CNTK Beta (Windows+Linux) 2015-12-8

Rating:        Based on 1 rating
Reviewed:  0 reviews
Downloads: 4299
Change Set: c93a1b95562f4ca
Released: Dec 8, 2015
Updated: Dec 9, 2015 by chrisbasoglu
Dev status: Beta Help Icon

Recommended Download

Application CNTK-20151208-Windows-64bit-ACML5.3.1-CUDA7.0.zip
application, 145993K, uploaded Dec 9, 2015 - 3588 downloads

Other Available Downloads

Application CNTK-20151208-Linux-64bit-ACML5.3.1-CUDA7.0.tar.gz
application, 49948K, uploaded Dec 9, 2015 - 202 downloads
Application CNTK-20151208-Windows-64bit-ACML5.3.1-cpu-only.zip
application, 58713K, uploaded Dec 9, 2015 - 364 downloads
Application CNTK-20151208-Linux-64bit-ACML5.3.1-cpu-only.tar.gz
application, 44706K, uploaded Dec 9, 2015 - 145 downloads

Release Notes

This release of Computational Network Toolkit (CNTK) supports both Windows and Linux on 64bit platforms.

We only build release versions after milestones. We encourage users to enlist the source code to get most updated features and bug fixes, and to contribute back to the toolkit.

Windows Version:

Compiled with Microsoft Visual Studio 2013, ACML 5.3.1, and gpu versions with CUDA 7.0.

For GPU machine:
  1. Download and unzip CNTK-20151208-Windows-64bit-ACML5.3.1-CUDA7.0.zip to the folder where you want to install CNTK.
  2. If Visual C++ Redistributable for Visual Studio 2013 is not installed on your computer, install it from http://www.microsoft.com/en-us/download/details.aspx?id=40784.
  3. For the GPU built version, ensure the latest NVIDIA driver is installed for your CUDA-enabled GPU.
  4. You do not need to install the CUDA SDK, though it would be fine if you do so now or in the future.
  5. If you want to run CNTK on multiple GPUs or multiple machines, you will need to install MSMPI. Install the latest Microsoft MS-MPI SDK and runtime from https://msdn.microsoft.com/en-us/library/bb524831(v=vs.85).aspx
  6. Set the environment variable ACML_FMA=0
For CPU-only machine
  1. Download and unzip CNTK-20151208-Windows-64bit-ACML5.3.1-cpu-only.zip.
  2. Set the environment variable ACML_FMA=0
You are ready to run CNTK (cntk.exe)

Linux Version:

For GPU machine:
  1. Download and unzip CNTK-20150415-Linux-64bit-ACML5.3.1-CUDA7.0.tar.gz.
  2. Download and install CUDA for your Linux distro: https://developer.nvidia.com/cuda-downloads?sid=742762
  3. Install the GPU deployment kit from NVIDIA by going to https://developer.nvidia.com/gpu-deployment-kit. Download the .run file, make it executable and run it.
  4. Download and install the latest video driver for your CUDA-enabled GPU: http://www.nvidia.com/download/driverResults.aspx/84721/en-us
  5. Set your cuda path (you can follow: http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/#environment-setup ).
  6. If you want to run CNTK on multiple GPUs or multiple machines, you will need to install OpenMPI 1.10.0. Install OpenMPI 1.10.0 from http://www.open-mpi.org/software/ompi/v1.10/
  7. Set the environment variable ACML_FMA=0
For CPU-only machine
  1. Download and unzip CNTK-20151208-Linux-64bit-ACML5.3.1-cpu-only.tar.gz.
  2. Set the environment variable ACML_FMA=0
You are ready to run CNTK (bin/cntk)

Reviews for this release