The majority of this repository is cloned from @bcosenza which offers the SYCL-Bench(mark) suite.
However, this fork provides a Dockerfile --for reproducibility of the 4 SYCL implementations-- and a Jupyter notebook for interpretability and to highlight our methodology around how data is collected, analysed and presented.
We also offer the source-code used in our deep-dive of the matrix multiplication example, implemented in 3 different SYCL parallel execution constructs and a serial baseline [matmul_serial.cpp]; basic kernel parallelism (BKP) [matmul_bkp.cpp], work-group parallelism (WGP) [matmul_wgp.cpp], and hierarchical data-parallelism (HDP) [matmul_hdp.cpp].
The dynamic Jupyter notebook, found in sycl-performance.ipynb, shows how sycl-bench was run and the results plotted. A static webpage of the analysis presented in the paper is found here.
Docker was run with:
docker run --device=/dev/dri --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=1 -it --mount src=`pwd`,target=/workspace,type=bind -p 8888:8888 --net=host --security-opt seccomp=unconfined --group-add video beaujoh/syclbench:latest bash
followed by:
beakerx --allow-root
to start the Jupyter session.
If you have any questions please contact me :)
Beau Johnston
Computer Scientist @ Oak Ridge National Laboratory [email protected]
Visiting Fellow @ Australian National University [email protected]
SYCL Benchmark Suite, work in progress
Benchmarks support the following command line arguments:
--size=<problem-size>- total problem size. For most benchmarks, global range of work items. Default: 3072--local=<local-size>- local size/work group size, if applicable. Not all benchmarks use this. Default: 256--num-runs=<N>- the number of times that the problem should be run, e.g. for averaging runtimes. Default: 5--device=<d>- changes the SYCL device selector that is used. Supported values:cpu,gpu,default. Default:default--output=<output>- Specify where to store the output and how to format. If<output>=stdio, results are printed to standard output. For any other value,<output>is interpreted as a file where the output will be saved in csv format.--verification-begin=<x,y,z>- Specify the start of the 3D range that should be used for verifying results. Note: Most benchmarks do not implement this feature. Default:0,0,0--verification-range=<x,y,z>- Specify the size of the 3D range that should be used for verifying results. Note: Most benchmarks do not implement this feature. Default:1,1,1--no-verification- disable verification entirely--no-ndrange-kernels- do not run kernels based on ndrange parallel for
If you use SYCL-Bench, please cite the following papers:
@inproceedings{SYCL-Bench:Euro-Par:2020, author = {Lal, Sohan and Alpay, Aksel and Salzmann, Philip and Cosenza, Biagio and Hirsch, Alexander and Stawinoga, Nicolai and Thoman, Peter and Fahringer, Thomas and Heuveline, Vincent}, title = {{SYCL-Bench: A Versatile Single-Source Benchmark Suite for Heterogeneous Computing}}, year = {2020}, publisher = {Springer International Publishing}, booktitle = {Accepted for publication at Euro-Par 2020: 26th International European Conference on Parallel and Distributed Computing}, series = {Euro-Par ’20} }
@inproceedings{SYCL-Bench:IWOCL:2020, author = {Lal, Sohan and Alpay, Aksel and Salzmann, Philip and Cosenza, Biagio and Stawinoga, Nicolai and Thoman, Peter and Fahringer, Thomas and Heuveline, Vincent}, title = {{SYCL-Bench: A Versatile Single-Source Benchmark Suite for Heterogeneous Computing}}, year = {2020}, isbn = {9781450375313}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/3388333.3388669}, doi = {10.1145/3388333.3388669}, booktitle = {Proceedings of the International Workshop on OpenCL}, articleno = {10}, numpages = {1}, keywords = {Heterogeneous Computing, SYCL Benchmarks &Runtime}, location = {Munich, Germany}, series = {IWOCL ’20} }