Copyright © 2018 Jiang Xu

COSMIC


Heterogeneous Multiprocessor Benchmark Suite

Heterogeneous multiprocessor/multicore/manycore systems are promising alternatives to traditional homogeneous multiprocessor systems and can potentially offer better tradeoffs among energy efficiency, performance, flexibility, scalability, and cost. However exploring their heterogeneities is not well supported by traditional benchmarks which rely on compliers and operating systems developed for homogenous multiprocessor systems. To overcome these issues, this project developed a heterogeneous multiprocessor benchmark suite called COSMIC (Communication-Observant Schedulable Memory-Inclusive Computation).

Project Member: Rafael Kioji Vivas Maeda, Peng Yang, Bin Li (Intel), Ravi Iyer (Intel), Ramesh Illikkal (Intel), Jiang Xu, Haoran Li, Zhongyuan Tian, Zhehui Wang, Zhifei Wang, Xuanqi Chen, Jun Feng

Maintained by: Rafael Kioji Vivas Maeda (rkvivasmaeda@connect.ust.hk). If there is any inquiry or suggestion, please do not hesitate to contact us.

Past Member: Zhe Wang, Weichen Liu, Xiaowen Wu, Xuan Wang, Yaoyao Ye, Mahdi Nikdast, Luan H.K. Duong

OVERVIEW

COSMIC is systematically developed from the algorithms of typical multiprocessor applications. Besides implementing the applications in high-level programming languages, COSMIC uses a formal computational model TCG to explicitly capture the computation and communication requirements of multiprocessor applications. It provides a tool to automatically allocate memory and partition and schedule tasks based on user-defined methods.

COSMIC heterogeneous multiprocessor benchmark suite currently includes the following main components and set of applications.

Application Description
Cifar-train Training phase of a Deep Convolutional Neural Network for the Cifar-10 dateset.
Facerec-train Training phase of a Deep Convolutional Neural Network for face recognition.
AlexNet-inf Interfence phase of the DCNN with 5 convolutional layers.
Machine Learning - ALIP Machine learning based image indexing.
Machine Learning - FMP Financial market prediction using machine learning.
HPCG Conjugate gradient algorithm. Generates a linear system of a three-dimensional heat diffusion problem.
Pennant Lagrangian staggered-grid hydrodynamics algorithm on 2-D unstructured finite-volume mesh.
Snap Performance modelling of a modern discrete ordinates neutral particle transport application.
Stream Synthetic benchmark measuring the memory bandwidth and a corresponding computation rate for four simple vector kernels.
Molecular Dynamics Simulating molecular dynamics when molecules are shot to surfaces of solid atoms. Developed with Prof. Wenjing Ye.
Ray Tracing 3D scenes rendering. Developed with Prof. Pedro Sander.
Ultrasound Medical diagnostics using 2D/3D ultrasound imaging. Developed with Prof. Weichuan Yu.
LDPC Low-density parity-check code encoder. Developed with Prof. Wai Ho Mow.
TURBO Turbo code decoder. Developed with Prof. Wai Ho Mow.
Reed-Solomon Reed-Solomon code encoder and decoder.
FFT Fast Fourier Transform with complex number inputs.

Download User Manual

Or you can download the full package directly.

Download COSMIC 3.0 (14GB)

AGREEMENT

COSMIC is made openly available under the following license. Please cite the following paper if it is used.

  • Rafael Kioji Vivas Maeda, Qiong Cai, Jiang Xu, Zhe Wang, Zhongyuan Tian, “Fast and Accurate Exploration of Multi-Level Caches Using Hierarchical Reuse Distance,” in Proceedings of IEEE Symposium on High Performance Computer Architecture (HPCA), Austin, USA, February 2017.

COPYRIGHT AND LICENSE

Copyright © 2007-2018 The Hong Kong University of Science and Technology
All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met.

  • Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
  • Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
  • Neither the name of the Hong Kong University of Science and Technology nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE HONG KONG UNIVERSITY OF SCIENCE AND TECHNOLOGY ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE HONG KONG UNIVERSITY OF SCIENCE AND TECHNOLOGY BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.