cuDNN v2: Higher Performance for Deep Learning on GPUs

Originally published at:

The cuDNN library team is excited to announce the second version of cuDNN, NVIDIA’s library of GPU-accelerated primitives for deep neural networks (DNNs). We are proud that the cuDNN library has seen broad adoption by the deep learning research community and is now integrated into major deep learning toolkits such as CAFFE, Theano and Torch.…

Is there any comparison between Theano with cuDNN and Caffe with cuDNN, say for simple CNNs or multilayer perceptrons? Caffe is supposedly built for performance but I would like to see how much edge it has over Theano with both packages interfacing CUDNN.

Have you found any benchmark since then?