Welcome to the world of CIBENCH!
Here you can enjoy the collection of 12 benchmarks and their optimization methods, they are:
- Bzip2-1.0.6
- Backprop,
- Hotspot3D,
- LavaMD,
- Srad_v2 from Rodinia Benchmark Suite
- FFT from GNU scientific library
- msb(msgrate) from NERSC8 Trinity Benchmarks
- USQCD Chroma
- Hmmer,
- H264ref,
- Povray from SPEC CPU2006 Benchmark Suite
- Hoard
Above benchmarks (besides FFT and Chroma) can be compiled individually in their folders, or you can compile them together by following commands:
To compile the original benchmarks:
$ make org
To compile the optimized benchmarks:
$ make opt
You can easily adjust your compiler preferences by modifying
To install USQCD Chroma and GSL, please following the guidance inside corresponding folders.
Due to the copyright of SPEC Benchmark Suite, we shall not release the code of Hmmer, H264ref and Povray here, please refer to our paper "What Every Scientific Programmer Should Know AboutCompiler Optimizations?" for the detailed optimization methods.
We list the inputs we used for your references:
For hmmer:
$hmmer: ./hmmer nph3.hmm swiss41
For h264ref:
$h264ref: ./h264ref -d foreman_ref_encoder_baseline.cfg
For povray:
$povray: ./povray SPEC-benchmark-ref.ini
If you have insights or questions towards CIBENCH please send email to [email protected], thanks!