Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
latency_checker: a latency benchmark for multi-core multi-node systems Copyright (C) 2012 Sébastien Boisvert This benchmark tool measures the round trip latency. It is useful to benchmark the latency of multi-core multi-node systems where each node has 1 active network device and where all the processor cores of a node heavily utilize the same network device. The code is in ansi C (ISO C 1990) and utilizes MPI (message passing interface). Compile it with make And use it with mpiexec -n 4 -output-filename Result latency_checker You will get one file per process (Result.1.0, Result.1.1, Result.1.2 and Result.1.3). The option -output-filename is specific to Open-MPI, you may need to change that if you use another MPI library/runtime. Each process will perform 100 000 exchanges. Each message has 4 000 bytes. A exchange is done when process A sends a message to process B followed by process B replying to process A. The round trip latency is the elapsed time required by this process. The number of exchanges can be changed with -exchanges 100000 and the message size can be changed with -message-size 4000 . Written by Sébastien Boisvert 2012-09-09 This benchmark is distributed with the GNU General Public License, version 3.