Here's another follow up on the hardware and a downgrade to version 0.93.
Even with 0.93, I still get the same missing symbol error.
Below is my hardware config through the driver querying program which comes
with the CUDA SDK.
Device 0: "GeForce GTX 260"
CUDA Driver Version: 3.0
CUDA Runtime Version: 3.0
CUDA Capability Major revision number: 1
CUDA Capability Minor revision number: 3
Total amount of global memory: 938803200 bytes
Number of multiprocessors: 27
Number of cores: 216
Total amount of constant memory: 65536 bytes
Total amount of shared memory per block: 16384 bytes
Total number of registers available per block: 16384
Warp size: 32
Maximum number of threads per block: 512
Maximum sizes of each dimension of a block: 512 x 512 x 64
Maximum sizes of each dimension of a grid: 65535 x 65535 x 1
Maximum memory pitch: 2147483647 bytes
Texture alignment: 256 bytes
Clock rate: 1.30 GHz
Concurrent copy and execution: Yes
Run time limit on kernels: No
Integrated: No
Support host page-locked memory mapping: Yes
Compute mode: Default (multiple host
threads can use this device simultaneously)
On Tue, Mar 23, 2010 at 3:15 PM, Paul Rigor (uci) <paul.ri...@uci.edu>wrote:
> Another follow up,
>
> Since I am using the latest drivers and toolkit from NVidia (195 and 3.x,
> respectively), I also attempted to compile pycuda against the emulation
> driver. Still, I get the very same missing symbol error. I've been
> meticulous with the LD_LIBRARY_PATH as well and can verify that the
> dependencies are accessible.
>
> libboost_python.so.1.42.0 => /usr/local/lib/libboost_python.so.1.42.0
> (0x00007faeb6ec9000)
> libboost_thread.so.1.42.0 => /usr/local/lib/libboost_thread.so.1.42.0
> (0x00007faeb6cb4000)
> libcudartemu.so.3 => /usr/local/cuda/lib64/libcudartemu.so.3
> (0x00007faeb6a7d000)
>
>
> Note that although the boost libraries are not installed under lib64, they
> are in fact 64-bit:
>
>
> ldd /usr/local/lib/libboost_python.so.1.42.0
> linux-vdso.so.1 => (0x00007fff05dff000)
> libutil.so.1 => /lib64/libutil.so.1 (0x00007f1da1cb7000)
> libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f1da1a9a000)
> libdl.so.2 => /lib64/libdl.so.2 (0x00007f1da1896000)
> librt.so.1 => /lib64/librt.so.1 (0x00007f1da168e000)
> libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00007f1da1383000)
> libm.so.6 => /lib64/libm.so.6 (0x00007f1da1100000)
> libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f1da0ee6000)
> libc.so.6 => /lib64/libc.so.6 (0x00007f1da0b77000)
> /lib64/ld-linux-x86-64.so.2 (0x00000036fd400000)
>
> Thanks again,
> Paul
>
> On Tue, Mar 23, 2010 at 2:50 PM, Paul Rigor (uci) <paul.ri...@uci.edu>wrote:
>
>> As a follow up,
>>
>> My system is a fedora 11 with the latest kernel
>> (2.6.30.10-105.2.23.fc11.x86_64), nvidia drivers and boost library (1.42).
>> I've double checked that the library dependencies are accessible and
>> present. I've also listed the symbol table in libcudart.so and the symbol
>> ('cuTexRefSetAddress') reported missing does indeed exists.
>>
>> The following lists what the pycuda driver wrapper is linked against:
>> linux-vdso.so.1 => (0x00007fffb39ff000)
>> libboost_python.so.1.42.0 => /usr/local/lib/libboost_python.so.1.42.0
>> (0x00007f8bc42f6000)
>> libboost_thread.so.1.42.0 => /usr/local/lib/libboost_thread.so.1.42.0
>> (0x00007f8bc40e1000)
>> libcudart.so.3 => /usr/local/cuda/lib64/libcudart.so.3
>> (0x00007f8bc3e86000)
>> libpython2.6.so.1.0 => /usr/lib64/libpython2.6.so.1.0
>> (0x00007f8bc3ad9000)
>> libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00007f8bc37ce000)
>> libm.so.6 => /lib64/libm.so.6 (0x00007f8bc354b000)
>> libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f8bc3331000)
>> libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f8bc3114000)
>> libc.so.6 => /lib64/libc.so.6 (0x00007f8bc2da6000)
>> libutil.so.1 => /lib64/libutil.so.1 (0x00007f8bc2ba3000)
>> libdl.so.2 => /lib64/libdl.so.2 (0x00007f8bc299e000)
>> librt.so.1 => /lib64/librt.so.1 (0x00007f8bc2796000)
>> /lib64/ld-linux-x86-64.so.2 (0x00000036fd400000)
>>
>>
>> Thanks,
>> Paul
>>
>> On Tue, Mar 23, 2010 at 2:29 PM, Paul Rigor (uci) <paul.ri...@uci.edu>wrote:
>>
>>> Hi gang,
>>>
>>>
>>> I have both CUDA and PyCUDA related questions.
>>>
>>> For CUDA, I'm using the latest drivers (195.x.x) and latest toolkit/sdk
>>> (3.0). My problem is compiling the set of test programs. I remember in the
>>> past that a hack had to be performed -- ie, comment out some function
>>> declarations in cuda's math_functions.h. I've received the following error:
>>>
>>> /usr/local/cuda/bin/../include/math_functions.h:422: error: inline
>>> function ‘int __signbit(double)’ cannot be declared weak
>>> /usr/local/cuda/bin/../include/math_functions.h:427: error: inline
>>> function ‘int __signbitf(float)’ cannot be declared weak
>>> /usr/local/cuda/bin/../include/math_functions.h:440: error: inline
>>> function ‘int __signbitl(long double)’ cannot be declared weak
>>>
>>> The work around is to comment these lines out -- but is this still the
>>> best way to go about it?
>>>
>>> For PyCUDA, I've downloaded the latest beta (0.94.xxx). The compilation
>>> of the wrapper for the driver goes smoothly. However, the test fails with
>>> the following undefined symbol error. Even with the 190.xxx.xxx drivers and
>>> the fix I mentioned above, I still get this error =\ Any one encountering
>>> similar issues?
>>>
>>> Traceback (most recent call last):
>>> File "test/test_driver.py", line 4, in <module>
>>> from pycuda.tools import mark_cuda_test
>>> File "/usr/lib64/python2.6/site-packages/pycuda/tools.py", line 30, in
>>> <module>
>>> import pycuda.driver as cuda
>>> File "/usr/lib64/python2.6/site-packages/pycuda/driver.py", line 1, in
>>> <module>
>>> from _driver import *
>>> ImportError: /usr/lib64/python2.6/site-packages/pycuda/_driver.so:
>>> undefined symbol: cuTexRefSetAddress
>>>
>>>
>>>
>>>
>>> --
>>> Paul Rigor
>>> Pre-doctoral BIT Fellow and Graduate Student
>>> Institute for Genomics and Bioinformatics
>>> Donald Bren School of Information and Computer Sciences
>>> University of California, Irvine
>>> http://www.ics.uci.edu/~prigor
>>>
>>
>>
>>
>> --
>> Paul Rigor
>> Pre-doctoral BIT Fellow and Graduate Student
>> Institute for Genomics and Bioinformatics
>> Donald Bren School of Information and Computer Sciences
>> University of California, Irvine
>> http://www.ics.uci.edu/~prigor
>>
>
>
>
> --
> Paul Rigor
> Pre-doctoral BIT Fellow and Graduate Student
> Institute for Genomics and Bioinformatics
> Donald Bren School of Information and Computer Sciences
> University of California, Irvine
> http://www.ics.uci.edu/~prigor
>
--
Paul Rigor
Pre-doctoral BIT Fellow and Graduate Student
Institute for Genomics and Bioinformatics
Donald Bren School of Information and Computer Sciences
University of California, Irvine
http://www.ics.uci.edu/~prigor
_______________________________________________
PyCUDA mailing list
pyc...@host304.hostmonster.com
http://host304.hostmonster.com/mailman/listinfo/pycuda_tiker.net
