The VGOGTE Parallel Benchmarks in C++

There is no official repository available.

Implementations

PTHREADS - This directory contains the parallel Pthread version of the benchmarks.

ARGO - This directory contains the parallel ArgoDSM/Pthread version of the benchmarks.

Benchmarks

Concurrent queue (CQ) - Insert/Delete nodes in a queue
Array Swap (SPS)      - Random swap of array elements
Persistent Cache (PC) - Update entries in persistent hash table
RB-tree               - Insert/Delete nodes in RB-Tree
TATP                  - Update location trans. from TATP
TPCC                  - New Order trans. from TPCC

Software Requirements

ARGO - Assumes you have already installed argodsm.

Building

Enter the directory from the version desired and execute:

$ cd <benchmark_name>
$ make

Executing

Run each benchmark:

$ (mpirun $OMPIFLAGS) ./binary_name

Each benchmark can also be configured with the number of threads (NUM_THREADS), number of operations per thread (NUM_OPS), etc.. The benchmarks also have specific parameters, such as number of sub operations per operation, number of sub elements per datum, etc.

References

Please cite the following works if you use these benchmarks:

1. Vaibhav Gogte, Stephan Diestelhorst, William Wang, Satish Narayanasamy, Peter M. Chen, and Thomas F. Wenisch. 2018. Persistency for synchronization-free regions. In Proceedings of the 39th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI 2018). ACM, New York, NY, USA, 46-61. DOI: https://doi.org/10.1145/3192366.3192367

2. Aasheesh Kolli, Vaibhav Gogte, Ali Saidi, Stephan Diestelhorst, Peter M. Chen, Satish Narayanasamy, and Thomas F. Wenisch. 2017. Language-level persistency. In Proceedings of the 44th Annual International Symposium on Computer Architecture (ISCA '17). ACM, New York, NY, USA, 481-493. DOI: https://doi.org/10.1145/3079856.3080229