In order to assure software performance, software needs to be tested accordingly as early as possible - only weaknesses diagnosed early can be assessed quickly and cheaply. ContiPerf enables performance testing already in early development phases and in an easy-to-learn manner: A developer writes a performance test in form of a JUnit 4 test case and adds performance test execution settings as well as performance requirements in form of Java annotations. When JUnit is invoked by an IDE, build script or build server, ContiPerf activates, performs the tests and creates an HTML report. The report provides a detailed overview of execution, requirements and measurements, even providing a latency distribution chart. A large feature set for execution settings and performance requirements is available, e.g. Ramp up, warm up, individual pause timing, concurrent exection of test groups and more.
This is a simple, modular PHP class designed to record execution time benchmarks throughout different parts of a script. Compatible with virtually any PHP script, the code can be added/removed quickly and with ease. Accurate to the microsecond.
C++ Benchmarking Library
This project is now maintained at GitHub: https://github.com/DigitalInBlue/Celero Celero is a cross-platform open source C++ benchmarking library. Written with an API similar to Google Test, Celero utilizes C++11 features, offers tested benchmarking capabilities with microsecond precision, baselining, fixtures, and easy test setup. Use the baseline feature to measure/benchmark algorithm performance against a known case to more accurately rate performance of solutions.
Java and C Comparative Performance Benchmarks
A collection of software benchmarks developed to compare the performance of Java with C on identical code. No language libraries were used to avoid implementation differences. Some of the benchmarks are also implemented in Python and Scala. There are benchmarks for bit twiddling, numerical computing, data structure manipulation, concurrent computing, callouts to native libraries, and, graphics processing units (GPU) utilization.
SarBox helps user to parse or analyze sar data
The software program (SarBox) parses and analyzes SAR data in charts or text formats. SAR is cumulative activity counters in the operating system. It is the most comprehensive command which helps to monitor performance of various Linux / Solaris subsystems (CPU, Memory, I/O, Disk, Network, etc) all in same log. This makes it difficult to analyze since every counter data is dumped into the log file. SarBox helps to analyze this log file and present data in graphical charts or text format. To collect sar statistics from an environment, use below syntax [user@host ~]# sar -A 30 2 > sar_linux.log “30 2” reports for every 30 seconds and total of 2 times, written in sar_linux.log.
ViennaProfiler allows to compare execution times of full programs or parts of a program. Execution times are stored in a MySQL database and conveniently analyzed via a web-interface. A C++ client API is provided, other languages can also be used.
The goal of this project is to develop a framework to test and measure the performance of IEEE 802.11 based devices. Currently we aims to address the issues related to protocol conformance testing for a device.
A free Fortran 2008, 2018 coarrays course with notes and exercises
Coarrays are native Fortran means for SPMD parallel programming. At runtime multiple copies of the executable (called images) are executing asynchronously. The F2008 standard provides coarray syntax, remote calls, coarray data objects, allocatable coarrays, syncronisation, atomics, etc. F2018 adds new functionality, e.g. collectives, teams, events, more atomics. Coarrays are used extensively in CASUP library for HPC (https://cgpack.sourceforge.io). The course is aimed at experienced Fortran users. Working knowledge of Fortran is assumed. The course is designed for a 1 day study with an instructor. Multiple exercises with full solutions are accompanied by notes. This course is taught to the University of Bristol HPC users. Exercises have been verified with Intel and GCC/OpenCoarrays. All course materials are released under BSD license. We welcome contributions, provided you are happy to release your work under BSD license. We welcome comments and problem reports.
lsperf is an utility listing performance of various computer components.