Merge pull request #123 from nouiz/check_conv_const_shape_gpu

theano/sandbox/cuda/blas.py

@@ -456,7 +456,7 @@ class GpuConv(Op):
         return ['cuda_ndarray.cuh','<stdio.h>']
 
     def c_code_cache_version(self):
-        return (0,15) # raise this whenever modifying any of the support_code_files
+        return (0, 16) # raise this whenever modifying any of the support_code_files
 
     def c_support_code_apply(self, node, nodename):
         # REMEMBER TO RAISE c_code_cache_version when changing any of these files

theano/sandbox/cuda/conv.cu

@@ -82,7 +82,12 @@ CudaNdarray_conv_valid(const CudaNdarray *img, const CudaNdarray * kern,
     const int img_size_byte = img_size*sizeof(float);
     const int kern_size_byte = kern_size*sizeof(float);
     const int out_size_byte = out_size*sizeof(float);
-    assert((THEANO_KERN_WID == CudaNdarray_HOST_DIMS(kern)[3]) || (THEANO_KERN_WID==0));
+    if (!((THEANO_KERN_WID == CudaNdarray_HOST_DIMS(kern)[3]) || (THEANO_KERN_WID==0))){
+      PyErr_Format(PyExc_ValueError, "ERROR: This GpuConv code was compiled for"
+		   " %d kernel columns, but the kernel we received had %d columns!",
+		   THEANO_KERN_WID, CudaNdarray_HOST_DIMS(kern)[3]);
+      return -1;
+    }
 
     bool subsample = subsample_rows!=1 || subsample_cols!=1;
     bool img_contiguous = CudaNdarray_is_c_contiguous(img);
@@ -761,7 +766,12 @@ CudaNdarray_conv_full(const CudaNdarray *img, const CudaNdarray * kern, CudaNdar
     
     //const int out_size_byte = out_size*sizeof(float); // unused 
 
-    assert((THEANO_KERN_WID == CudaNdarray_HOST_DIMS(kern)[3]) || (THEANO_KERN_WID==0));
+    if (!((THEANO_KERN_WID == CudaNdarray_HOST_DIMS(kern)[3]) || (THEANO_KERN_WID==0))){
+      PyErr_Format(PyExc_ValueError, "ERROR: This GpuConv code was compiled for"
+		   " %d kernel columns, but the kernel we received had %d columns!",
+		   THEANO_KERN_WID, CudaNdarray_HOST_DIMS(kern)[3]);
+      return -1;
+    }
     bool subsample = subsample_rows!=1 || subsample_cols!=1;
 
     bool img_contiguous = CudaNdarray_is_c_contiguous(img);

theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py

-import sys, time
+import sys
+import time
+import unittest
+
+
 import numpy
 
 from nose.plugins.skip import SkipTest
@@ -86,7 +90,7 @@ def _params_allgood_header():
 
 def _params_allgood(ishape, kshape, mode, subsample=(1,1), img_stride=(1,1),
         kern_stride=(1,1), version=-1, verbose=0, random=True, print_=None,
-        id=None, rtol=1e-5, atol = 1e-8, nb_iter=0, ones=False):
+        id=None, rtol=1e-5, atol = 1e-8, nb_iter=0, ones=False, compile_kshp=None):
     #
     # This function is the core of several of the big unit-test drivers,
     # but it can also be used very directly on its own to test a specific
@@ -94,6 +98,11 @@ def _params_allgood(ishape, kshape, mode, subsample=(1,1), img_stride=(1,1),
     #
     # See `test_example` (above) for an example of how to use this directly.
     #
+    # :param kshape: (4d)The shape of the kernel at run time.
+    # :param compile_kshp: (2d) hardcode the shape of the kernel in the generated code
+    #                      This is supposed to be faster, but we need to check
+    #                      That we raise an error if the input have the wrong shape.
+    #
     if ones:
         assert not random
         npy_img = theano._asarray(numpy.ones(ishape), dtype='float32')
@@ -124,7 +133,10 @@ def _params_allgood(ishape, kshape, mode, subsample=(1,1), img_stride=(1,1),
         t1 = time.time()
         i = cuda_tensor4()
         k = cuda_tensor4()
-        op = theano.sandbox.cuda.blas.GpuConv(border_mode=mode,subsample=subsample, version=version, verbose=verbose)(i,k)
+        op = theano.sandbox.cuda.blas.GpuConv(border_mode=mode,
+                                              subsample=subsample,
+                                              version=version,
+                                              verbose=verbose, kshp=compile_kshp)(i,k)
         f=theano.function([i,k],op, mode=theano_mode)
         gpuval = f(img,kern)
         t2 = time.time()
@@ -180,7 +192,8 @@ def _params_allgood(ishape, kshape, mode, subsample=(1,1), img_stride=(1,1),
 
     return rval
 
-def exec_conv(version, shapes, verbose, random, mode, print_=None, rtol=1e-5, ones=False):
+def exec_conv(version, shapes, verbose, random, mode,
+              print_=None, rtol=1e-5, ones=False):
     if verbose>0:
         _params_allgood_header()
     nb_failed = 0
@@ -550,7 +563,7 @@ def test_full():
 
 def test_subsample():
     # implement when
-    shapes = [ 
+    shapes = [
             ((1, 1, 1, 1), (1, 1, 1, 1), (1,1), (1,1), (1,1))
             , ((1, 1, 1, 1), (1, 1, 1, 1), (2,2), (1,1), (1,1))
             , ((4, 2, 10, 10), (3, 2, 2, 2), (1, 3), (1,1), (1,1))
@@ -571,9 +584,7 @@ def test_subsample():
     ones = False
     if ones:
         random = False
-    
-    #test
-    random = False
+
     exec_conv(version_valid, shapes, verbose, random, 'valid', print_=print_, ones=ones)
     exec_conv(version_full, shapes, verbose, random, 'full', print_=print_, ones=ones)
 
@@ -583,6 +594,36 @@ def test_subsample():
 #    print >> sys.stderr, "WARNING TODO: test_logical_shapes not implemented (i.e. imshp_logical, kshp_logical, kshp_logical_top_aligned)"
 
 
+class TestConv2DGPU(unittest.TestCase):
+    def test_invalid_input_shape(self):
+        """
+        Tests that when the shape gived at build time is not the same as
+        run time we raise an error
+        """
+        verbose = 0
+        random = True
+        print_ = False
+        ones = False
+        if ones:
+            random = False
+
+        global theano_mode
+        theano_mode_orig = theano_mode
+        try:
+            if theano.config.mode in ['DebugMode', 'DEBUG_MODE']:
+                theano_mode = theano.compile.mode.get_mode('FAST_RUN').including('gpu')
+                for mode in ['valid', 'full']:
+                    for shapes in [((3, 2, 8, 8), (4, 2, 5, 5), (8, 8)),
+                                   ((3, 2, 8, 8), (4, 2, 5, 5), (5, 8)),
+                                   #((3, 2, 8, 8), (4, 2, 5, 5), (8, 5)),
+                                   # We use only the number of columns.
+                                   ]:
+
+                        self.assertRaises(ValueError, _params_allgood, shapes[0], shapes[1],
+                                          verbose=verbose, random=random, mode=mode,
+                                          print_=print_, ones=ones, compile_kshp=shapes[2])
+        finally:
+            theano_mode = theano_mode_orig
 def _test_dummy():
     ishape = (1, 1, 5, 5)
     kshape = (1, 1, 3, 3)