merge

theano/compile/debugmode.py

@@ -1414,10 +1414,12 @@ class DebugMode(Mode):
             check_c_code=None,
             check_py_code=None,
             check_isfinite=None,
-            require_matching_strides=None):
+            require_matching_strides=None,
+            linker=None):
         """Initialize member variables.
 
         If any of these arguments (except optimizer) is not None, it overrides the class default.
+        The linker arguments is not used. It is set their to allow Mode.requiring() and some other fct to work with DebugMode too.
         """
         super(DebugMode, self).__init__(
                 optimizer=optimizer,

theano/sandbox/cuda/__init__.py

@@ -12,8 +12,6 @@ from basic_ops import (GpuFromHost, HostFromGpu, GpuElemwise,
 import opt
 import cuda_ndarray
 
-import theano.compile.sandbox
-
 import os
 import theano.config as config
 from theano.compile import optdb

theano/sandbox/cuda/elemwise.py

+"""
+This file implement 3 different version of the elemwise op on the gpu. Only NaiveAlgo is used and it is not very naive now. 
+
+The elemwise fct are also used with scalar operation! So it can happen that ndim is 0 as with all scalar type.
+"""
+
+
 import StringIO, sys
 import numpy
 from theano import Op, Type, Apply, Variable, Constant

theano/sandbox/cuda/tests/test_basic_ops.py

 import sys, time
-from theano.compile.sandbox.sharedvalue import shared
-from theano.compile.sandbox.pfunc import pfunc
+
+from theano import shared
+from theano.compile.pfunc import pfunc
 from theano import tensor
 
 import numpy
+import theano
+import theano.tensor as T
 
 # Skip test if cuda_ndarray is not available.
 from nose.plugins.skip import SkipTest
@@ -13,6 +16,7 @@ except ImportError:
     raise SkipTest('Optional package cuda_ndarray not available')
 
 import theano.sandbox.cuda as tcn
+import cuda_ndarray as cuda
 import theano.compile.mode
 
 mode_with_gpu = theano.compile.mode.get_default_mode().including('gpu')
@@ -20,6 +24,63 @@ mode_with_gpu = theano.compile.mode.get_default_mode().including('gpu')
 def tes_use():
     tcn.use()
 
+def test_sum():
+    """
+    test sum pattern 1, 11, 10, 100, 110, 001, 111, 1011, 1111
+    TODO: test with broadcast
+    """
+
+    for shape, pattern in [((5,),[0]),
+                           ((5,4),[0,1]),((5,4),[0]),
+                           ((5,4,3),[0]),((5,4,3),[0,1]),((5,4,3),[2]),((5,4,3),[0,1,2]),
+                           ((5,4,3,2),[0,1,2,3]), ((5,4,3,2),[0,2,3])]:
+        a = tensor.TensorType('float32',(False,)*len(shape))()
+        b = T.Sum(pattern)(a)
+        val = numpy.random.rand(numpy.prod(shape)).reshape(shape)
+#        val = numpy.ones(shape)
+#        val = numpy.arange(numpy.prod(shape)).reshape(shape)
+        val = numpy.asarray(val,dtype='float32')
+        f = theano.function([a],b, mode=mode_with_gpu)
+        f2 = theano.function([a],b)
+        assert tcn.GpuSum in [x.op.__class__ for x in f.maker.env.toposort()]
+        assert T.Sum in [x.op.__class__ for x in f2.maker.env.toposort()]
+        assert numpy.allclose(f2(val),f(val))
+        
+
+        #test with broadcast
+    for shape, pattern in [((5,),[0]),
+                           ((5,4),[0,1]),((5,4),[0]),
+                           ((5,4,3),[0]),((5,4,3),[0,1]),((5,4,3),[2]),((5,4,3),[0,1,2]),
+                           ((5,4,3,2),[0,1,2,3]), ((5,4,3,2),[0,2,3])]:
+        shape = numpy.asarray(shape)*2
+        a = tensor.TensorType('float32',(False,)*len(shape))()
+        a2 = tcn.CudaNdarrayType((False,)*len(shape))()
+        b = T.Sum(pattern)(a)
+        b2 = T.Sum(pattern)(a2)
+        val = numpy.random.rand(numpy.prod(shape)).reshape(shape)
+#        val = numpy.ones(shape)
+#        val = numpy.arange(numpy.prod(shape)).reshape(shape)
+        val = numpy.asarray(val,dtype='float32')
+        val2 = cuda.CudaNdarray(val)
+        if len(shape)==1:
+            val = val[::2]
+            val2 = val2[::2]
+        elif len(shape)==2:
+            val = val[::2,::2]
+            val2 = val2[::2,::2]
+        elif len(shape)==3:
+            val = val[::2,::2,::2]
+            val2 = val2[::2,::2,::2]
+        elif len(shape)==4:
+            val = val[::2,::2,::2,::2]
+            val2 = val2[::2,::2,::2,::2]
+        f = theano.function([a],b)
+        f2 = theano.function([a2],b2, mode=mode_with_gpu)
+        assert tcn.GpuSum in [x.op.__class__ for x in f2.maker.env.toposort()]
+        assert T.Sum in [x.op.__class__ for x in f.maker.env.toposort()]
+        assert numpy.allclose(f2(val2),f(val))
+        
+
 def test_elemwise0():
 
     a = tcn.shared_constructor(numpy.random.rand(4,4), 'a')

theano/sandbox/cuda/var.py

@@ -2,7 +2,7 @@ import numpy
 
 from theano import Op, Type, Apply, Variable, Constant
 from theano import tensor
-from theano.compile.sandbox.sharedvalue import shared, SharedVariable, shared_constructor
+from theano.compile import shared, SharedVariable, shared_constructor
 
 from theano.sandbox.cuda.type import CudaNdarrayType
 from theano.sandbox.cuda.type_support import filter as type_support_filter