A SpAMM implementation inside a contiguous memory chunk

Introduction

This is an implementation of SpAMM inside a contiguous memory chunk. This chunking allows us to keep a fine grained granularity at the leaf level (much too fine for decent performance if allocated non-contiguously) while improving memory access performance. As a neat side effect, the chunks are already packed and "serialized" and can be migrated as is to other nodes in an MPI/Charm++/... setup.

Build Instructions

Simply run

cmake .
make

There are some tests, that are run with

make test

Current build status

The current status of our test builds on Travis-ci is