Merge pull request #7 from f0k/cormm-cleantest

Cleanup CUDA convolution tests

theano/sandbox/cuda/tests/test_conv_cuda_ndarray.py

@@ -262,7 +262,6 @@ def exec_conv(version, shapes, verbose, random, mode,
 
     failed_version = set()
     failed_id = []
-    # I put -1 in case we forget to add version in the test to.
     for ver in version:
         for id, (ishape, kshape, subshape,
                  istride, kstride) in enumerate(shapes):
@@ -616,7 +615,7 @@ def test_valid_9_10():
               print_=print_, ones=ones, rtol=1.1e-5)
 
 
-def test_valid():
+def test_valid(conv_gemm=False):
     seed_rng()
     shapes = get_valid_shapes()
 
@@ -625,29 +624,32 @@ def test_valid():
     # I put -2 to test the reference version.
     version = [-2, -1, 6]
     verbose = 0
-#    version=[1]
 
     random = True
     print_ = False
     ones = False
     if ones:
         random = False
 
-#    exec_conv(version, shapes, verbose, random, 'valid',
-#              print_=print_, ones=ones, rtol=1.1e-5)
-
-    mode = theano_mode.including("conv_gemm")
-
-    version = [-1]
-    # Add tests with strided inputs by still square images and filters.
-    shapes += get_shapes2(scales_img=(2, 2), img_stride=(2, 2))
-    shapes += get_shapes2(scales_kern=(2, 2), kern_stride=(2, 2))
+    if conv_gemm:
+        # Test the GpuCorrMM version
+        mode = theano_mode.including("conv_gemm")
+        cls = cuda.blas.GpuCorrMM
+        version = [-1]  # dummy version; not used by GpuCorrMM so one version is enough
+        # Add tests with strided inputs by still square images and filters.
+        shapes += get_shapes2(scales_img=(2, 2), img_stride=(2, 2))
+        shapes += get_shapes2(scales_kern=(2, 2), kern_stride=(2, 2))
+    else:
+        mode = cls = None
     exec_conv(version, shapes, verbose, random, 'valid',
               print_=print_, ones=ones, rtol=1.1e-5,
-              theano_mode=mode, cls=cuda.blas.GpuCorrMM)
+              theano_mode=mode, cls=cls)
+
+def test_gemm_valid():
+    test_valid(conv_gemm=True)
 
 
-def test_full():
+def test_full(conv_gemm=False):
     seed_rng()
     shapes = get_basic_shapes()
     shapes += get_shapes2()
@@ -704,22 +706,24 @@ def test_full():
 #    shapes=shapes[:277]
     version = [-2, -1, 0, 1, 2, 3, 4, 5]
     verbose = 0
-#    version=[4]
     random = True
 
-#    exec_conv(version, shapes, verbose, random, 'full')
-
-    # Test the GpuCorrMM version
-    mode = theano_mode.including("conv_gemm")
-
-    shapes = shapes[0:10]
+    if conv_gemm:
+        # Test the GpuCorrMM version
+        mode = theano_mode.including("conv_gemm")
+        cls = cuda.blas.GpuCorrMM
+        version = [-1]  # dummy version; not used by GpuCorrMM so one version is enough
+    else:
+        mode = cls = None
     exec_conv(version, shapes, verbose, random, 'full',
-              theano_mode=mode, cls=cuda.blas.GpuCorrMM)
+              theano_mode=mode, cls=cls)
+
+def test_gemm_full():
+    test_full(conv_gemm=True)
 
 
-def test_subsample():
+def test_subsample(conv_gemm=False):
     seed_rng()
-    # implement when
     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)),
@@ -741,20 +745,23 @@ def test_subsample():
     if ones:
         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)
-
-    # Test the GpuCorrMM version
-    mode = theano_mode.including("conv_gemm")
+    if conv_gemm:
+        # Test the GpuCorrMM version
+        mode = theano_mode.including("conv_gemm")
+        cls = cuda.blas.GpuCorrMM
+        version_valid = version_full = [-1]  # dummy version; not used by GpuCorrMM so one version is enough
+    else:
+        mode = cls = None
 
     exec_conv(version_valid, shapes, verbose, random, 'valid',
               print_=print_, ones=ones,
-              theano_mode=mode, cls=cuda.blas.GpuCorrMM)
+              theano_mode=mode, cls=cls)
     exec_conv(version_full, shapes, verbose, random, 'full',
               print_=print_, ones=ones,
-              theano_mode=mode, cls=cuda.blas.GpuCorrMM)
+              theano_mode=mode, cls=cls)
+
+def test_gemm_subsample():
+    test_subsample(conv_gemm=True)
 
 
 class TestConv2DGPU(unittest.TestCase):
@@ -829,7 +836,7 @@ def test_invalid_input_shape(self):
 
 
 
-def test_gemm():
+def test_gemm_directly():
     """
     input: (batch size, channels, rows, columns)
     filters: (number of filters, channels, rows, columns)