CRONO v0.9 : A Benchmark Suite for Multithreaded Graph Algorithms Executing on Futuristic Multicores
This is a pre-release repository containing 10 graph analytic benchmarks. An in-built synthetic graph generater can be included by statically configuring certain benchmarks. These benchmarks can run on real multicore machines as well as the Graphite Multicore Simulator.
If you use these benchmarks, please cite:
CRONO : A Benchmark Suite for Multithreaded Graph Algorithms Executing on Futuristic Multicores, Masab Ahmad, Farrukh Hijaz, Qingchuan Shi, Omer Khan, IEEE International Symposium on Workload Characteriz ation (IISWC), Oct 2015, Atlanta, Georgia, USA.
Paper pdf is located at: http://www.engr.uconn.edu/~omer.khan/pubs/crono-iiswc15.pdf
- Linux (Tested on Ubuntu 14.04)
- g++ 4.6 (Tested with g++ 4.7)
- Ubiquitous graph based search, planning, and clustering algorithms
- Benchmarks use adjacency list representation for input graphs
- Most benchmarks scale to 256 threads, some scale upto 1024 threads as well
- Easy to compile and use
- The input graphs are generic, i.e., the benchmarks do not assume any pre-processing or optimizations, such as graph compression and vertex/edge reordering
Checkout the Repo:
git clone https://github.com/masabahmad/CRONO
To generate the executable for a benchmark, run
make inside the CRONO directory, then execute each benchmark using the syntax specified by the individual README.md.
"Lease/release: architectural support for scaling contended data structures", SK Haider (Microsoft Research), W Hasenplaugh (MIT), D Alistarh (Microsoft Research), PPoPP '16 Proceedings of the 21st ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming.
"Dictionary Sharing: An Efficient Cache Compression Scheme for Compressed Caches", B Panda (INRIA), A Seznec (INRIA/IRISA), MICRO '16, 49th Annual IEEE/ACM International Symposium on Microarchitecture, 2016.