执行基于neon的深度强化学习用到GPU出现的问题

[pililong44]

(.venv)tommy@tommy:~/neon/simple_dqn$ ./train.sh roms/breakout.bin
A.L.E: Arcade Learning Environment (version 0.5.0)
[Powered by Stella]
Use -help for help screen.
Warning: couldn't load settings file: ./ale.cfg
Game console created:
ROM file: roms/breakout.bin
Cart Name: Breakout - Breakaway IV (1978) (Atari)
Cart MD5: f34f08e5eb96e500e851a80be3277a56
Display Format: AUTO-DETECT ==> NTSC
ROM Size: 2048
Bankswitch Type: AUTO-DETECT ==> 2K

Running ROM file...
Random seed is 0
2015-12-24 22:35:55,889 Using minimal action set with size 4
2015-12-24 22:35:55,889 Replay memory size: 1000000
2015-12-24 22:35:55,955 Initialized NervanaGPU
2015-12-24 22:35:55,955 Backend: gpu, RNG seed: None
2015-12-24 22:35:56,088 Replay memory size: 10000
2015-12-24 22:35:56,089 Results are written to results/breakout.csv
2015-12-24 22:35:56,089 Populating replay memory with 50000 random moves
Traceback (most recent call last):
File "src/main.py", line 127, in <module>
stats.write(0, "random")
File "/home/tommy/neon/simple_dqn/src/statistics.py", line 90, in write
qvalues = self.net.predict(self.validation_states)
File "/home/tommy/neon/simple_dqn/src/deepqnetwork.py", line 168, in predict
self._setInput(states)
File "/home/tommy/neon/simple_dqn/src/deepqnetwork.py", line 91, in _setInput
self.be.divide(self.input, 255, self.input)
File "/home/tommy/neon/neon/backends/backend.py", line 941, in divide
return OpTreeNode.build("div", a, b, out=out)
File "/home/tommy/neon/neon/backends/backend.py", line 2020, in build
return OpTreeNode({"op": "assign"}, out, node).execute()
File "/home/tommy/neon/neon/backends/backend.py", line 2037, in execute
return backend.execute(self)
File "/home/tommy/neon/neon/backends/nervanagpu.py", line 981, in execute
return call_compound_kernel(self._get_rand_state(), *stack)
File "/home/tommy/neon/neon/backends/float_ew.py", line 892, in call_compound_kernel
kernel = _get_compound_kernel(tuple(type_args))
File "<decorator-gen-16>", line 2, in _get_compound_kernel
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pycuda/tools.py", line 430, in context_dependent_memoize
result = func(*args)
File "/home/tommy/neon/neon/backends/float_ew.py", line 604, in _get_compound_kernel
module = SourceModule(code, options=[])
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pycuda/compiler.py", line 259, in __init__
arch, code, cache_dir, include_dirs)
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pycuda/compiler.py", line 249, in compile
return compile_plain(source, options, keep, nvcc, cache_dir, target)
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pycuda/compiler.py", line 84, in compile_plain
checksum.update(get_nvcc_version(nvcc).encode("utf-8"))
File "<decorator-gen-5>", line 2, in get_nvcc_version
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pytools/__init__.py", line 430, in _deco
result = func(*args)
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pycuda/compiler.py", line 16, in get_nvcc_version
result, stdout, stderr = call_capture_output(cmdline)
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pytools/prefork.py", line 197, in call_capture_output
return forker[0].call_capture_output(cmdline, cwd, error_on_nonzero)
File "/home/tommy/neon/.venv/local/lib/python2.7/site-packages/pytools/prefork.py", line 54, in call_capture_output
% ( " ".join(cmdline), e))
pytools.prefork.ExecError: error invoking 'nvcc --version': [Errno 2] No such file or directory

问题，该程序是基于python中的neon深度学习库的一种深度强化学习程序的执行过程，在cpu情况下可以正常运行，可是应用gpu时却出现上述的问题，请大神们帮帮忙，看看小白这里该怎么处理这类问题。。