Why an error occurs when executing a program with a GPU and there are no errors when executing a program with a CPU.

Program:

""" Created on Tue Feb 14 16:41:40 2017 @author: George """ # -*- coding: utf-8 -*- from theano import function, config, shared, sandbox import theano.tensor as T import numpy import time vlen = 10 * 30 * 768 # 10 x #cores x # threads per core iters = 1000 rng = numpy.random.RandomState(22) x = shared(numpy.asarray(rng.rand(vlen), config.floatX)) f = function([], T.exp(x)) print(f.maker.fgraph.toposort()) t0 = time.time() for i in range(iters): r = f() t1 = time.time() print("Looping %d times took %f seconds" % (iters, t1 - t0)) print("Result is %s" % (r,)) if numpy.any([isinstance(x.op, T.Elemwise) for x in f.maker.fgraph.toposort()]): print('Used the cpu') else: print('Used the gpu')

Running a program with a GPU:

 Python 3.5.2 |Anaconda custom (32-bit)| (default, Jul 5 2016, 11:45:57) [MSC v.1900 32 bit (Intel)] Type "copyright", "credits" or "license" for more information. IPython 5.1.0 -- An enhanced Interactive Python. ? -> Introduction and overview of IPython's features. %quickref -> Quick reference. help -> Python's own help system. object? -> Details about 'object', use 'object??' for extra details. runfile('H:/Users/George/untitled10.py', wdir='H:/Users/George') Using gpu device 0: GeForce GTX 980 (CNMeM is enabled with initial size: 50.0% of memory, cuDNN not available) Traceback (most recent call last): File "<ipython-input-1-f00bb712112e>", line 1, in <module> runfile('H:/Users/George/untitled10.py', wdir='H:/Users/George') File "H:\Users\George\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 866, in runfile execfile(filename, namespace) File "H:\Users\George\Anaconda3\lib\site-packages\spyder\utils\site\sitecustomize.py", line 102, in execfile exec(compile(f.read(), filename, 'exec'), namespace) File "H:/Users/George/untitled10.py", line 8, in <module> from theano import function, config, shared, sandbox File "H:\Users\George\Anaconda3\lib\site-packages\theano\__init__.py", line 111, in <module> theano.sandbox.cuda.tests.test_driver.test_nvidia_driver1() File "H:\Users\George\Anaconda3\lib\site-packages\theano\sandbox\cuda\tests\test_driver.py", line 31, in test_nvidia_driver1 profile=False) File "H:\Users\George\Anaconda3\lib\site-packages\theano\compile\function.py", line 320, in function output_keys=output_keys) File "H:\Users\George\Anaconda3\lib\site-packages\theano\compile\pfunc.py", line 479, in pfunc output_keys=output_keys) File "H:\Users\George\Anaconda3\lib\site-packages\theano\compile\function_module.py", line 1777, in orig_function defaults) File "H:\Users\George\Anaconda3\lib\site-packages\theano\compile\function_module.py", line 1641, in create input_storage=input_storage_lists, storage_map=storage_map) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\link.py", line 690, in make_thunk storage_map=storage_map)[:3] File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\vm.py", line 1003, in make_all no_recycling)) File "H:\Users\George\Anaconda3\lib\site-packages\theano\sandbox\cuda\__init__.py", line 256, in make_thunk compute_map, no_recycling) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\op.py", line 970, in make_thunk no_recycling) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\op.py", line 879, in make_c_thunk output_storage=node_output_storage) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\cc.py", line 1200, in make_thunk keep_lock=keep_lock) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\cc.py", line 1143, in __compile__ keep_lock=keep_lock) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\cc.py", line 1595, in cthunk_factory key=key, lnk=self, keep_lock=keep_lock) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\cmodule.py", line 1142, in module_from_key module = lnk.compile_cmodule(location) File "H:\Users\George\Anaconda3\lib\site-packages\theano\gof\cc.py", line 1506, in compile_cmodule preargs=preargs) File "H:\Users\George\Anaconda3\lib\site-packages\theano\sandbox\cuda\nvcc_compiler.py", line 362, in compile_str nvcc_stdout, nvcc_stderr = decode_iter(p.communicate()[:2]) File "H:\Users\George\Anaconda3\lib\site-packages\theano\compat\__init__.py", line 46, in decode_iter yield x.decode() UnicodeDecodeError: 'utf-8' codec can't decode byte 0xd1 in position 3: invalid continuation byte
  • Use code design and tags - gil9red

2 answers 2

The fact that the UnicodeDecodeError error UnicodeDecodeError caused by the decode_iter(p.communicate()[:2]) line indicates that the external program ( nvcc compiler caused by theano ) uses a different encoding from utf-8 for its output ( x.decode() uses sys.getdefaultencoding() equal to utf-8 in Python 3). This is a known bug in theano .

"Running the program from the CPU" does not cause the nvcc compiler, so you do not see this error.

If upgrading to the latest version of theano does not resolve the error, then look at the nvcc_stdout , nvcc_stderr byte string values ​​(using %debug ) to find a workaround in your case.

More information on how to choose to choose the encoding for the subprocess

    Overcame error: UnicodeDecodeError: 'utf-8' codec can't decode byte 0xd1 in position 3: invalid continuation byte changing in the file: Anaconda3 \ lib \ site-packages \ theano \ compat__init __. Py line: yield x.decode () to string yield x.decode ('cp1251').

    The result of the test program:

    Looping 1000 times took 0.998402 seconds Result is [1.23178029 1.61879349 1.52278066 ..., 2.20771813 2.29967761 1.62323296] Used the gpu

    George