Skip to content

Latest commit

 

History

History
66 lines (49 loc) · 2.83 KB

README.md

File metadata and controls

66 lines (49 loc) · 2.83 KB
Raw

MPI Based DistNTF

Install Instructions

This program depends on:

  • Armadillo library for linear algebra operations.
  • OpenBLAS for optimized kernels. If building with OpenBLAS and MKL is discoverable by cmake, use -DCMAKE_IGNORE_MKL=1 while building.

Once the above steps are completed, set the following environment variables.

export ARMADILLO_INCLUDE_DIR=/home/rnu/libraries/armadillo-6.600.5/include/
export LIB=$LIB:/home/rnu/libraries/openblas/lib:
export INCLUDE=$INCLUDE:/home/rnu/libraries/openblas/include:$ARMADILLO_INCLUDE_DIR:
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/rnu/libraries/openblas/lib/:
export NMFLIB_DIR=/ccs/home/ramki/rhea/research/nmflib/
export INCLUDE=$INCLUDE:$ARMADILLO_INCLUDE_DIR:$NMFLIB_DIR
export CPATH=$CPATH:$INCLUDE:
export MKLROOT=/ccs/compilers/intel/rh6-x86_64/16.0.0/mkl/

If you have got MKL, please source mklvars.sh before running make/cmake

  • Create a build directory.
  • Change to the build directory
  • In case of MKL, source the $MKLROOT/bin/mkl_vars.sh intel64
  • run cmake [PATH TO THE CMakeList.txt].
  • make

Refer to the build scripts found in the build directory for a sample script.

Input types

PLANC only supports dense input tensors.

Other Macros

cmake macros 2. -DCMAKE_BUILD_TYPE=Debug: For creating debug builds. 3. -DCMAKE_WITH_BARRIER_TIMING: Default is set to barrier timing. To disable build with -DCMAKE_WITH_BARRIER_TIMING:BOOL=OFF. 4. -DCMAKE_BUILD_CUDA: Default is off. 5. -DCMAKE_IGNORE_MKL: Ignores MKL when discoverable by cmake. If using OpenBLAS, set -DCMAKE_IGNORE_MKL=1 while building.

Code level macros - Defined in distutils.h

  1. MPI_VERBOSE - Be doubly sure about what you do. It prints all intermediary matrices. Try this only for very very small tensors that is of size less than 10.
  2. WRITE_RAND_INPUT - for dumping the generated random tensor.

IO options

PLANC uses MPI-IO to read in dense input tensors.

PLANC uses MPI-IO for outputting the factor matrices into a single file per factor matrix.

Command Line options

For a full list of command line options please run distntf -h. Some common options are listed below.

  • -p Sets the processor grid dimensions.
  • -k Sets the low-rank parameter for the approximation.
  • -d Sets the input dimensions.
  • -e Switch to toggle error calculation. Off by default.
  • -t Maximum number of iterations to run the algorithms for.
  • -a Algorithm to run.
  • -i Input tensor to run on. Can be a synthetic input via options rand_lowrank/rand_uniform/rand_normal or a path to a file.
  • -dimtree Toggles the dimension tree optimization to speedup MTTKRP calculations. On by default.

Citation

Please refer to the papers section for the appropriate reports to cite.