I downloaded pycuda-2013.1.1 and modified ./configure like :
siteconf.py:
BOOST_INC_DIR = [‘/usr/local/include/boost’]
BOOST_LIB_DIR = [‘/usr/local/lib’]
BOOST_COMPILER = ‘gcc46’
USE_SHIPPED_BOOST = False
BOOST_PYTHON_LIBNAME = [‘boost_python’]
BOOST_THREAD_LIBNAME = [‘boost_thread’]
CUDA_TRACE = False
CUDA_ROOT = ‘/usr/local/cuda-5.5/bin’
CUDA_ENABLE_GL = False
CUDA_ENABLE_CURAND = True
CUDADRV_LIB_DIR = [‘/usr/local/cuda-5.5/lib64’]
CUDADRV_LIBNAME = [‘cuda’]
CUDART_LIB_DIR = [‘/usr/local/cuda-5.5/lib’, ‘/usr/local/cuda-5.5/lib64’]
CUDART_LIBNAME = [‘cudart’]
CURAND_LIB_DIR = [‘/usr/local/cuda-5.5/lib’, ‘/usr/local/cuda-5.5/lib64’]
CURAND_LIBNAME = [‘curand’]
CXXFLAGS =
LDFLAGS =
but, python did not add correct libdevice:
root@Computer:/home/guolong/sailfish/examples# ./ldc_2d.py
[ 2112 INFO Master/Computer] Machine master starting with PID 2467
[ 2216 INFO Master/Computer] Sailfish version: 5beab7bbfe85df5621899780dd0a992fcc312b2e
[ 2216 INFO Master/Computer] Handling subdomains: [0]
[ 2217 INFO Master/Computer] Subdomain → GPU map: {0: 0}
[ 2341 INFO Subdomain/0] Initializing subdomain.
Process Subdomain/0:
Traceback (most recent call last):
File “/usr/lib/python2.7/multiprocessing/process.py”, line 258, in _bootstrap
self.run()
File “/usr/lib/python2.7/multiprocessing/process.py”, line 114, in run
self._target(*self._args, **self._kwargs)
File “/home/guolong/sailfish/sailfish/master.py”, line 60, in _start_subdomain_runner
runner.run()
File “/home/guolong/sailfish/sailfish/subdomain_runner.py”, line 1277, in run
self._init_compute()
File “/home/guolong/sailfish/sailfish/subdomain_runner.py”, line 698, in _init_compute
self._update_compute_code()
File “/home/guolong/sailfish/sailfish/subdomain_runner.py”, line 694, in _update_compute_code
self.module = self.backend.build(code)
File “/home/guolong/sailfish/sailfish/backend_cuda.py”, line 216, in build
cache_dir=cache) #options=[‘-Xopencc’, ‘-O0’]) #, options=[‘–use_fast_math’])
File “/usr/local/lib/python2.7/dist-packages/pycuda-2013.1.1-py2.7-linux-x86_64.egg/pycuda/compiler.py”, line 261, in init
arch, code, cache_dir, include_dirs)
File “/usr/local/lib/python2.7/dist-packages/pycuda-2013.1.1-py2.7-linux-x86_64.egg/pycuda/compiler.py”, line 251, in compile
return compile_plain(source, options, keep, nvcc, cache_dir)
File “/usr/local/lib/python2.7/dist-packages/pycuda-2013.1.1-py2.7-linux-x86_64.egg/pycuda/compiler.py”, line 134, in compile_plain
cmdline, stdout=stdout.decode(“utf-8”), stderr=stderr.decode(“utf-8”))
CompileError: nvcc compilation of /tmp/tmpGGrnyl/kernel.cu failed
[command: nvcc --cubin --prec-div=false --prec-sqrt=false -arch sm_30 -I/usr/local/lib/python2.7/dist-packages/pycuda-2013.1.1-py2.7-linux-x86_64.egg/pycuda/cuda kernel.cu]
[stderr:
nvcc fatal : Path to libdevice library not specified
]