SETI MB CUDA for Linux

[pp]

Also, as BOINC starts 2 normal CPU crunchers on my C2D E4400 and additionally SETI-CUDA grabs one of CPU for 100%, the crunchers start fighting for second CPU and all thing goes very slowly including X-server response time.

Finally, I tried Slackware64-current, and so I tried x86_64 version of Seti-Cuda with the same hardware.
The symptoms are the same - slowdown of X-server response time, so I don't like it....

I see the same behaviour on Gentoo AMD64. I've disabled all use of CPU in my profile, still the client  takes 100% CPU with its single CUDA-process. For now I manually lower the speed of the CPU to avoid having the fan go crazy on me :-) BOINC version is 6.4.5. Any insight into this would be appreciated.
/pp

[sunu]

Can you give a link to your host?

[pp]

Can you give a link to your host?

If that was meant for me you need to explain that to me like the noob I am.    You want me to provide more detailed system info? Just let me know what you need.

[sunu]

For example my host: http://setiathome.berkeley.edu/show_host_detail.php?hostid=3281360

Can you give a link to yours?

[pp]

I feel stupid now :-)
http://setiathome.berkeley.edu/show_host_detail.php?hostid=4974779

[sunu]

Ok pp, regarding your first message, if you're willing to run the CUDA app you'll have to accept a certain slowdown in your desktop.

Now tell me what exactly do you want to run? You'll get the highest RAC (credits) running two astropulse instances in your CPU, and multibeam in your GPU.

What version of nvidia drivers do you have? The latest are 185.18.14

Also a good move would be to upgrade your boinc. I see that they have released the 6.6.36 version. I haven't used it and I don't know how it is, but I've read some bad things with various 6.6.3x versions in windows. I use 6.6.20, it's supposedly beta but I don't have any problems. Direct link http://boincdl.ssl.berkeley.edu/dl/boinc_6.6.20_x86_64-pc-linux-gnu.sh . If you want the latest go here http://boinc.berkeley.edu/download_all.php .

[pp]

I don't mind the slowdown, it's a spare machine I only use for gaming sporadically. I participated in Seti Classic for 6 years until they shut down and I only started with Boinc a few days ago when a friend asked for some Linux assistance and my competitive spirit was awaken again. I'm still trying to learn the terminology and the overall framework structure.
The old software versions are the once currently in the Gentoo Portage tree and I rarely have use for something outside the tree but I'll look into it if you recommend it. It didn't take long to find this forum through some googling so I'm sure there are plenty of information in here that I will find useful but some pointers to good threads covering the topics you recommend would be appreciated. Like what clients to use and how to set the whole thing up. I assume app_info.xml is the key somehow. Is "multibeam" the SETI@Home Enhanced client? What's the difference between Astropulse and Astopulse v5 and which one should I use? Sorry for all the newbie questions... 

[sunu]

Especially for cuda, it's best to use the latest versions of boinc and nvidia drivers. If portage doesn't have them then you'll have to get them elsewhere. For boinc I gave you the links above.

Your host page identifies the GPU driver as 1600416869. This number isn't even remotely close to the numbering system that nvidia uses so I can't understand what you have. The latest official driver is 185.18.14. If portage doesn't have it, grab it from nvidia's page or from ftp://download.nvidia.com/XFree86/Linux-x86_64/185.18.14/ (get the pkg2 one).

Yes multibeam is the seti@home enhanced client. Astropulse_v5 is a newer version of plain astropulse. I don't think there are workunits for the plain astropulse anymore so there is no point installing it. Right now there aren't also any workunits for Astropulse_v5 because we crunched them all  ,  but eventually there will be.

What sse capabilities does your cpu have?

[arkayn]

My guess would be SSE3 or PNI on that host.

[pp]

Yes, this is an SSE3 capable Athlon64. My nvidia drivers are 180.60 and if you know what the boinc client are looking for to identify them, I can inform the maintainer.
I compiled the trunk version of boinc successfully yesterday but I'll have to go through that Makefile in detail before I swap to it. I'm experimenting with GCC 4.4.0 so I'm trying to keep the number of installations outside Portage to a minimum for now to make it easier to iron out the bugs. (No, I'm not a Gentoo developer but I try to help out whenever I can.)

The app_info.xml that came with the CUDA client contains several identical sections but with different version numbers. Aren't they redundant?

# grep version_num app_info.xml
<version_num>528</version_num>
<version_num>603</version_num>
<version_num>605</version_num>
<version_num>606</version_num>
<version_num>608</version_num>

I downloaded the SSE3-capable version of Astropulse v5 and there's a new app_info.xml included with it. Can I just add that info to my current app_info.xml to have boinc use both programs? To have it use two instances on my dual core, I assume that's controlled by the "100% CPU" setting in my S@H profile?

Thanks for your help and please leave a few Astropulse WUs for me ;-)

/pp

[sunu]

Yes, this is an SSE3 capable Athlon64. My nvidia drivers are 180.60 and if you know what the boinc client are looking for to identify them, I can inform the maintainer.

It doesn't matter much what boinc shows. Get the newer nvidia driver if you can.

I compiled the trunk version of boinc successfully yesterday but I'll have to go through that Makefile in detail before I swap to it.

Why compile? Use the ready packages from the links I gave you above.

The app_info.xml that came with the CUDA client contains several identical sections but with different version numbers. Aren't they redundant?

# grep version_num app_info.xml
<version_num>528</version_num>
<version_num>603</version_num>
<version_num>605</version_num>
<version_num>606</version_num>
<version_num>608</version_num>

Some of them maybe, but leave them as they are. It doesn't matter.

I downloaded the SSE3-capable version of Astropulse v5 and there's a new app_info.xml included with it. Can I just add that info to my current app_info.xml to have boinc use both programs?

Get the 64bit one if you haven't already. It's much faster than the 32bit one. Yes add the info to your current app_info.xml.

To have it use two instances on my dual core, I assume that's controlled by the "100% CPU" setting in my S@H profile?

That's a bit more complicated. If you want to run 2+1 (that is 2 astropulse in your cpu and 1 multibeam in your gpu) you'll need to get the 185.18.14 nvidia driver just to begin with. More tinkering will be needed but we are here to help. If you stay with your current 180.60 you can only use 1+1 (1cpu + 1gpu).

[pp]

Thanks sunu. Plenty of valuable information to work with there. I'll see if I can try to isolate the cpu hog first before I try Astropulse.

About the compilation... that's what working with Gentoo for a few years does to you. I didn't even consider the precompiled binary...  Well, no harm done, it was a fun experience...

[sunu]

I'll see if I can try to isolate the cpu hog first before I try Astropulse.

What cpu hog?

[Richard Haselgrove]

Didn't early Linux CUDA builds take 100% CPU? I though that was cured by the 185 driver.

[pp]

I'll see if I can try to isolate the cpu hog first before I try Astropulse.

What cpu hog?

I was referring to the behaviour dtiger reported last in his post a few pages back in this thread: http://lunatics.kwsn.net/linux/seti-mb-cuda-for-linux.msg14847.html#msg14847
setiathome-CUDA is using 100% of one of my my regular CPU's cores even though no actual work is done there. Richard's idea about nvidia-drivers is interesting. Was that ever confirmed? I actually found this forum looking for a solution to that problem...