All GPU missing after second GTX-460 added
13 Oct 2012 16:45:56 UTC
Topic 196566
(moderation:
Hi hope someone can point me in the right direction
I have added a second GTX460 as that seemed the easiest way to get more WUs crunched without a lot of investment, I have capacity on the power supply - 650w.
No SLI used (or needed).
The nvidia xserver settings tool shows the new card to be working, and the original card is working ok.
The only difference between the cards is the PCIe Link Width shows x4 for the new card, and x16 for the original. This is expected as the motherboard runs that slot at x4.
win7 and ubuntu seem to run ok.
http://einsteinathome.org/host/4918234 is the host.
and this is the startup messages.
Sat 13 Oct 2012 15:25:10 BST Starting BOINC client version 6.10.17 for x86_64-pc-linux-gnu
Sat 13 Oct 2012 15:25:10 BST log flags: file_xfer, sched_ops, task
Sat 13 Oct 2012 15:25:10 BST Libraries: libcurl/7.19.7 OpenSSL/0.9.8k zlib/1.2.3.3 libidn/1.15
Sat 13 Oct 2012 15:25:10 BST Data directory: /var/lib/boinc-client
Sat 13 Oct 2012 15:25:10 BST Processor: 4 GenuineIntel Intel(R) Core(TM) i3 CPU 530 @ 2.93GHz [Family 6 Model 37 Stepping 2]
Sat 13 Oct 2012 15:25:10 BST Processor: 4.00 MB cache
Sat 13 Oct 2012 15:25:10 BST 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 pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni dtes64 monito
Sat 13 Oct 2012 15:25:10 BST OS: Linux: 2.6.32-44-generic
Sat 13 Oct 2012 15:25:10 BST Memory: 3.74 GB physical, 22.10 GB virtual
Sat 13 Oct 2012 15:25:10 BST Disk: 45.85 GB total, 2.80 GB free
Sat 13 Oct 2012 15:25:10 BST Local time is UTC +1 hours
Sat 13 Oct 2012 15:25:11 BST No usable GPUs found
Sat 13 Oct 2012 15:25:11 BST PrimeGrid Application uses missing NVIDIA GPU
Sat 13 Oct 2012 15:25:11 BST Einstein@Home Application uses missing NVIDIA GPU
Sat 13 Oct 2012 15:25:11 BST Einstein@Home Missing coprocessor for task p2030.20111001.G195.96-01.74.C.b2s0g0.00000_2344_1
Sat 13 Oct 2012 15:25:11 BST Einstein@Home Missing coprocessor for task p2030.20111001.G195.84-01.97.C.b6s0g0.00000_2112_1
If I find a solution I will post back but I´m a bit lost what to try next.
Cheers
Mike
All GPU missing after second GTX-460 added
)
Well I think it is fixed, or at least for the moment.
Sat 13 Oct 2012 20:16:05 BST NVIDIA GPU 0: GeForce GTX 460 (driver version unknown, CUDA version 5000, compute capability 2.1, 768MB, 163 GFLOPS peak)
Sat 13 Oct 2012 20:16:05 BST NVIDIA GPU 1: GeForce GTX 460 (driver version unknown, CUDA version 5000, compute capability 2.1, 768MB, 163 GFLOPS peak)
and now i have two cuda racing each other to complete a WU each.
I´m not sure yet exactly what the fix is but testing the second card with win7 with a monitor connected, and then added the
1
to the cc_config.xml file -and a restart back to linux now sees two GTX460 searching away.
It will be interesting to see if the PCI Link width x4 is slower than the x16.
Has something happened to the
)
Has something happened to the driver location/access rights when installing the second GPU?
BOINC needs (at least read) access to the drivers in order to recognise the GPUs.
Gruß,
Gundolf
[edit]You posted the second message while I was composing mine ;-)[/edit]
Computer sind nicht alles im Leben. (Kleiner Scherz)
I guess it was possible some
)
I guess it was possible some change with the drivers however looking back over the stdoutdae.txt
the two GPUs were visible just once immediately after the install and it looked like they actually started working.
13-Oct-2012 15:17:28 [---] Local time is UTC +1 hours
13-Oct-2012 15:17:29 [---] NVIDIA GPU 0: GeForce GTX 460 (driver version unknown, CUDA version 5000, compute capability 2.1, 768MB, 163 GFLOPS peak)
13-Oct-2012 15:17:29 [---] NVIDIA GPU 1: GeForce GTX 460 (driver version unknown, CUDA version 5000, compute capability 2.1, 768MB, 163 GFLOPS peak)
13-Oct-2012 15:17:29 [---] Not using a proxy
13-Oct-2012 15:17:29 [Cosmology@Home] URL http://www.cosmologyathome.org/; Computer ID 156022; resource share 100
13-Oct-2012 15:17:29 [LHC@home 1.0] URL http://lhcathomeclassic.cern.ch/sixtrack/; Computer ID 9958753; resource share 100
13-Oct-2012 15:17:29 [PrimeGrid] URL http://www.primegrid.com/; Computer ID 247238; resource share 20
13-Oct-2012 15:17:29 [Einstein@Home] URL http://einstein.phys.uwm.edu/; Computer ID 4918234; resource share 100
13-Oct-2012 15:17:29 [Einstein@Home] General prefs: from Einstein@Home (last modified 07-Oct-2012 01:08:03)
13-Oct-2012 15:17:29 [Einstein@Home] Host location: none
13-Oct-2012 15:17:29 [Einstein@Home] General prefs: using your defaults
13-Oct-2012 15:17:29 [---] Reading preferences override file
13-Oct-2012 15:17:29 [---] Preferences limit memory usage when active to 1916.27MB
13-Oct-2012 15:17:29 [---] Preferences limit memory usage when idle to 3449.28MB
13-Oct-2012 15:17:29 [---] Preferences limit disk usage to 5.27GB
BOINC initialization completed, beginning process execution...
13-Oct-2012 15:17:30 [Einstein@Home] Restarting task p2030.20110928.G177.21-04.93.C.b6s0g0.00000_400_1 using einsteinbinary_BRP4 version 129
13-Oct-2012 15:17:31 [LHC@home 1.0] Restarting task wlxscan0_w5cbb__36__s__64.31_59.32__0_0.5__6__15_1_sixvf_boinc49571_0 using sixtrack version 44401
13-Oct-2012 15:17:31 [LHC@home 1.0] Restarting task wlxscan0_w5cbb__35__s__64.31_59.32__11_11.5__6__88.5_1_sixvf_boinc49502_1 using sixtrack version 44401
13-Oct-2012 15:17:31 [LHC@home 1.0] Restarting task wlxscan0_w5cbb__35__s__64.31_59.32__11_11.5__6__82.5_1_sixvf_boinc49498_0 using sixtrack version 44401
13-Oct-2012 15:17:33 [Einstein@Home] Restarting task p2030.20111001.G194.18-04.94.C.b0s0g0.00000_80_0 using einsteinbinary_BRP4 version 128
13-Oct-2012 15:17:33 [Einstein@Home] Starting p2030.20111001.G194.18-04.94.C.b0s0g0.00000_56_1
13-Oct-2012 15:17:34 [Einstein@Home] Starting task p2030.20111001.G194.18-04.94.C.b0s0g0.00000_56_1 using einsteinbinary_BRP4 version 128
13-Oct-2012 15:17:34 [Einstein@Home] Sending scheduler request: To fetch work.
13-Oct-2012 15:17:34 [Einstein@Home] Reporting 12 completed tasks, requesting new tasks for GPU
13-Oct-2012 15:17:40 [Einstein@Home] Scheduler request failed: Couldn't resolve host name
I had not connected the Ethernet at that point, the case was open etc - so just shut it down gracefully, closed up the case, connected up the Ethernet.
Occasionally after a kernel update, the drivers need updating/re-installing - but the nvidia console also shows no GPU, so still none the wiser.
Anyways thanks for having a read, early indications are the second GPU (pcie-x4) is eating a WU in about 90 minutes, compared with 65 minutes for the first GPU (x16) and 1000 minutes for the i3 - but of course the i3 could be doing three WUs in parallel.
Cheers
Mike
I dont know how it works
)
I dont know how it works under Linux, but I guess that the first time the system boots after installing a GPU the drivers will be "reconfigured" to refect that there are a new GPU and while the system is doing this it could lead BOINC to fail accessing the GPUs...
Its ussually recomended to set BOINC to not start automatically on system restart when you are going to make changes in the hardware, to avoid this glitches during drivers installation/reconfiguration...
By the way if both GPUs are the same model, you should not need the 1 option in the cc_cofig (anyway it doesn't do any harm to keep it there)