SETI MB CUDA for Linux

[kpolberg]

Well didn't work to well, actually cuda seems to be working fine. But XBMC together with seti cuda was no hit

[Freggel_Buster]

I tried to Install  the 32bit CUDA app from Crunch3r.(setiathome-CUDA-6.08.i686)

But i need a Hand to Install it properly.

I followed the Instructions , like this.

1. I use the Nvidia Driver V180.44 from Ubuntu 9.04 32Bit
2. I put the whole cudalibs folder to /usr/lib
3. modified the ld.so.conf and start ldconfig
4. install the CUDA Toolkit V2.1
5. copy the setiathome-CUDA-6.08.i686-pc-linux-gnu and app_info.xml Files
to /var/lib/boinc-client/projects/setiathome.berkeley.edu
6. ldd setiathome-CUDA-6.08.i686-pc-linux-gnu shows no Errors or Warrnings

But it doesn't Work for me. :-(
Got this:

Do 16 Apr 2009 00:08:25 CEST||Starting BOINC client version 6.4.5 for i686-pc-linux-gnu
Do 16 Apr 2009 00:08:25 CEST||log flags: task, file_xfer, sched_ops
Do 16 Apr 2009 00:08:25 CEST||Libraries: libcurl/7.18.2 OpenSSL/0.9.8g zlib/1.2.3.3 libidn/1.10
Do 16 Apr 2009 00:08:25 CEST||Data directory: /var/lib/boinc-client
Do 16 Apr 2009 00:08:25 CEST|SETI@home|Found app_info.xml; using anonymous platform
Do 16 Apr 2009 00:08:26 CEST||Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4800+ [Family 15 Model 107 Stepping 1]
Do 16 Apr 2009 00:08:26 CEST||Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt rdtscp lm 3dnowext 3dnow pni cx16 lahf_lm cmp_legacy svm extapic cr8_legacy 3dnowprefetch
Do 16 Apr 2009 00:08:26 CEST||OS: Linux: 2.6.28-11-generic
Do 16 Apr 2009 00:08:26 CEST||Memory: 1.97 GB physical, 2.97 GB virtual
Do 16 Apr 2009 00:08:26 CEST||Disk: 22.91 GB total, 16.43 GB free
Do 16 Apr 2009 00:08:26 CEST||Local time is UTC +2 hours
Do 16 Apr 2009 00:08:26 CEST||Not using a proxy
Do 16 Apr 2009 00:08:26 CEST||Can't load library libcudart
Do 16 Apr 2009 00:08:26 CEST||No coprocessors
Do 16 Apr 2009 00:08:26 CEST|Einstein@Home|URL: http://einstein.phys.uwm.edu/; Computer ID: 1867036; location: home; project prefs: default
Do 16 Apr 2009 00:08:26 CEST|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 4879716; location: (none); project prefs: default
Do 16 Apr 2009 00:08:26 CEST|Milkyway@home|URL: http://milkyway.cs.rpi.edu/milkyway/; Computer ID: 56211; location: home; project prefs: default
Do 16 Apr 2009 00:08:26 CEST||General prefs: from Milkyway@home (last modified 08-Apr-2009 00:18:12)
Do 16 Apr 2009 00:08:26 CEST||Computer location: home
Do 16 Apr 2009 00:08:26 CEST||General prefs: no separate prefs for home; using your defaults
Do 16 Apr 2009 00:08:26 CEST||Reading preferences override file
Do 16 Apr 2009 00:08:26 CEST||Preferences limit memory usage when active to 1006.22MB
Do 16 Apr 2009 00:08:26 CEST||Preferences limit memory usage when idle to 1811.20MB
Do 16 Apr 2009 00:08:26 CEST||Preferences limit disk usage to 10.00GB
Do 16 Apr 2009 00:08:26 CEST||file projects/setiathome.berkeley.edu/libcudart.so.2 not found
Do 16 Apr 2009 00:08:26 CEST||file projects/setiathome.berkeley.edu/libcufft.so.2 not found
Do 16 Apr 2009 00:08:26 CEST||[error] No URL for file transfer of libcudart.so.2
Do 16 Apr 2009 00:08:26 CEST||[error] No URL for file transfer of libcufft.so.2

[sunu]

Unfortunately the 32bit CUDA linux client is bad. Use the 64bit app if you can.

[icy-flame]

I have a core i7 920 + 9400GT 512MB, but it took about 30min (wall clock) to do a WU with CUDA.

Given the time it took to finish and one of the CPU cores were running at full load, I suspect it is working in fall back mode, but I do not see any error message in boinc.log.

Is there anything or anywhere I should check to find out where the problems is?

Code: [Select]

# ldd setiathome-CUDA-6.08.x86_64-pc-linux-gnu
        libcufft.so.2 => /usr/lib64/libcufft.so.2 (0x00002b1499d87000)
        libcudart.so.2 => /usr/lib64/libcudart.so.2 (0x00002b149a0a1000)
        libcuda.so.1 => /usr/lib64/libcuda.so.1 (0x00002b149a2df000)
        libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00000031b0e00000)
        libm.so.6 => /lib64/libm.so.6 (0x00000031ac200000)
        libpthread.so.0 => /lib64/libpthread.so.0 (0x00000031aca00000)
        libc.so.6 => /lib64/libc.so.6 (0x00000031abe00000)
        libdl.so.2 => /lib64/libdl.so.2 (0x00000031ac600000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00000031b0a00000)
        librt.so.1 => /lib64/librt.so.1 (0x00000031b0600000)
        libz.so.1 => /usr/lib64/libz.so.1 (0x00000031ace00000)
        /lib64/ld-linux-x86-64.so.2 (0x00000031aae00000)


Code: [Select]

14:52 [---] Starting BOINC client version 6.6.20 for x86_64-pc-linux-gnu
10-May-2009 13:54:52 [---] This a development version of BOINC and may not function properly
10-May-2009 13:54:52 [---] log flags: task, file_xfer, sched_ops
10-May-2009 13:54:52 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8g zlib/1.2.3 c-ares/1.5.1
10-May-2009 13:54:52 [---] Data directory: /usr/lib/BOINC
10-May-2009 13:54:52 [SETI@home] Found app_info.xml; using anonymous platform
10-May-2009 13:54:52 [---] Processor: 8 GenuineIntel Intel(R) Core(TM) i7 CPU 920  @ 2.67GHz [Family 6 Model 26 Stepping 4]
10-May-2009 13:54:52 [---] Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall nx rdtscp lm constant_tsc ida nonstop_tsc pni monitor ds_cpl vmx est tm2 cx16 xtpr popcnt lahf_lm
10-May-2009 13:54:52 [---] OS: Linux: 2.6.18-128.1.10.el5
10-May-2009 13:54:52 [---] Memory: 11.72 GB physical, 13.69 GB virtual
10-May-2009 13:54:52 [---] Disk: 888.98 GB total, 836.73 GB free
10-May-2009 13:54:52 [---] Local time is UTC +1 hours
10-May-2009 13:54:52 [---] CUDA device: GeForce 9400 GT (driver version 0, CUDA version 1.1, 511MB, est. 16GFLOPS)
10-May-2009 13:54:52 [---] Not using a proxy
10-May-2009 13:54:52 [SETI@home] URL: http://setiathome.berkeley.edu/; Computer ID: 4917711; location: (none); project prefs: default
10-May-2009 13:54:52 [SETI@home] General prefs: from SETI@home (last modified 02-Mar-2009 19:24:50)
10-May-2009 13:54:52 [SETI@home] Host location: none
10-May-2009 13:54:52 [SETI@home] General prefs: using your defaults
10-May-2009 13:54:52 [---] Preferences limit memory usage when active to 6001.50MB
10-May-2009 13:54:52 [---] Preferences limit memory usage when idle to 10802.69MB
10-May-2009 13:54:52 [---] Preferences limit disk usage to 0.50GB
10-May-2009 14:04:31 [SETI@home] Starting 27fe09ab.28487.20931.9.8.139_1
10-May-2009 14:31:47 [SETI@home] Computation for task 27fe09ab.28487.20931.9.8.139_1 finished

[Jason G]

I have a core i7 920 + 9400GT 512MB, but it took about 30min (wall clock) to do a WU with CUDA.

Given the time it took to finish and one of the CPU cores were running at full load, I suspect it is working in fall back mode, but I do not see any error message in boinc.log.

My 20% OC'd 9600GSO averages about 18 minutes, please compare specs against these two cards, but your time sounds about right to me  .  That's still faster than CPU with AKv8 , If I recall correctly.

[icy-flame]

I guess my GPU just not powerful enough then. What still doesn't look right is I have one WU been going for 3 hours, only 90% progress.

I have an identical box (hardware and OS), running AK_V8 with boinc 6.4.5, typical WU finish in 60 minutes wall clock. But since HT is enabled (i.e. eight WU running in parallel on four physical cores), the actual CPU time is closer to 30 min each.

[Jason G]

Yeah,  I had some 8600GT's when all this Ciuda kerfuffle started, and that really brought everything to its knees, really disappointing even though I got the cards for nothing, from a friend who was throwing them out.

Because I got my 9600 GSO's really cheap (i.e. from Santa  ), I thought that I'd try to squeeze what I could out of them.  It seems while they do fairly ordinary at stock clock, they seem pretty good cranked up a bit.  You might consider experimenting with 'ATITool' to determine best OC without artefacts, and see how 'RthDribl' runs in large config with Boinc suspended. 

If the performance then when OC'd doesn't satisfy you, I'd say we're both hitting the limits of the wallet rather than silicon 

[Edit: Having said all that, driver crashes are still possible, instigating fall-back mode, but I would've thought that'd be a Windows thing for some reason.. Being a Windows man of course  ]

[icy-flame]

3 hrs per WU on a 9400GT, that is just disappointing

AK_V8 on my CPU can do better than that.

[sunu]

Maybe those wus were VLAR?

[Gecko_R7]

Maybe those wus were VLAR?

Exactly.  VLAR on 9400GT would certainly be disappointing.

Also keep in mind that a stock 9400GT might disappoint overall if being compared to production from a high-end CPU, or even a mid-level one w/ lots of L2 cache and a modest over-clock.  Those with Celeron's, or (older Core2 hosts w/ lower clock speeds) and small cache may actually see see 9400GT as quicker than a CPU core.

IMO, comparing 94,95 & 9600 performance to CPU is very relative to the CPU being compared, especially using "reference speed" GPU cards.

[dtiger]

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....

[sunu]

dtiger you can try the latest boinc 6.6.29. It has the option to use the GPU only when the pc is not used. I haven't used it so I don't know if it works.

Also with cuda 2.2 the linux client now uses a fraction of a cpu core, like the windows client does, and not 100% of a core (not that it'll make X faster, but it's a step forward for us, linux users).

[s52d]

Hi!

GTX 260 on Q6600 machine, Slackware. NVIDIA-Linux-x86_64-185.18.04 driver.
Tried 32 bit, worked fine. (beside returning errors, but this is driver problem, not my machine)


It is now Slackware64, 64bit. Used to work fine in text mode, but KDE kills it. Most WU takes a lot of time, in error we can see:

http://setiweb.ssl.berkeley.edu/result.php?resultid=1248087935

Cuda error 'cudaMalloc((void**) &dev_WorkData' in file './cudaAcceleration.cu' in line 293 : out of memory.
setiathome_CUDA: CUDA runtime ERROR in device memory allocation (Step 1 of 3). Falling back to HOST CPU processing...

Any hint what to do?  XFCE instead of KDD? Going out of KDE while PC is not used?
A bit of searching...

Aha, 16 bit color depth works, 24 does not.
    DefaultDepth    16
in xorg.conf, and seems to work fine.


A bit later: again, memory... Aaargh.


BR
Iztok

[sunu]

Try the latest driver 185.18.14. It's now the official one.

What version of cuda libraries do you use? 2.1? 2.2?

You have many errors here. Can you describe under what conditions did they occur?

[s52d]

Hi!

Main news: It is probably not DefaultDepth  but restarting boinc/X11 and freeing up memory ;-)

Errors: PC got mad and I lost number of WUs due to "computing error" during testing.
older errors from 32bit adventure.

Anyhow, I found occasionally few copies of setiathome-CUDA-6.08.x86_64-pc-linux-gnu in the memory (preferences: no memory).
Quick-and-Dirty solution: crontab+shell script to run ps |wc , if there are 3 or more copies, restart whole boinc.
triggered once yesterday, now I have 24 hours of uptime.

Slackware-current64,  2.6.29.4 kernel customized to PC, boinc 6.6.31.
libs as in driver/client packages.
CUDA device: GeForce GTX 260 (driver version 0, compute capability 1.3, 895MB, est. 104GFLOPS)

from dmesg:
NVRM: loading NVIDIA UNIX x86_64 Kernel Module  185.18.04  Thu Apr 16 21:41:04 PDT 2009
IRQ 16/nvidia: IRQF_DISABLED is not guaranteed on shared IRQs
NVRM: Xid (0001:00): 13, 0008 00000000 00005039 00000180 0000046c 00000008


Thanks for pointing out new 185.18.14 .... Will try on next reboot of X11.

BR
Iztok