Right, I seem to have resolved this. The issue was on my end, starting
with a nonexistent directory in .bashrc (on both computers) being thrown
into the compilation. That became obvious with -v, it seems removing
that line in .bashrc makes everything work.
Output is attached anyway. Thanks a lot!
On 11-10-11 12:17 PM, Andreas Kloeckner wrote:
On Tue, 11 Oct 2011 10:08:04 +0000, Irwin
Non-text part: multipart/mixed
I've attached siteconf.py and the output of
make. The first command that causes issues is:
gcc -pthread -fno-strict-aliasing -fwrapv -Wall -g -fPIC -DPYGPU_PACKAGE=pycuda
-DHAVE_CURAND=1 -DBOOST_PYTHON_SOURCE=1 -DPYGPU_PYCUDA=1
-DBOOST_MULTI_INDEX_DISABLE_SERIALIZATION=1 -Dboost=pycudaboost -Isrc/cpp
-I/usr/local/lib/python2.6/dist-packages/numpy/core/include -I/usr/include/python2.6 -c
src/cpp/cuda.cpp -o build/temp.linux-x86_64-2.6/src/cpp/cuda.o
All of that looks just fine. Can you rerun this command in the PyCUDA
directory with -v added? This is to figure out how the hell /usr/bin
ends up on the include path.