{ "id": 13447, "title": "Problems with installing theano", "url": "/competitions/diabetic-retinopathy-detection/discussion/13447", "author_name": "", "post_date": "2015-04-16T21:28:07.917Z", "votes": null, "comment_count": 8, "views": 5245, "content": "

I am desperately trying to install Theano library, but keep failing :( Could someone kindly help me with it?

\n

after installation of theano and setting environment veriables to this

\n

export THEANO_FLAGS='floatX=float32,device=gpu0,nvcc.fastmath=True,optimizer_including=cudnn'

\n

export CUDA_ROOT=/path/to/cudnn

\n

I type python -c "import theano" and get this

\n

Using gpu device 0: GeForce GTX TITAN Black
1 #include <Python.h>
2 #include <iostream>
3 #include "cudnn.h"
4 //////////////////////
5 //// Support Code
6 //////////////////////
7
8 #if PY_MAJOR_VERSION >= 3
9 #define PyInt_FromLong PyLong_FromLong
10 #endif
11
12
13 namespace {
14 struct __struct_compiled_op_f90a7a397ede0965c66895e171db1485 {
15 PyObject* __ERROR;
16
17 PyObject* storage_V1;
18
19
20 __struct_compiled_op_f90a7a397ede0965c66895e171db1485() {}
21 ~__struct_compiled_op_f90a7a397ede0965c66895e171db1485(void) {
22 cleanup();
23 }
24
25 int init(PyObject* __ERROR, PyObject* storage_V1) {
26 Py_XINCREF(storage_V1);
27 this->storage_V1 = storage_V1;
28
29
30
31 this->__ERROR = __ERROR;
32 return 0;
33 }
34 void cleanup(void) {
35 __label_1:
36
37 double __DUMMY_1;
38 __label_4:
39
40 double __DUMMY_4;
41
42 Py_XDECREF(this->storage_V1);
43 }
44 int run(void) {
45 int __failure = 0;
46
47 PyObject* py_V1;
48
49 PyObject* V1;
50
51 {
52
53 py_V1 = Py_None;
54 {Py_XINCREF(py_V1);}
55
56 V1 = NULL;
57
58 {
59 // Op class DnnVersion
60
61 #if defined(CUDNN_VERSION)
62 V1 = PyTuple_Pack(2, PyInt_FromLong(CUDNN_VERSION), PyInt_FromLong(cudnnGetVersion()));
63 #else
64 V1 = PyInt_FromLong(-1);
65 #endif
66 __label_3:
67
68 double __DUMMY_3;
69
70 }
71 __label_2:
72
73 if (!__failure) {
74
75 assert(py_V1->ob_refcnt > 1);
76 Py_DECREF(py_V1);
77 py_V1 = V1 ? V1 : Py_None;
78 Py_INCREF(py_V1);
79
80 PyObject* old = PyList_GET_ITEM(storage_V1, 0);
81 {Py_XINCREF(py_V1);}
82 PyList_SET_ITEM(storage_V1, 0, py_V1);
83 {Py_XDECREF(old);}
84 }
85
86 Py_XDECREF(V1);
87
88 {Py_XDECREF(py_V1);}
89
90 double __DUMMY_2;
91
92 }
93
94
95 if (__failure) {
96 // When there is a failure, this code puts the exception
97 // in __ERROR.
98 PyObject* err_type = NULL;
99 PyObject* err_msg = NULL;
100 PyObject* err_traceback = NULL;
101 PyErr_Fetch(&err_type, &err_msg, &err_traceback);
102 if (!err_type) {err_type = Py_None;Py_INCREF(Py_None);}
103 if (!err_msg) {err_msg = Py_None; Py_INCREF(Py_None);}
104 if (!err_traceback) {err_traceback = Py_None; Py_INCREF(Py_None);}
105 PyObject* old_err_type = PyList_GET_ITEM(__ERROR, 0);
106 PyObject* old_err_msg = PyList_GET_ITEM(__ERROR, 1);
107 PyObject* old_err_traceback = PyList_GET_ITEM(__ERROR, 2);
108 PyList_SET_ITEM(__ERROR, 0, err_type);
109 PyList_SET_ITEM(__ERROR, 1, err_msg);
110 PyList_SET_ITEM(__ERROR, 2, err_traceback);
111 {Py_XDECREF(old_err_type);}
112 {Py_XDECREF(old_err_msg);}
113 {Py_XDECREF(old_err_traceback);}
114 }
115 // The failure code is returned to index what code block failed.
116 return __failure;
117
118 }
119 };
120 }
121
122
123 static int __struct_compiled_op_f90a7a397ede0965c66895e171db1485_executor(__struct_compiled_op_f90a7a397ede0965c66895e171db1485* self) {
124 return self->run();
125 }
126
127 static void __struct_compiled_op_f90a7a397ede0965c66895e171db1485_destructor(void* executor, void* self) {
128 delete ((__struct_compiled_op_f90a7a397ede0965c66895e171db1485*)self);
129 }
130
131 //////////////////////
132 //// Functions
133 //////////////////////
134 static PyObject * instantiate(PyObject * self, PyObject *argtuple) {
135 assert(PyTuple_Check(argtuple));
136 if (2 != PyTuple_Size(argtuple)){
137 PyErr_Format(PyExc_TypeError, "Wrong number of arguments, expected 2, got %i", (int)PyTuple_Size(argtuple));
138 return NULL;
139 }
140 __struct_compiled_op_f90a7a397ede0965c66895e171db1485* struct_ptr = new __struct_compiled_op_f90a7a397ede0965c66895e171db1485();
141 if (struct_ptr->init( PyTuple_GET_ITEM(argtuple, 0),PyTuple_GET_ITEM(argtuple, 1) ) != 0) {
142 delete struct_ptr;
143 return NULL;
144 }
145 PyObject* thunk = PyCObject_FromVoidPtrAndDesc((void*)(&__struct_compiled_op_f90a7a397ede0965c66895e171db1485_executor), struct_ptr, __struct_compiled_op_f90a7a397ede0965c66895e171db1485_destructor);
146 return thunk; }
147
148 //////////////////////
149 //// Module init
150 //////////////////////
151 static PyMethodDef MyMethods[] = {
152 {"instantiate", instantiate, METH_VARARGS, "undocumented"} ,
153 {NULL, NULL, 0, NULL}
154 };
155 PyMODINIT_FUNC initf90a7a397ede0965c66895e171db1485(void){
156 (void) Py_InitModule("f90a7a397ede0965c66895e171db1485", MyMethods);
157 }
158
===============================
In file included from /usr/include/python2.7/Python.h:8:0,
from mod.cu:1:
/usr/include/python2.7/pyconfig.h:1161:0: warning: "_POSIX_C_SOURCE" redefined [enabled by default]
In file included from /usr/local/cuda-6.5/bin/..//include/host_config.h:119:0,
from /usr/local/cuda-6.5/bin/..//include/cuda_runtime.h:59,
from <command-line>:0:
/usr/include/features.h:163:0: note: this is the location of the previous definition
In file included from /usr/include/python2.7/Python.h:8:0,
from mod.cu:1:
/usr/include/python2.7/pyconfig.h:1183:0: warning: "_XOPEN_SOURCE" redefined [enabled by default]
In file included from /usr/local/cuda-6.5/bin/..//include/host_config.h:119:0,
from /usr/local/cuda-6.5/bin/..//include/cuda_runtime.h:59,
from <command-line>:0:
/usr/include/features.h:165:0: note: this is the location of the previous definition
mod.cu:3:19: fatal error: cudnn.h: No such file or directory
compilation terminated.

\n

['nvcc', '-shared', '-O3', '-use_fast_math', '-arch=sm_35', '-m64', '-Xcompiler', '-fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=b0c165044f8fd1e0f58745b1202dab6e,-D NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-D NPY_ARRAY_ENSURECOPY=NPY_ENSURECOPY,-D NPY_ARRAY_ALIGNED=NPY_ALIGNED,-D NPY_ARRAY_WRITEABLE=NPY_WRITEABLE,-D NPY_ARRAY_UPDATE_ALL=NPY_UPDATE_ALL,-D NPY_ARRAY_C_CONTIGUOUS=NPY_C_CONTIGUOUS,-D NPY_ARRAY_F_CONTIGUOUS=NPY_F_CONTIGUOUS,-fPIC', '-Xlinker', '-rpath,/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/cuda_ndarray', '-I/usr/lib/pymodules/python2.7/numpy/core/include', '-I/usr/include/python2.7', '-I/mnt/home.dokt/antonkos/code/Theano/theano/sandbox/cuda', '-o', '/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/tmpOc31Ki/f90a7a397ede0965c66895e171db1485.so', 'mod.cu', '-L/usr/lib', '-lpython2.7', '-lcudnn', '-lcudart']
ERROR (theano.gof.opt): SeqOptimizer apply <theano.sandbox.cuda.dnn.NoCuDNNRaise object at 0x3a32250>
ERROR (theano.gof.opt): Traceback:
ERROR (theano.gof.opt): Traceback (most recent call last):
File "theano/gof/opt.py", line 195, in apply
sub_prof = optimizer.optimize(fgraph)
File "theano/gof/opt.py", line 81, in optimize
ret = self.apply(fgraph, *args, **kwargs)
File "theano/sandbox/cuda/dnn.py", line 1680, in apply
if not dnn_available():
File "theano/sandbox/cuda/dnn.py", line 75, in dnn_available
v = version()
File "theano/sandbox/cuda/dnn.py", line 206, in version
theano.Mode(optimizer=None))
File "theano/compile/function.py", line 286, in function
output_keys=output_keys)
File "theano/compile/pfunc.py", line 511, in pfunc
on_unused_input=on_unused_input, output_keys=output_keys)
File "theano/compile/function_module.py", line 1484, in orig_function
defaults)
File "theano/compile/function_module.py", line 1335, in create
input_storage=input_storage_lists)
File "theano/gof/link.py", line 559, in make_thunk
output_storage=output_storage)[:3]
File "theano/gof/vm.py", line 996, in make_all
no_recycling))
File "theano/sandbox/cuda/__init__.py", line 258, in make_thunk
compute_map, no_recycling)
File "theano/gof/op.py", line 749, in make_thunk
output_storage=node_output_storage)
File "theano/gof/cc.py", line 1072, in make_thunk
keep_lock=keep_lock)
File "theano/gof/cc.py", line 1014, in __compile__
keep_lock=keep_lock)
File "theano/gof/cc.py", line 1441, in cthunk_factory
key=key, lnk=self, keep_lock=keep_lock)
File "theano/gof/cmodule.py", line 1075, in module_from_key
module = lnk.compile_cmodule(location)
File "theano/gof/cc.py", line 1353, in compile_cmodule
preargs=preargs)
File "theano/sandbox/cuda/nvcc_compiler.py", line 423, in compile_str
'for cmd', ' '.join(cmd))
Exception: ('The following error happened while compiling the node', <theano.sandbox.cuda.dnn.DnnVersion object at 0xc10f610>(), '\\n', 'nvcc return status', 1, 'for cmd', 'nvcc -shared -O3 -use_fast_math -arch=sm_35 -m64 -Xcompiler -fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=b0c165044f8fd1e0f58745b1202dab6e,-D NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-D NPY_ARRAY_ENSURECOPY=NPY_ENSURECOPY,-D NPY_ARRAY_ALIGNED=NPY_ALIGNED,-D NPY_ARRAY_WRITEABLE=NPY_WRITEABLE,-D NPY_ARRAY_UPDATE_ALL=NPY_UPDATE_ALL,-D NPY_ARRAY_C_CONTIGUOUS=NPY_C_CONTIGUOUS,-D NPY_ARRAY_F_CONTIGUOUS=NPY_F_CONTIGUOUS,-fPIC -Xlinker -rpath,/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/cuda_ndarray -I/usr/lib/pymodules/python2.7/numpy/core/include -I/usr/include/python2.7 -I/mnt/home.dokt/antonkos/code/Theano/theano/sandbox/cuda -o /mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/tmpOc31Ki/f90a7a397ede0965c66895e171db1485.so mod.cu -L/usr/lib -lpython2.7 -lcudnn -lcudart', '[<theano.sandbox.cuda.dnn.DnnVersion object at 0xc10f610>()]')

\n

when I run the command on which node building failed i.e.

\n

nvcc -shared -O3 -use_fast_math -arch=sm_35 -m64 -Xcompiler -fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=b0c165044f8fd1e0f58745b1202dab6e,-D NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-D NPY_ARRAY_ENSURECOPY=NPY_ENSURECOPY,-D NPY_ARRAY_ALIGNED=NPY_ALIGNED,-fPIC -Xlinker -rpath,/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/cuda_ndarray -I/usr/lib/pymodules/python2.7/numpy/core/include -I/usr/include/python2.7 -I/mnt/home.dokt/antonkos/code/Theano/theano/sandbox/cuda -o /mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/tmpOc31Ki/f90a7a397ede0965c66895e171db1485.so mod.cu -L/usr/lib -lpython2.7 -lcudnn -lcudart

\n

it says me nvcc fatal : Don't know what to do with 'NPY_ARRAY_ALIGNED=NPY_ALIGNED,-fPIC'

\n

I'm puzzled with this. Would appreciate a lot if someone can give relevant advises...

", "messages": [ { "id": "72002", "postDate": "04/16/2015 21:28:07", "content": "

I am desperately trying to install Theano library, but keep failing :( Could someone kindly help me with it?

\n

after installation of theano and setting environment veriables to this

\n

export THEANO_FLAGS='floatX=float32,device=gpu0,nvcc.fastmath=True,optimizer_including=cudnn'

\n

export CUDA_ROOT=/path/to/cudnn

\n

I type python -c "import theano" and get this

\n

Using gpu device 0: GeForce GTX TITAN Black
1 #include <Python.h>
2 #include <iostream>
3 #include "cudnn.h"
4 //////////////////////
5 //// Support Code
6 //////////////////////
7
8 #if PY_MAJOR_VERSION >= 3
9 #define PyInt_FromLong PyLong_FromLong
10 #endif
11
12
13 namespace {
14 struct __struct_compiled_op_f90a7a397ede0965c66895e171db1485 {
15 PyObject* __ERROR;
16
17 PyObject* storage_V1;
18
19
20 __struct_compiled_op_f90a7a397ede0965c66895e171db1485() {}
21 ~__struct_compiled_op_f90a7a397ede0965c66895e171db1485(void) {
22 cleanup();
23 }
24
25 int init(PyObject* __ERROR, PyObject* storage_V1) {
26 Py_XINCREF(storage_V1);
27 this->storage_V1 = storage_V1;
28
29
30
31 this->__ERROR = __ERROR;
32 return 0;
33 }
34 void cleanup(void) {
35 __label_1:
36
37 double __DUMMY_1;
38 __label_4:
39
40 double __DUMMY_4;
41
42 Py_XDECREF(this->storage_V1);
43 }
44 int run(void) {
45 int __failure = 0;
46
47 PyObject* py_V1;
48
49 PyObject* V1;
50
51 {
52
53 py_V1 = Py_None;
54 {Py_XINCREF(py_V1);}
55
56 V1 = NULL;
57
58 {
59 // Op class DnnVersion
60
61 #if defined(CUDNN_VERSION)
62 V1 = PyTuple_Pack(2, PyInt_FromLong(CUDNN_VERSION), PyInt_FromLong(cudnnGetVersion()));
63 #else
64 V1 = PyInt_FromLong(-1);
65 #endif
66 __label_3:
67
68 double __DUMMY_3;
69
70 }
71 __label_2:
72
73 if (!__failure) {
74
75 assert(py_V1->ob_refcnt > 1);
76 Py_DECREF(py_V1);
77 py_V1 = V1 ? V1 : Py_None;
78 Py_INCREF(py_V1);
79
80 PyObject* old = PyList_GET_ITEM(storage_V1, 0);
81 {Py_XINCREF(py_V1);}
82 PyList_SET_ITEM(storage_V1, 0, py_V1);
83 {Py_XDECREF(old);}
84 }
85
86 Py_XDECREF(V1);
87
88 {Py_XDECREF(py_V1);}
89
90 double __DUMMY_2;
91
92 }
93
94
95 if (__failure) {
96 // When there is a failure, this code puts the exception
97 // in __ERROR.
98 PyObject* err_type = NULL;
99 PyObject* err_msg = NULL;
100 PyObject* err_traceback = NULL;
101 PyErr_Fetch(&err_type, &err_msg, &err_traceback);
102 if (!err_type) {err_type = Py_None;Py_INCREF(Py_None);}
103 if (!err_msg) {err_msg = Py_None; Py_INCREF(Py_None);}
104 if (!err_traceback) {err_traceback = Py_None; Py_INCREF(Py_None);}
105 PyObject* old_err_type = PyList_GET_ITEM(__ERROR, 0);
106 PyObject* old_err_msg = PyList_GET_ITEM(__ERROR, 1);
107 PyObject* old_err_traceback = PyList_GET_ITEM(__ERROR, 2);
108 PyList_SET_ITEM(__ERROR, 0, err_type);
109 PyList_SET_ITEM(__ERROR, 1, err_msg);
110 PyList_SET_ITEM(__ERROR, 2, err_traceback);
111 {Py_XDECREF(old_err_type);}
112 {Py_XDECREF(old_err_msg);}
113 {Py_XDECREF(old_err_traceback);}
114 }
115 // The failure code is returned to index what code block failed.
116 return __failure;
117
118 }
119 };
120 }
121
122
123 static int __struct_compiled_op_f90a7a397ede0965c66895e171db1485_executor(__struct_compiled_op_f90a7a397ede0965c66895e171db1485* self) {
124 return self->run();
125 }
126
127 static void __struct_compiled_op_f90a7a397ede0965c66895e171db1485_destructor(void* executor, void* self) {
128 delete ((__struct_compiled_op_f90a7a397ede0965c66895e171db1485*)self);
129 }
130
131 //////////////////////
132 //// Functions
133 //////////////////////
134 static PyObject * instantiate(PyObject * self, PyObject *argtuple) {
135 assert(PyTuple_Check(argtuple));
136 if (2 != PyTuple_Size(argtuple)){
137 PyErr_Format(PyExc_TypeError, "Wrong number of arguments, expected 2, got %i", (int)PyTuple_Size(argtuple));
138 return NULL;
139 }
140 __struct_compiled_op_f90a7a397ede0965c66895e171db1485* struct_ptr = new __struct_compiled_op_f90a7a397ede0965c66895e171db1485();
141 if (struct_ptr->init( PyTuple_GET_ITEM(argtuple, 0),PyTuple_GET_ITEM(argtuple, 1) ) != 0) {
142 delete struct_ptr;
143 return NULL;
144 }
145 PyObject* thunk = PyCObject_FromVoidPtrAndDesc((void*)(&__struct_compiled_op_f90a7a397ede0965c66895e171db1485_executor), struct_ptr, __struct_compiled_op_f90a7a397ede0965c66895e171db1485_destructor);
146 return thunk; }
147
148 //////////////////////
149 //// Module init
150 //////////////////////
151 static PyMethodDef MyMethods[] = {
152 {"instantiate", instantiate, METH_VARARGS, "undocumented"} ,
153 {NULL, NULL, 0, NULL}
154 };
155 PyMODINIT_FUNC initf90a7a397ede0965c66895e171db1485(void){
156 (void) Py_InitModule("f90a7a397ede0965c66895e171db1485", MyMethods);
157 }
158
===============================
In file included from /usr/include/python2.7/Python.h:8:0,
from mod.cu:1:
/usr/include/python2.7/pyconfig.h:1161:0: warning: "_POSIX_C_SOURCE" redefined [enabled by default]
In file included from /usr/local/cuda-6.5/bin/..//include/host_config.h:119:0,
from /usr/local/cuda-6.5/bin/..//include/cuda_runtime.h:59,
from <command-line>:0:
/usr/include/features.h:163:0: note: this is the location of the previous definition
In file included from /usr/include/python2.7/Python.h:8:0,
from mod.cu:1:
/usr/include/python2.7/pyconfig.h:1183:0: warning: "_XOPEN_SOURCE" redefined [enabled by default]
In file included from /usr/local/cuda-6.5/bin/..//include/host_config.h:119:0,
from /usr/local/cuda-6.5/bin/..//include/cuda_runtime.h:59,
from <command-line>:0:
/usr/include/features.h:165:0: note: this is the location of the previous definition
mod.cu:3:19: fatal error: cudnn.h: No such file or directory
compilation terminated.

\n

['nvcc', '-shared', '-O3', '-use_fast_math', '-arch=sm_35', '-m64', '-Xcompiler', '-fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=b0c165044f8fd1e0f58745b1202dab6e,-D NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-D NPY_ARRAY_ENSURECOPY=NPY_ENSURECOPY,-D NPY_ARRAY_ALIGNED=NPY_ALIGNED,-D NPY_ARRAY_WRITEABLE=NPY_WRITEABLE,-D NPY_ARRAY_UPDATE_ALL=NPY_UPDATE_ALL,-D NPY_ARRAY_C_CONTIGUOUS=NPY_C_CONTIGUOUS,-D NPY_ARRAY_F_CONTIGUOUS=NPY_F_CONTIGUOUS,-fPIC', '-Xlinker', '-rpath,/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/cuda_ndarray', '-I/usr/lib/pymodules/python2.7/numpy/core/include', '-I/usr/include/python2.7', '-I/mnt/home.dokt/antonkos/code/Theano/theano/sandbox/cuda', '-o', '/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/tmpOc31Ki/f90a7a397ede0965c66895e171db1485.so', 'mod.cu', '-L/usr/lib', '-lpython2.7', '-lcudnn', '-lcudart']
ERROR (theano.gof.opt): SeqOptimizer apply <theano.sandbox.cuda.dnn.NoCuDNNRaise object at 0x3a32250>
ERROR (theano.gof.opt): Traceback:
ERROR (theano.gof.opt): Traceback (most recent call last):
File "theano/gof/opt.py", line 195, in apply
sub_prof = optimizer.optimize(fgraph)
File "theano/gof/opt.py", line 81, in optimize
ret = self.apply(fgraph, *args, **kwargs)
File "theano/sandbox/cuda/dnn.py", line 1680, in apply
if not dnn_available():
File "theano/sandbox/cuda/dnn.py", line 75, in dnn_available
v = version()
File "theano/sandbox/cuda/dnn.py", line 206, in version
theano.Mode(optimizer=None))
File "theano/compile/function.py", line 286, in function
output_keys=output_keys)
File "theano/compile/pfunc.py", line 511, in pfunc
on_unused_input=on_unused_input, output_keys=output_keys)
File "theano/compile/function_module.py", line 1484, in orig_function
defaults)
File "theano/compile/function_module.py", line 1335, in create
input_storage=input_storage_lists)
File "theano/gof/link.py", line 559, in make_thunk
output_storage=output_storage)[:3]
File "theano/gof/vm.py", line 996, in make_all
no_recycling))
File "theano/sandbox/cuda/__init__.py", line 258, in make_thunk
compute_map, no_recycling)
File "theano/gof/op.py", line 749, in make_thunk
output_storage=node_output_storage)
File "theano/gof/cc.py", line 1072, in make_thunk
keep_lock=keep_lock)
File "theano/gof/cc.py", line 1014, in __compile__
keep_lock=keep_lock)
File "theano/gof/cc.py", line 1441, in cthunk_factory
key=key, lnk=self, keep_lock=keep_lock)
File "theano/gof/cmodule.py", line 1075, in module_from_key
module = lnk.compile_cmodule(location)
File "theano/gof/cc.py", line 1353, in compile_cmodule
preargs=preargs)
File "theano/sandbox/cuda/nvcc_compiler.py", line 423, in compile_str
'for cmd', ' '.join(cmd))
Exception: ('The following error happened while compiling the node', <theano.sandbox.cuda.dnn.DnnVersion object at 0xc10f610>(), '\\n', 'nvcc return status', 1, 'for cmd', 'nvcc -shared -O3 -use_fast_math -arch=sm_35 -m64 -Xcompiler -fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=b0c165044f8fd1e0f58745b1202dab6e,-D NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-D NPY_ARRAY_ENSURECOPY=NPY_ENSURECOPY,-D NPY_ARRAY_ALIGNED=NPY_ALIGNED,-D NPY_ARRAY_WRITEABLE=NPY_WRITEABLE,-D NPY_ARRAY_UPDATE_ALL=NPY_UPDATE_ALL,-D NPY_ARRAY_C_CONTIGUOUS=NPY_C_CONTIGUOUS,-D NPY_ARRAY_F_CONTIGUOUS=NPY_F_CONTIGUOUS,-fPIC -Xlinker -rpath,/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/cuda_ndarray -I/usr/lib/pymodules/python2.7/numpy/core/include -I/usr/include/python2.7 -I/mnt/home.dokt/antonkos/code/Theano/theano/sandbox/cuda -o /mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/tmpOc31Ki/f90a7a397ede0965c66895e171db1485.so mod.cu -L/usr/lib -lpython2.7 -lcudnn -lcudart', '[<theano.sandbox.cuda.dnn.DnnVersion object at 0xc10f610>()]')

\n

when I run the command on which node building failed i.e.

\n

nvcc -shared -O3 -use_fast_math -arch=sm_35 -m64 -Xcompiler -fno-math-errno,-Wno-unused-label,-Wno-unused-variable,-Wno-write-strings,-DCUDA_NDARRAY_CUH=b0c165044f8fd1e0f58745b1202dab6e,-D NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION,-D NPY_ARRAY_ENSURECOPY=NPY_ENSURECOPY,-D NPY_ARRAY_ALIGNED=NPY_ALIGNED,-fPIC -Xlinker -rpath,/mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/cuda_ndarray -I/usr/lib/pymodules/python2.7/numpy/core/include -I/usr/include/python2.7 -I/mnt/home.dokt/antonkos/code/Theano/theano/sandbox/cuda -o /mnt/home.dokt/antonkos/.theano/compiledir_Linux-3.2--amd64-x86_64-with-debian-7.8--2.7.3-64/tmpOc31Ki/f90a7a397ede0965c66895e171db1485.so mod.cu -L/usr/lib -lpython2.7 -lcudnn -lcudart

\n

it says me nvcc fatal : Don't know what to do with 'NPY_ARRAY_ALIGNED=NPY_ALIGNED,-fPIC'

\n

I'm puzzled with this. Would appreciate a lot if someone can give relevant advises...

", "rawMarkdown": "", "votes": null }, { "id": "72352", "postDate": "04/18/2015 21:08:06", "content": "

Not exactly sure where your problem lies, but this blog post and nvidia documentation really helped me to get cuda c and theano (and pylearn2 as well) to play nice with each other.  

\n

http://markus.com/install-theano-on-aws/

\n

http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/#axzz3XhIiPlrK

", "rawMarkdown": "", "votes": null }, { "id": "72353", "postDate": "04/18/2015 21:12:07", "content": "

I want to use cudnn for theano, however I can't simply copy *.so & *.h files in /usr/lib64 & /usr/include folders because I don't have rights for that. It seems theano has some bug in case if one sets it through CUDA_ROOT or LD_LIBRARY_PATH or just PATH variables. 

\n

Currently I am trying to fix this issue in theano source code.

\n\n

UPD. I am not trying to install it on Amazon server. This is some local cluster without admin rights.

\n

-K.

", "rawMarkdown": "", "votes": null }, { "id": "72498", "postDate": "04/19/2015 19:00:49", "content": "

There is a few suspicious things over there, I think:

\n

export CUDA_ROOT=/path/to/cudnn
- did you set exactly the above string to CUDA_ROOT?

\n

export THEANO_FLAGS='floatX=float32,device=gpu0,nvcc.fastmath=True,optimizer_including=cudnn'
- you may use .theanoarc file to configure theano, because that is more transparent to use.

\n

I can't simply copy *.so & *.h files in /usr/lib64 & /usr/include folders because I don't have rights for that
- you will be able if you'll try to do it as root

", "rawMarkdown": "", "votes": null }, { "id": "72514", "postDate": "04/19/2015 20:01:19", "content": "

Here's what I added to my .bashrc to get Theano to use cuDNN (without needing root):

\n

export PATH=$PATH:/usr/local/cuda/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/mnt/storage/users/sedielem/lib/cudnn-R2
export LIBRARY_PATH=$LIBRARY_PATH:/mnt/storage/users/sedielem/lib/cudnn-R2
export CPATH=/mnt/storage/users/sedielem/lib/cudnn-R2

\n

The directory /mnt/storage/users/sedielem/lib/cudnn-R2 contains all the files from the archive you can download from NVIDIA.

\n

Note that depending on your distro, the CUDA path may be different as well.

", "rawMarkdown": "", "votes": null }, { "id": "72522", "postDate": "04/19/2015 21:12:12", "content": "

[quote=George Solymosi;72498]

\n

There is a few suspicious things over there, I think:

\n

export CUDA_ROOT=/path/to/cudnn
- did you set exactly the above string to CUDA_ROOT?

\n

[/quote]

\n

yes, CUDA_ROOT indeed should be set to the directory with CUDA.

\n

I had this when I was trying different options, but after going through theano's source code I understood that setting CUDA_ROOT to directory with cudnn is wrong thing.

\n

[quote]

\n

export THEANO_FLAGS='floatX=float32,device=gpu0,nvcc.fastmath=True,optimizer_including=cudnn'
- you may use .theanoarc file to configure theano, because that is more transparent to use.

\n

I can't simply copy *.so & *.h files in /usr/lib64 & /usr/include folders because I don't have rights for that
- you will be able if you'll try to do it as root

\n

[/quote]

\n

The problem is that I can't get root rights on that machine, so solution with simply putting cudnn's *.so & *.h in proper place is not suitable for me. Browsing through internet I understood that actually nobody who tried was able to put cudnn to different directory than CUDA_ROOT and make theano working with cudnn. Going through theano's source code I figured out that theano actually assumes that cudnn is in CUDA_ROOT. So, I modified theano to be able to find cudnn in different directory from CUDA_ROOT. Now, my version of theano works with CUDNN_ROOT additionally, i.e. if such environment variable exists theano additionally looks into this directory when it needs cudnn for compilation.

", "rawMarkdown": "", "votes": null }, { "id": "72523", "postDate": "04/19/2015 21:14:18", "content": "

[quote=sedielem;72514]

\n

Here's what I added to my .bashrc to get Theano to use cuDNN (without needing root):

\n

export PATH=$PATH:/usr/local/cuda/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/mnt/storage/users/sedielem/lib/cudnn-R2
export LIBRARY_PATH=$LIBRARY_PATH:/mnt/storage/users/sedielem/lib/cudnn-R2
export CPATH=/mnt/storage/users/sedielem/lib/cudnn-R2

\n

The directory /mnt/storage/users/sedielem/lib/cudnn-R2 contains all the files from the archive you can download from NVIDIA.

\n

Note that depending on your distro, the CUDA path may be different as well.

\n

[/quote]

\n\n

I saw these solutions on theano web-page dedicated to cudnn, however for some reason none of these worked for me. 

", "rawMarkdown": "", "votes": null }, { "id": "73150", "postDate": "04/22/2015 12:12:00", "content": "

Hi, I am new here.

\n

I am just wondering is cuDNN library allowed to use in this competition?

", "rawMarkdown": "", "votes": null }, { "id": "79396", "postDate": "05/20/2015 07:09:25", "content": "

I fiddled around with al lot of these kinds of errors and solutions. It was only after I put all the settings in a .theanoarc file that I got it to work without errors.

", "rawMarkdown": "", "votes": null } ], "comments": [ { "id": 72352, "author_name": "gobrewers14", "author_url": "", "post_date": "04/18/2015 21:08:06", "content": "

Not exactly sure where your problem lies, but this blog post and nvidia documentation really helped me to get cuda c and theano (and pylearn2 as well) to play nice with each other.  

\n

http://markus.com/install-theano-on-aws/

\n

http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/#axzz3XhIiPlrK

", "votes": null, "replies": [] }, { "id": 72353, "author_name": "", "author_url": "", "post_date": "04/18/2015 21:12:07", "content": "

I want to use cudnn for theano, however I can't simply copy *.so & *.h files in /usr/lib64 & /usr/include folders because I don't have rights for that. It seems theano has some bug in case if one sets it through CUDA_ROOT or LD_LIBRARY_PATH or just PATH variables. 

\n

Currently I am trying to fix this issue in theano source code.

\n\n

UPD. I am not trying to install it on Amazon server. This is some local cluster without admin rights.

\n

-K.

", "votes": null, "replies": [] }, { "id": 72498, "author_name": "", "author_url": "", "post_date": "04/19/2015 19:00:49", "content": "

There is a few suspicious things over there, I think:

\n

export CUDA_ROOT=/path/to/cudnn
- did you set exactly the above string to CUDA_ROOT?

\n

export THEANO_FLAGS='floatX=float32,device=gpu0,nvcc.fastmath=True,optimizer_including=cudnn'
- you may use .theanoarc file to configure theano, because that is more transparent to use.

\n

I can't simply copy *.so & *.h files in /usr/lib64 & /usr/include folders because I don't have rights for that
- you will be able if you'll try to do it as root

", "votes": null, "replies": [] }, { "id": 72514, "author_name": "sedielem", "author_url": "", "post_date": "04/19/2015 20:01:19", "content": "

Here's what I added to my .bashrc to get Theano to use cuDNN (without needing root):

\n

export PATH=$PATH:/usr/local/cuda/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/mnt/storage/users/sedielem/lib/cudnn-R2
export LIBRARY_PATH=$LIBRARY_PATH:/mnt/storage/users/sedielem/lib/cudnn-R2
export CPATH=/mnt/storage/users/sedielem/lib/cudnn-R2

\n

The directory /mnt/storage/users/sedielem/lib/cudnn-R2 contains all the files from the archive you can download from NVIDIA.

\n

Note that depending on your distro, the CUDA path may be different as well.

", "votes": null, "replies": [] }, { "id": 72522, "author_name": "", "author_url": "", "post_date": "04/19/2015 21:12:12", "content": "

[quote=George Solymosi;72498]

\n

There is a few suspicious things over there, I think:

\n

export CUDA_ROOT=/path/to/cudnn
- did you set exactly the above string to CUDA_ROOT?

\n

[/quote]

\n

yes, CUDA_ROOT indeed should be set to the directory with CUDA.

\n

I had this when I was trying different options, but after going through theano's source code I understood that setting CUDA_ROOT to directory with cudnn is wrong thing.

\n

[quote]

\n

export THEANO_FLAGS='floatX=float32,device=gpu0,nvcc.fastmath=True,optimizer_including=cudnn'
- you may use .theanoarc file to configure theano, because that is more transparent to use.

\n

I can't simply copy *.so & *.h files in /usr/lib64 & /usr/include folders because I don't have rights for that
- you will be able if you'll try to do it as root

\n

[/quote]

\n

The problem is that I can't get root rights on that machine, so solution with simply putting cudnn's *.so & *.h in proper place is not suitable for me. Browsing through internet I understood that actually nobody who tried was able to put cudnn to different directory than CUDA_ROOT and make theano working with cudnn. Going through theano's source code I figured out that theano actually assumes that cudnn is in CUDA_ROOT. So, I modified theano to be able to find cudnn in different directory from CUDA_ROOT. Now, my version of theano works with CUDNN_ROOT additionally, i.e. if such environment variable exists theano additionally looks into this directory when it needs cudnn for compilation.

", "votes": null, "replies": [] }, { "id": 72523, "author_name": "", "author_url": "", "post_date": "04/19/2015 21:14:18", "content": "

[quote=sedielem;72514]

\n

Here's what I added to my .bashrc to get Theano to use cuDNN (without needing root):

\n

export PATH=$PATH:/usr/local/cuda/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/mnt/storage/users/sedielem/lib/cudnn-R2
export LIBRARY_PATH=$LIBRARY_PATH:/mnt/storage/users/sedielem/lib/cudnn-R2
export CPATH=/mnt/storage/users/sedielem/lib/cudnn-R2

\n

The directory /mnt/storage/users/sedielem/lib/cudnn-R2 contains all the files from the archive you can download from NVIDIA.

\n

Note that depending on your distro, the CUDA path may be different as well.

\n

[/quote]

\n\n

I saw these solutions on theano web-page dedicated to cudnn, however for some reason none of these worked for me. 

", "votes": null, "replies": [] }, { "id": 73150, "author_name": "rbnwkt", "author_url": "", "post_date": "04/22/2015 12:12:00", "content": "

Hi, I am new here.

\n

I am just wondering is cuDNN library allowed to use in this competition?

", "votes": null, "replies": [] }, { "id": 79396, "author_name": "dfjmulder", "author_url": "", "post_date": "05/20/2015 07:09:25", "content": "

I fiddled around with al lot of these kinds of errors and solutions. It was only after I put all the settings in a .theanoarc file that I got it to work without errors.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "72002": "", "72352": "", "72353": "", "72498": "", "72514": "", "72522": "", "72523": "", "73150": "", "79396": "" }, "source": "meta" }