A problem-independent scheme of accelerating the ADMM that does not require the user to write any parallel code
C Cuda Python C++
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
core
packing
.gitignore
README.md
SConstruct
cuda.py

README.md

parADMM Engine

This beta release is associated with Testing fine-grained parallelism for the ADMM on a factor-graph (Hao, Oghbaee, Rostami, Derbinsky, Bento).

The code includes a demo of how to use parADMM for a GPU or multiple cpu cores that is based on the circle packing example found in the numerical section of the paper above.

To compile, run: scons

Available flags:

  • --verbose shows debug output (including timing data)
  • --openmp compiles with OpenMP support
  • --cuda=/path/to/cuda compiles with CUDA support

To run:

  • out/packing (cpu and OpenMP)
  • out/packing-gpu (gpu)