Help !!!! Seti & 9800GT

CGS

CGS

CGS

Associate
Joined
26 Nov 2003
Posts
146
Location
Garden of England
Just got hold of a new POV 9800GT 512MB green edition to stick in a Dell Dimension at work
Have fitted it ok loaded latest drivers but just cant get it to run cuda without computation errors :(
Using latest optimised settings and PC has worked fine with old 8500gt for a while now (very slow but no errors).
Have tried clocking card always (core shader linked unlinked mem etc) using rivertune at driver level with no joy.
Temp isnt a problem as has not exceeded 60Deg C
Would be grateful if anyone can help or maybe i just got a duff GPU :(
 
This damn upgrade fever is very catching :D

What is the boinc messages tab saying?
On the seti site you should be able to see exactly what the error is? Have a look at your results for the machine in question and filter by errors
Have you tried using the latest lunatic's release?
It might be that you are using CUDA3 rather than CUDA 2.3 if you are using the latest drivers..
Im not really sure what is going on to be honest as I haven't run the stock app for so long now..
Stop boinc
run the lunatics installer
select to use cuda 2.3
and see how you go..

Im guessing by the way - but that is what I would do as a first few steps..
 
edgedemon said:
This damn upgrade fever is very catching :D

What is the boinc messages tab saying?
On the seti site you should be able to see exactly what the error is? Have a look at your results for the machine in question and filter by errors
Have you tried using the latest lunatic's release?
It might be that you are using CUDA3 rather than CUDA 2.3 if you are using the latest drivers..
Im not really sure what is going on to be honest as I haven't run the stock app for so long now..
Stop boinc
run the lunatics installer
select to use cuda 2.3
and see how you go..

Im guessing by the way - but that is what I would do as a first few steps..
Click to expand...

Already using lunatics and tried both cuda settings
Cant check results as seti still down and run out of cuda wu's now even rescheduled vlars etc only seeing computation errors through boinc manager and watching wu count to upto 2% and die :(
At a bit of a loss atm :(
 
I had this problem with the green 9800 that I got- halz had some suggestions that should get it to work- its got to do with clocks, but I can't remember them......
 
I would suspend boinc so you don't trash any more workunits
Then in Boinc Manager projects tab select seti
On the left select 'Your Account'
On the web page that opens select Account from the top
Scroll down and selct view next to 'Computers on this account'
Find your PC that is producing the errors
Select tasks and then filter by 'error' and try and find some that were uploaded today
If you click on the task number it will give you the details about the task and under
Stderr output

there will be some text which hopefully will give you a clue as to why this is happening.
Post back here and hopefully area51 or another person more knowledgeable then me will be able to help, or you can post on the seti forums with the error message as there are shedloads of experts there...
 
edgedemon said:
I would suspend boinc so you don't trash any more workunits
Then in Boinc Manager projects tab select seti
On the left select 'Your Account'
On the web page that opens select Account from the top
Scroll down and selct view next to 'Computers on this account'
Find your PC that is producing the errors
Select tasks and then filter by 'error' and try and find some that were uploaded today
If you click on the task number it will give you the details about the task and under
Stderr output

there will be some text which hopefully will give you a clue as to why this is happening.
Post back here and hopefully area51 or another person more knowledgeable then me will be able to help, or you can post on the seti forums with the error message as there are shedloads of experts there...
Click to expand...

Like I said earlier none have uploaded yet will post one here when they do
 
This is one of the error reports as they have finally uploaded.
Dont me much to me hopefully does to one of you guys

Stderr output

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
setiathome_CUDA: Found 1 CUDA device(s):
Device 1 : GeForce 9800 GT
totalGlobalMem = 536543232
sharedMemPerBlock = 16384
regsPerBlock = 8192
warpSize = 32
memPitch = 2147483647
maxThreadsPerBlock = 512
clockRate = 1375000
totalConstMem = 65536
major = 1
minor = 1
textureAlignment = 256
deviceOverlap = 1
multiProcessorCount = 14
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce 9800 GT is okay
SETI@home using CUDA accelerated device GeForce 9800 GT
V12 modification by Raistmer
Priority of worker thread rised successfully
Priority of process adjusted successfully
Total GPU memory 536543232 free GPU memory 498860032
setiathome_enhanced 6.02 Visual Studio/Microsoft C++

Build features: Non-graphics CUDA VLAR autokill enabled FFTW USE_SSE x86
CPUID: Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz

Cache: L1=64K L2=2048K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
libboinc: 6.3.22

Work Unit Info:
...............
WU true angle range is : 0.443717
After app init: total GPU memory 536543232 free GPU memory 498860032
Cuda error 'cufftExecC2C' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_fft.cu' in line 143 : unspecified launch failure.
Cuda error 'cudaAcc_GetPowerSpectrum_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_PowerSpectrum.cu' in line 56 : unspecified launch failure.
Cuda error 'cudaAcc_GetPowerSpectrum_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_PowerSpectrum.cu' in line 56 : unspecified launch failure.
Cuda error 'cudaAcc_summax32_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_summax.cu' in line 147 : unspecified launch failure.
Cuda error 'cudaAcc_summax32_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_summax.cu' in line 147 : unspecified launch failure.
Cuda error 'cudaMemcpy(PowerSpectrumSumMax, dev_PowerSpectrumSumMax, cudaAcc_NumDataPoints / fftlen * sizeof(*dev_PowerSpectrumSumMax), cudaMemcpyDeviceToHost)' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_summax.cu' in line 160 : unspecified launch failure.

</stderr_txt>
 
:eek::eek::eek::eek:

The only time I have seen anything like that was on a faulty GX2 where 1 core was broken. It wasn't the same messages, but similar. Each time a wu would load it would get to 12 secs and error instantly..

Maybe one for the seti forums?
 
not tried it on stock yet but will do next week, will also stick it in different desktop and try then.
It did play intensive game fine with no crashes so hopefully not a dodgy core
 
Let us know the outcome.

Always best to start at stock settings (hardware & software) - just to make sure everything is working as required and also to give you a performance baselilne to judge how the Opt Apps and Overclocks are performing.

:D
 
CGS said:
This is one of the error reports as they have finally uploaded.
Dont me much to me hopefully does to one of you guys

Stderr output

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
setiathome_CUDA: Found 1 CUDA device(s):
Device 1 : GeForce 9800 GT
totalGlobalMem = 536543232
sharedMemPerBlock = 16384
regsPerBlock = 8192
warpSize = 32
memPitch = 2147483647
maxThreadsPerBlock = 512
clockRate = 1375000
totalConstMem = 65536
major = 1
minor = 1
textureAlignment = 256
deviceOverlap = 1
multiProcessorCount = 14
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce 9800 GT is okay
SETI@home using CUDA accelerated device GeForce 9800 GT
V12 modification by Raistmer
Priority of worker thread rised successfully
Priority of process adjusted successfully
Total GPU memory 536543232 free GPU memory 498860032
setiathome_enhanced 6.02 Visual Studio/Microsoft C++

Build features: Non-graphics CUDA VLAR autokill enabled FFTW USE_SSE x86
CPUID: Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz

Cache: L1=64K L2=2048K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
libboinc: 6.3.22

Work Unit Info:
...............
WU true angle range is : 0.443717
After app init: total GPU memory 536543232 free GPU memory 498860032
Cuda error 'cufftExecC2C' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_fft.cu' in line 143 : unspecified launch failure.
Cuda error 'cudaAcc_GetPowerSpectrum_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_PowerSpectrum.cu' in line 56 : unspecified launch failure.
Cuda error 'cudaAcc_GetPowerSpectrum_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_PowerSpectrum.cu' in line 56 : unspecified launch failure.
Cuda error 'cudaAcc_summax32_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_summax.cu' in line 147 : unspecified launch failure.
Cuda error 'cudaAcc_summax32_kernel' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_summax.cu' in line 147 : unspecified launch failure.
Cuda error 'cudaMemcpy(PowerSpectrumSumMax, dev_PowerSpectrumSumMax, cudaAcc_NumDataPoints / fftlen * sizeof(*dev_PowerSpectrumSumMax), cudaMemcpyDeviceToHost)' in file 'd:/BoincSeti_Prog/sinbad_repositories/LunaticsUnited/SETI_CUDA_MB_exp/client/cuda/cudaAcc_summax.cu' in line 160 : unspecified launch failure.

</stderr_txt>
Click to expand...

Can you unhide your S@H machine .... I can't see the tasks and find out what the error code that was reported by the client was......

When installing the optimised apps, you really should start with stock, download a small amount of work, set no new tasks, process the work, upload it and THEN install the optimised stuff (when you have an empty cache). I'm not entirely sure you can go straight for optimised without having the stock running first.....
 
when i was having issues recently with my laptop gpu, it turned out to be the latest drivers, i had to role back to 197.16 and it worked fine again. if its the same issue as i had try loading gpu-z, and checking for a tick against CUDA. for some reason the tick went away with the newer drivers and seti didn't work, didn't stop it trying and erroring loads of WU till i spotted it.
 
Thanx for the pointers I will have a play at work next week :)
Think the drivers might be a good one to try as i did update drivers when changed from 8500gt to 9800gt.
As its my own work pc Im gonna move card to different PC and start again.
Gonna put ATI in My work one so will asking how to setup milkyway next :P.
Will unhide computers when i remember how :)
 
Have you tried connecting to another project with Nvidia apps like Einstein? Or F@H?
If those WU's error as well then it's time to start looking at the card or drivers rather than BOINC or Seti settings.

As others have suggested try the default app at stock settings first though.
 
have had another play
Changed nvidia drivers to lower version that was working fine with 8500gt
Uninstalled optimized lunatics
Tried again
GPU wu's now seem to run but don't complete (well one has)
Now get "Completed, validation inconclusive" on pending.
Is this good, bad getting there etc ? :)
Have halted gpu tasks for now

Is it possible to install lunatics optimised for cpu only leaving gpu on standard?
 
You must log in or register to reply here.
Back
Top Bottom