CLaxon is a set of OpenCL benchmarks testing the performance of various massive
parallel algorithms relevant for safety-critical systems. The aim of this
project is to aid the evaluation of OpenCL-capable architectures. This could be
in the context of architectural research or in the developments of (part of) an
OpenCL compiler. Optionally, these benchmarks can be useful for (automated)
regression testing of compilers or GPU simulators.
Characteristic for CLaxon is that kernels will always run with the same input
data, and are thus expected to roughly produce the same output data. No
validation code is included currently, but patches welcome.
It is _not_ my aim to establish a new benchmark suite used for performance
evaluation of commercial architectures. Those who wish to evaluate such
GPU architectures comparatively are encouraged to run SPEC Accel instead.
This work represents the collective work of Rodinia[1], Parboil[2],
KinectFusion[3] and my own.
Requirements:
- CMake >= 2.8.10
- OpenCL >= 1.1
- (Optional) ninja-build (or ninja on Debian-derivatives)
On Fedora, I recommend installing the following packages to satisfy the OpenCL
dependencies:
- libopencl-headers
- ocl-icd
- ocl-icd-devel
Build instructions:
- cmake -G Ninja .
- ninja
Acknowledgements:
data/frnn/frnn_stanbun_000.txt: a projection of the Stanford bunny
pointcloud, courtesy of Stanford University Computer Graphics Laboratory.
[1] S. Che, M. Boyer, J. Meng, D. Tarjan, J. W. Sheaffer, Sang-Ha Lee and
K. Skadron. "Rodinia: A Benchmark Suite for Heterogeneous Computing". IEEE
International Symposium on Workload Characterization, Oct 2009.
[2] “Parboil benchmark suite,” 2010,
http://impact.crhc.illinois.edu/Parboil/parboil.aspx.
[3] R. A. Newcombe, S. Izadi, O. Hilliges, D. Molyneaux, D. Kim, A. J. Davison,
P. Kohi, J. Shotton, S. Hodges, and A. Fitzgibbon, “Kinectfusion: Real-time
dense surface mapping and tracking,” in 2011 10th IEEE International
Symposium on Mixed and Augmented Reality, Oct 2011.