This work provids thread level pipelining of MATLAB functions as well as general concurrency. Resulting in significant speedup of current MATLAB based SDR implementations on multicore systems, or other HPC based applications. Our lab has demonstrated by a 20x speed increase for a 802.11a receiver, running in real-time.
Required Packages/Software
- Python2
- libboost1.54-all-dev
- cmake (For testing only)
- CppHeaderParser
sudo pip install CppHeaderParser - MATLAB R2014a+ (Have not tested with previous versions)
Note I have only tested this on linux, but should work on other platforms assuming you can meet the package requirements.
Acknowledgements
- Thanks to Cameron Desrochers nonblocking queue code
readerwriterqueue. It has some speed advantage overboost::lockfree::spsc_queue