU-seti). Now testing for "error ephidemia".

BR
Iztok

a bit later: happened.
two instances of seti-CUDA running, kill -TERM to boinc, 2 minutes to restart from crontab.
A stream of WUs gone with:


<message>
process exited with code 193 (0xc1, -63)
</message> 
<stderr_txt>

SETI@home MB CUDA 608 Linux 64bit SM 1.0 - r06 by Crunch3r :p

setiathome_CUDA: Found 1 CUDA device(s):
   Device 1 : GeForce GTX 260
           totalGlobalMem = 938803200
           sharedMemPerBlock = 16384 
           regsPerBlock = 16384     
           warpSize = 32
           memPitch = 262144
           maxThreadsPerBlock = 512
           clockRate = 1350000     
           totalConstMem = 65536
           major = 1             
           minor = 3
           textureAlignment = 256
           deviceOverlap = 1
           multiProcessorCount = 27
setiathome_CUDA: CUDA Device 1 specified, checking...
   Device 1: GeForce GTX 260 is okay
SETI@home using CUDA accelerated device GeForce GTX 260
setiathome_enhanced 6.01 Revision: 402 g++ (GCC) 4.1.2 20070925 (Red Hat 4.1.2-33)
libboinc: BOINC 6.5.0

Work Unit Info:
...............
WU true angle range is :  0.410807
Optimal function choices:
-----------------------------------------------------
name               
-----------------------------------------------------
              v_BaseLineSmooth (no other)
            v_GetPowerSpectrum 0.00019 0.00000
                   v_ChirpData 0.01607 0.00000
                  v_Transpose4 0.00651 0.00000
               FPU opt folding 0.00152 0.00000

SETI@home MB CUDA 608 Linux 64bit SM 1.0 - r06 by Crunch3r :p

setiathome_CUDA: Found 1 CUDA device(s):
   Device 1 : GeForce GTX 260
           totalGlobalMem = 938803200
           sharedMemPerBlock = 16384 
           regsPerBlock = 16384
           warpSize = 32
           memPitch = 262144
           maxThreadsPerBlock = 512
           clockRate = 1350000
           totalConstMem = 65536
           major = 1
           minor = 3
           textureAlignment = 256
           deviceOverlap = 1
           multiProcessorCount = 27
setiathome_CUDA: CUDA Device 1 specified, checking...
   Device 1: GeForce GTX 260 is okay
SIGSEGV: segmentation violation
Stack trace (16 frames):

setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x47cba9]
/lib64/libpthread.so.0[0x7f3f289e2f30]
/usr/lib64/libcuda.so.1[0x7f3f29454020]
/usr/lib64/libcuda.so.1[0x7f3f29459d84]
/usr/lib64/libcuda.so.1[0x7f3f2942310f]
/usr/lib64/libcuda.so.1[0x7f3f291aeb3b]
/usr/lib64/libcuda.so.1[0x7f3f291bf46b]
/usr/lib64/libcuda.so.1[0x7f3f291a7211]
/usr/lib64/libcuda.so.1(cuCtxCreate+0xaa)[0x7f3f291a0faa]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x5ace4b]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x40d4ca]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x419f23]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x424c7d]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x407f60]
/lib64/libc.so.6(__libc_start_main+0xe6)[0x7f3f28682526]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu(__gxx_personality_v0+0x241)[0x407be9]

Exiting...

</stderr_txt>

[sunu]

6.3.20 is then next to try, after SETI starts sending new job again.

Do you mean 6.6.20? 6.3.20 is pretty old now.

Try something else. Put in your app_info.xml

   <avg_ncpus>1.0000</avg_ncpus>
   <max_ncpus>1.0000</max_ncpus>

in all AKv8 entries and also make all cuda entries 1.0000.  Then put in your cc_config.xml

<ncpus>5</ncpus>

Test it and see how it goes.

[s52d]

Done.

let me wait a day or two if it stikes again.

Thanks,

Iztok

Of course: <core_client_version>6.6.20</core_client_version>

[Sp@r0]

Hello,

I have some problem with "SETI@Home MB NVIDIA CUDA V6.08", I run my PC under Fedora 11 x86_64, i have a GTS 250. I have setup the CUDA 2.2 driver with the toolkit and the SDK.
The SDK sample apps run without problems.

I use without problem the CPU optimized applications  AK V8 Linux 64 SSSE3  and it run very well, but with the cuda apps i have 2 big problems
* I have got calculation error on all my WU's within a second
* Boinc did not detect Cuda card if i launch it as service but it works if i launch it form a terminal

I have try lots of things with differents version of Cuda but now i'm seek, i do not know the reason of the calcultaion error because i do not know if there is an error log file

Can you give me some tips to solve my problem

PS : I m sorry for the quality of my english
 

[sunu]

What version is the nvidia driver you use?

[Sp@r0]

i use the folling version :
Cuda 2.2
Nvidia driver : 185.18.14
Boinc : 6.4.7 (come from yum, a lite bit strange the current version is 6.4.5 on the boinc's website)

[sunu]

Follow all steps (1-4) below:

1)  Use a newer boinc version. The latest is 6.6.36, http://boinc.berkeley.edu/download_all.php . I haven't checked it, I use 6.6.20, direct download link http://boinc.berkeley.edu/dl/boinc_6.6.20_x86_64-pc-linux-gnu.sh
2)  Make sure all the appropriate cuda libs from 2.2 toolkit

libcudart.so
libcudart.so.2
libcudart.so.2.2
libcufft.so
libcufft.so.2
libcufft.so.2.2

are in the projects/setiathome.berkeley.edu directory.

3)  Edit accordingly your ld.so.conf or the corresponding ld-something file of your distro with the above location of the cuda libs.

4)  Place a copy of the cuda client in one of the following locations:

/usr/local/sbin
/usr/local/bin
/usr/sbin
/usr/bin
/sbin
/bin
/usr/games

[Sp@r0]

Thanks a lots for your tips sunu,

it run very well know i have got my firsts valids WUs on my GTS250 (about 5~10 min for a WU).It use about 2 % of CPU time to run so that i can crunch 2 other seti WU on my C2D E7300.

Bye

[sunu]

Happy crunching Sp@r0 

[Tye]

Follow all steps (1-4) below:

Hi sunu, I'm running Ubuntu 9.04 64-bit and 2 9600 GSO's with the 4 driver packages that "nvidia-180-kernel-source_185.18.14-0ubuntu1_amd64.deb" is part of, and the 185.18.14 drivers seem to be working, but not for CUDA with Seti.  I've followed your steps, but keep getting workunits with the following errors (after <1 sec run time each):   Any ideas that might help???

http://setiathome.berkeley.edu/result.php?resultid=1293981772

<core_client_version>6.6.20</core_client_version>
<![CDATA[
<message>
process exited with code 193 (0xc1, -63)
</message>
<stderr_txt>

SETI@home MB CUDA 608 Linux 64bit SM 1.0 - r06 by Crunch3r :p

setiathome_CUDA: Found 2 CUDA device(s):
   Device 1 : GeForce 9600 GSO
           totalGlobalMem = 804585472
           sharedMemPerBlock = 16384
           regsPerBlock = 8192
           warpSize = 32
           memPitch = 262144
           maxThreadsPerBlock = 512
           clockRate = 1350000
           totalConstMem = 65536
           major = 1
           minor = 1
           textureAlignment = 256
           deviceOverlap = 1
           multiProcessorCount = 12
   Device 2 : GeForce 9600 GSO 512
           totalGlobalMem = 536608768
           sharedMemPerBlock = 16384
           regsPerBlock = 8192
           warpSize = 32
           memPitch = 262144
           maxThreadsPerBlock = 512
           clockRate = 1600000
           totalConstMem = 65536
           major = 1
           minor = 1
           textureAlignment = 256
           deviceOverlap = 1
           multiProcessorCount = 6
setiathome_CUDA: CUDA Device 1 specified, checking...
   Device 1: GeForce 9600 GSO is okay
SIGSEGV: segmentation violation
Stack trace (16 frames):
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x47cba9]
/lib/libpthread.so.0[0x7f8e8b5d8080]
/usr/lib/libcuda.so.1[0x7f8e8c04f020]
/usr/lib/libcuda.so.1[0x7f8e8c054d84]
/usr/lib/libcuda.so.1[0x7f8e8c01e10f]
/usr/lib/libcuda.so.1[0x7f8e8bda9b3b]
/usr/lib/libcuda.so.1[0x7f8e8bdba46b]
/usr/lib/libcuda.so.1[0x7f8e8bda2211]
/usr/lib/libcuda.so.1(cuCtxCreate+0xaa)[0x7f8e8bd9bfaa]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x5ace4b]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x40d4ca]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x419f23]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x424c7d]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu[0x407f60]
/lib/libc.so.6(__libc_start_main+0xe6)[0x7f8e8b2755a6]
setiathome-CUDA-6.08.x86_64-pc-linux-gnu(__gxx_personality_v0+0x241)[0x407be9]

Exiting...

</stderr_txt>
]]>

[sunu]

Hi Tye!

I don't have much time right now, I'll post back in a few hours.

[Tye]

Hi Tye!

I don't have much time right now, I'll post back in a few hours.

Thanks sunu - I'll check back tonight.  Hope everything's not too busy anymore.

[sunu]

Hi again Tye.

First of all get rid of that package. I don't know what it is or what it contains but it's better to use the "original" nvidia drivers. Go to synaptic and uninstall everything related to nvidia (you can keep xserver-xorg-video-nv). Some tips from nvidia:

If you wish to install the NVIDIA Linux graphics driver on a Debian GNU/Linux or Ubuntu system that ships with Xorg 7.x, please ensure that your system meets the following requirements:

    * development tools like make (build-essential) and gcc are installed
    * the linux-headers package matching the installed Linux kernel is installed
    * the pkg-config and xserver-xorg-dev packages are installed
    * the nvidia-glx package has been uninstalled with the --purge option and the files /etc/init.d/nvidia-glx and /etc/init.d/nvidia-kernel do not exist

If you use Ubuntu, please also ensure that the linux-restricted-modules or linux-restricted-modules-common packages have been uninstalled. Alternatively, you can edit the /etc/default/linux-restricted-modules or /etc/default/linux-restricted-modules-common configuration file and disable the NVIDIA linux-restricted kernel modules (nvidia, nvidia_legacy) via:

    DISABLED_MODULES="nv nvidia_new"

Additionally, delete the following file if it exists:

    /lib/linux-restricted-modules/.nvidia_new_installed

Get the driver from http://www.nvidia.com/object/linux_display_amd64_185.18.14.html . If you need help on the installation post back.


Ubuntu 9.04's kernel has some problems with our apps. Either compile your own kernel or use 8.10's kernel (2.6.27-14). Again post back if you need step by step instructions for that.

Also please post your xorg.conf and xorg.0.log.

[Tye]

Hmmm, I'll go ahead and try that later today, but I can tell you that my Ubuntu 9.04 will CUDA just fine with 180.44, though I lose 2 of my 4 CPUs to the overhead with the 180 drivers and CUDA for linux, using either the 2.1 *or* 2.2 libcudart and libcufft.  Also, I successfully compiled the nvidia CUDA SDK stuff and all those apps work fine, so I was hoping that would carry over to the Seti CUDA app, but I guess not...

I'm not where I can try using the driver from nvidia right now, but I'll work that later today and see what it does.  It'll hopefully work, but in that case it'll feel like "magic", since I don't know how all the CUDA SDK apps could work, but not the seti CUDA app...   

I've had the driver direct from Nvidia installed before so that should be no problem - just didn't have the other stuff in place to enable the CUDA to work correctly, so I rolled back to 180.44 and then saw this 185 package so figured that might work 'better'.   Guess not...