Create the repo of UVM benchmarks

Benchmarks

Traditional Benchmarks

1. BFS

   Modify Rodinia BFS dataset Generation program accordingly. Current generated data can be used by our BFS directly.
   BFS_Original

2. Add Rodinia UVM version

   Several benchmarks will be modified later Rodinia Original

Machine Learning Benchmarks

1. Bayesian Network learning algorithm

   Modify the original code and add data_generation code for generating new numbers of node from the given data file. BN_Original

2. Kmeans

   Modify the original code. Kmeans_Original

3. Logistic Regression

   Modify the orginal code. Logistic Regression Original

4. CNN

   Modify the original code. CNN Original

5. KNN

   Modify the original code. KNN_Original

6. SVM

   Modify the original code. SVM_Original

If you publish using this software or a modified version of it, we would appreciate your citing the following paper:

Gu, Yongbin, et al. “UVMBench: A Comprehensive Benchmark Suite for Researching Unified Virtual Memory in GPUs.” ArXiv:2007.09822 [Cs], July 2020. arXiv.org, http://arxiv.org/abs/2007.09822.