add GPUCorrMM conv3d gradweights alternative

theano/gpuarray/opt.py

@@ -1862,6 +1862,37 @@ def local_abstractconv_gemm_gradweights_alt(node):
         return None
 
 
+@local_optimizer([AbstractConv3d_gradWeights])
+def local_abstractconv3d_gemm_gradweights_alt(node):
+    if not isinstance(node.op, AbstractConv3d_gradWeights):
+        return None
+    img, topgrad, shape = node.inputs
+    if not isinstance(img.type, GpuArrayType) or \
+            not isinstance(topgrad.type, GpuArrayType):
+        return None
+    ctx = infer_context_name(img, topgrad)
+    border_mode = node.op.border_mode
+    subsample = node.op.subsample
+    filter_dilation = node.op.filter_dilation
+
+    if border_mode == 'valid' and subsample == (1, 1, 1) and filter_dilation == (1, 1, 1):
+        rval = GpuCorr3dMM(border_mode,
+                           subsample,
+                           filter_dilation)(
+            gpu_contiguous(img.dimshuffle(1, 0, 2, 3, 4)),
+            gpu_contiguous(topgrad.dimshuffle(1, 0, 2, 3, 4)))
+
+        if node.op.filter_flip:
+            rval = rval[:, :, ::-1, ::-1, ::-1]
+
+        rval = rval.dimshuffle(1, 0, 2, 3, 4)
+        rval = tensor.patternbroadcast(rval, node.outputs[0].broadcastable)
+        rval = as_gpuarray_variable(rval, context_name=ctx)
+        return [rval]
+    else:
+        return None
+
+
 @local_optimizer([AbstractConv3d_gradWeights])
 def local_abstractconv3d_gradweights_gemm(node):
     if not isinstance(node.op, AbstractConv3d_gradWeights):
@@ -2648,6 +2679,7 @@ conv_metaopt.register([local_abstractconv_gradinputs_gemm_alt])
 conv_metaopt.register([local_abstractconv_cudnn_alternative])
 conv_metaopt.register([local_abstractconv3d2d])
 conv_metaopt.register([local_abstractconv3d_alt])
+conv_metaopt.register([local_abstractconv3d_gemm_gradweights_alt])
 abstractconv_groupopt.register('conv_metaopt', conv_metaopt, 'conv_meta', position=0)
 
 # Register cuDNN batch normalization implementation