Hi, I'm very interested in your paper Fast Convolutional Nets With fbfft: A GPU Performance Evaluation. I have looked at your code but can hardly know how to use it. Could you provide some test files for me to test?

cc@fft:~/code/fbcunn/build$ cmake ..
STATUS,=
STATUS,=
STATUS,=
-- Found Torch7 in /home/cc/torch/install
-- Found CUDA: /usr/local/cuda (found suitable version "8.0", minimum required is "8.0")
STATUS,/usr/local/cuda/lib64/libcudart.so=
STATUS,=
STATUS,/usr/local/cuda/lib64/libcublas.so;/usr/local/cuda/lib64/libcublas_device.a=
-- Configuring done
CMake Error at /home/cc/torch/install/share/cmake/torch/FindCUDA.cmake:1711 (add_library):
Cannot find source file:

cuda/util/CachedDeviceProperties.cpp

Tried extensions .c .C .c++ .cc .cpp .cxx .m .M .mm .h .hh .h++ .hm .hpp
.hxx .in .txx
Call Stack (most recent call first):
CMakeLists.txt:382 (CUDA_ADD_LIBRARY)

-- Build files have been written to: /home/cc/code/fbcunn/build

https://github.com/facebook/fbcuda/blob/1a5dbe62410cf639f224dcb9ea1643fe216d5ccb/CudaUtils.cuh#L23

/// Finds the minimum `val` across the warp
template <typename T>
__device__ __forceinline__ T warpReduceMin(T val) {
  return warpReduce<T, Max<T> >(val, Min<T>());
}

should be

  return warpReduce<T, Min<T> >(val, Min<T>());

right?