Skip to content


Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time
CUV Documentation



CUV is a C++ template and Python library which makes it easy to use NVIDIA(tm)


Supported Platforms:

  • This library was only tested on Ubuntu Karmic, Lucid and Maverick. It uses
    mostly standard components (except PyUBLAS) and should run without major
    modification on any current linux system.

Supported GPUs:

  • By default, code is generated for the lowest compute architecture. We
    recommend you change this to match your hardware. Using ccmake you can set
    the build variable "CUDA_ARCHITECTURE" for example to -arch=compute_20
  • All GT 9800 and GTX 280 and above
  • GT 9200 without convolutions. It might need some minor modifications to
    make the rest work. If you want to use that card and have problems, just
    get in contact.
  • On 8800GTS, random numbers and convolutions wont work.


  • Like for example Matlab, CUV assumes that everything is an n-dimensional
    array called "tensor"
  • Tensors can have an arbitrary data-type and can be on the host (CPU-memory)
    or device (GPU-memory)
  • Tensors can be column-major or row-major (1-dimensional tensors are, by
    convention, row-major)
  • The library defines many functions which may or may not apply to all
    possible combinations. Variations are easy to add.
  • For convenience, we also wrap some of the functionality provided by Alex
    Krizhevsky on his website ( with
    permission. Thanks Alex for providing your code!

Python Integration

  • CUV plays well with python and numpy. That is, once you wrote your fast GPU
    functions in CUDA/C++, you can export them using Boost.Python. You can use
    Numpy for pre-processing and fancy stuff you have not yet implemented, then
    push the Numpy-matrix to the GPU, run your operations there, pull again to
    CPU and visualize using matplotlib. Great.

Implemented Functionality

  • Simple Linear Algebra for dense vectors and matrices (BLAS level 1,2,3)
  • Helpful functors and abstractions
  • Sparse matrices in DIA format and matrix-multiplication for these matrices
  • I/O functions using boost.serialization
  • Fast Random Number Generator
  • Up to now, CUV was used to build dense and sparse Neural Networks and
    Restricted Boltzmann Machines (RBM), convolutional or locally connected.


  • Tutorials are available on
  • The documentation can be generated from the code or accessed on the


  • We are eager to help you getting started with CUV and improve the library
    continuously! If you have any questions, feel free to contact Hannes Schulz
    (schulz at ais dot uni-bonn dot de) or Andreas Mueller (amueller at ais dot
    uni-bonn dot de). You can find the website of our group at http://



For C++ libs, you will need:

  • cmake (and cmake-curses-gui for easy configuration)
  • libboost-dev >= 1.37
  • libblas-dev
  • libtemplate-perl -- (we might get rid of this dependency soon)
  • NVIDIA CUDA (tm), including SDK. We support versions 3.X and 4.0
  • thrust library - included in CUDA since 4.0 (otherwise available from http:
  • doxygen (if you want to build the documentation yourself)

For Python Integration, you additionally have to install

  • pyublas -- from
  • python-nose -- for python testing
  • python-dev

Optionally, install dependent libraries

  • cimg-dev for visualization of matrices (grayscale only, ATM)

Obtaining CUV

You should check out the git repository

   $ git clone git://

Installation Procedure

Building a debug version:

 $ cd cuv-version-source
 $ mkdir -p build/debug
 $ cd build/debug
 $ cmake -DCMAKE_BUILD_TYPE=Debug ../../
 $ ccmake .          # adjust paths to your system (cuda, thrust, pyublas, ...)!
                     # turn on/off optional libraries (CImg, ...)
 $ make -j
 $ ctest             # run tests to see if it went well
 $ sudo make install
 $ export PYTHONPATH=`pwd`/src      # only if you want python bindings

Building a release version:

 $ cd cuv-version-source
 $ mkdir -p build/release
 $ cd build/release
 $ cmake -DCMAKE_BUILD_TYPE=Release ../../
 $ ccmake .          # adjust paths to your system (cuda, thrust, pyublas, ...)!
                     # turn on/off optional libraries (CImg, ...)
 $ make -j
 $ ctest             # run tests to see if it went well
 $ sudo make install
 $ export PYTHONPATH=`pwd`/src      # only if you want python bindings

On Debian/Ubuntu systems, you can skip the sudo make install step and instead

 $ cpack -G DEB
 $ sudo dpkg -i cuv-VERSION.deb

Building the documentation

 $ cd build/debug    # change to the build directory
 $ make doc

Sample Code

We show two brief examples. For further inspiration, please take a look at the
test cases implemented in the src/tests directory.

Pushing and pulling of memory

C++ Code:

 #include <cuv.hpp>
     using namespace cuv;

     int main(void){
         tensor<float,host_memory_space> h(256);  // reserves space in host memory
         tensor<float,dev_memory_space>  d(256);  // reserves space in device memory

         fill(h,0);                          // terse form
         apply_0ary_functor(h,NF_FILL,0.f);    // more verbose

         d=h;                                // push to device
         sequence(d);                        // fill device vector with a sequence

         h=d;                                // pull to host
         for(int i=0;i<h.size();i++)
             assert(d[i] == h[i]);

Python Code:

 import cuv_python as cp
 import numpy as np

 h = np.zeros((1,256))                                   # create numpy matrix
 d = cp.dev_tensor_float(h)                              # constructs by copying numpy_array

 h2 = np.zeros((1,256)).copy("F")                        # create numpy matrix
 d2 = cp.dev_tensor_float_cm(h2)                         # creates dev_tensor_float_cm (column-major float) object

 cp.fill(d,1)                                            # terse form
 cp.apply_nullary_functor(d,cp.nullary_functor.FILL,1)   # verbose form

 h =                                                # pull and convert to numpy
 assert(np.sum(h) == 256)
 d.dealloc()                                             # explicitly deallocate memory (optional)

Simple Matrix operations


 #include <cuv.hpp>
 using namespace cuv;

 int main(void){
     tensor<float,dev_memory_space,column_major> C(2048,2048),A(2048,2048),B(2048,2048);

     fill(C,0);         // initialize to some defined value, not strictly necessary here

     apply_binary_functor(A,B,BF_MULT);  // elementwise multiplication
     A *= B;                             // operators also work (elementwise)
     prod(C,A,B, 'n','t');               // matrix multiplication

Python Code

 import cuv_python as cp
 import numpy as np
 C = cp.dev_tensor_float_cm([2048,2048])   # column major tensor
 A = cp.dev_tensor_float_cm([2048,2048])
 B = cp.dev_tensor_float_cm([2048,2048])
 cp.fill(C,0)                       # fill with some defined values, not really necessary here
 cp.apply_binary_functor(B,A,cp.binary_functor.MULT) # elementwise multiplication
 B *= A                                              # operators also work (elementwise),A,B,'n','t')                              # matrix multiplication

The examples can be found in the "examples/" folder under "python" and "cpp"

Generated on Mon Jul 4 2011 12:04:53 for CUV by  doxygen 1.7.1


Matrix library for CUDA in C++ and Python






No packages published