Instant error while computing with new Geforce GTX 680
10 May 2012 14:58:54 UTC
Topic 196320
(moderation:
I just installed it and are using driver version 301.24. Everytime it starts a BRP4cuda work unit it instantly jumps to 100% and gives the "error while computing" message. Is the GTX 680 incompatible with einstein@home or is it a driver problem?
7.0.25
- exit code -1073741515 (0xc0000135)
]]>
Instant error while computing with new Geforce GTX 680
)
Just tried the 301.10 driver too and reinstalled boinc. Same issue. Anyone who can help me with this?
CPUgrid also quits/crashes straight away when trying to run:
7.0.25
Den angivne sti blev ikke fundet. (0x3) - exit code 3 (0x3)
# Using device 0
# There is 1 device supporting CUDA
# Device 0: "GeForce GTX 680"
# Clock rate: 0.71 GHz
# Total amount of global memory: -2147483648 bytes
# Number of multiprocessors: 8
# Number of cores: 64
SWAN : Module load result [.fastfill.cu.] [200]
SWAN: FATAL : Module load failed
Assertion failed: 0, file swanlib_nv.c, line 390
This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
]]>
GPUgrid does not accept 6xx
)
GPUgrid does not accept 6xx series yet. As for here they work, not sure why yours isn't. Did you restart after updating driver?
Hi! Thanks for reporting
)
Hi!
Thanks for reporting this.
As 5pot will be able to confirm, there is nothing in principle that will prevent the 680 from running BRP4 jobs at E@H, his host is doing rather well under Win 7:
http://einsteinathome.org/task/287715923.
5pot, what driver are you using atm?
Cheers
HB
Hi, check if the
)
Hi,
check if the application is really executable. I had the problem under Linux. Everytime the 7.0.25 client downloaded a new application it would be a normal file instead of an executable. I seem to recall that there was at least one person who had the same problem on Windows as well. I downgraded to BOINC 6.10.58 - that fixed the problem.
mickydl
RE: Hi, check if the
)
7.0.25 already fixed this particular issue, but if the executable was downloaded with an earlier, defective BOINC version, just upgrading to 7.0.25 does not correct the problem with existing downloaded (non)executables, you would need to reset the project.
But I doubt this is the problem here, the error message would be different and at least the GPUgrid app is clearly executing, but then terminating.
Cheers
HB
Thanks for the replies. The
)
Thanks for the replies. The PC is a new build (i7 3770k, Asus Sabertooth Z77, Asus GTX 680, Intel 520 SSD, Enermax platimax 750W PSU - no overclocking yet)
Boinc was installed as 7.0.25. Because the GTX 680 was delayed, I had an ATI R770 card in it, but I was careful to completely uninstall the ATI driver before replacing it with the GTX 680 (yes I've rebooted several times and reset the project.)
3D games and SETI cuda works fine. The CPUgrid problem looks to be caused by incompatibility with Kepler GPUs.
This is the log for a SETI WU:
core_client_version>7.0.25
setiathome_CUDA: Found 1 CUDA device(s):
Device 1 : GeForce GTX 680
totalGlobalMem = -2147483648
sharedMemPerBlock = 49152
regsPerBlock = 65536
warpSize = 32
memPitch = 2147483647
maxThreadsPerBlock = 1024
clockRate = 705500
totalConstMem = 65536
major = 3
minor = 0
textureAlignment = 512
deviceOverlap = 1
multiProcessorCount = 8
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 680 is okay
SETI@home using CUDA accelerated device GeForce GTX 680
setiathome_enhanced 6.09 Visual Studio/Microsoft C++
libboinc: 6.3.22
Work Unit Info:
...............
WU true angle range is : 0.435162
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_GetPowerSpectrum 0.00018 0.00000
v_ChirpData 0.01235 0.00000
v_Transpose4 0.00615 0.00000
FPU opt folding 0.00336 0.00000
Flopcounter: 43136210480884.266000
Spike count: 3
Pulse count: 0
Triplet count: 0
Gaussian count: 1
called boinc_finish
]]>
RE: As 5pot will be able to
)
It looks like 5pot is using beta driver 301.25 and I just tried using that too. Still doesn't work :(
There is now a 301.34 to play
)
There is now a 301.34 to play with, see this post.
http://einsteinathome.org/node/196234&nowrap=true#117109
301.34 I believe is the same
)
301.34 I believe is the same driver, just with 670 and 690 "recognition code" built in. I THINK.
I currently think 301.25 is the best. Personally (for 680)
2 things you may want to try.
1) Run Driver Sweeper in Safe Mode, and get rid of ALL DRIVERS (BOTH ATI AND NVIDIA). I know you said you un-installed ATI's, but, I've read quite a few times that they can still "linger" around. After everything is uninstalled for sure, re-download 301.25 and go from there. Sometimes you need to start from scratch, especially when switching between ATI and NVIDIA.
2) If you're running a screen saver, I have also read that this can interact with running tasks (GPUgrid forums). There are people there who will have tasks fail because screen saver is running. EVEN IF SCREEN IS TURNED OFF
EDIT: What is interesting though, is that someone with your same processor reported this error, and eventually was able to get his to work. Unfortunately, he never responded back as to how he fixed it. (correlation not causation) He had 560Ti though.
This error pops up at Einstein, GPUgrid, and evem WCG's beta GPU thread. You may want to try checking your NVIDIA Panel settings, or Power Management settings and make sure everything is set to max performance type settings, if this is acceptable in your case.
This specific error is hard to track down. Hope this info helps. Good luck
Thought of this too. As a
)
Thought of this too. As a LAST RESORT!!!!! I see there have been two motherboard updates for your BIOS. Again, LAST RESORT!!!, you may wish to d/l the latest and give it a try.