This blog entry was contributed to by Ruud van der Pas, who works in the Oracle Linux Toolchain Team and works on gprofng on a daily basis.
Gprofng is the next-generation performance profiling tool that does not require recompilation of the code. As Oracle continues to bring new and important tools to support developers, this GNU binutils gprofng tool provides key insight into to the performance of applications written in C, C++, Java, Scala, or Fortran. Multithreading programming models are also supported by gprofng, such as Pthreads, OpenMP, and Java threads.
Learn how to use this important new tool with this free training video.
After watching this video, you will learn how to use gprofng to profile a simple program and analyze the results. The video shows how gprofng runs your application, collecting performance data into a results directory.
The video then walks you through how gprofng is used to analyze the collected performance data to identify bottlenecks and detect where the application spends its time. You will learn a range of options for controlling and customizing how the data is viewed. Typical gprofng command options used to view the performance data collected are shown, including how to create an html format for the results which can be viewed in a browser.
Sampling for data collection
An important difference between gprofng and other profiling tools is that it uses Program Counter (PC) sampling to collect data while the application is running. The application is halted at regular intervals and key information is stored with operational data to create a profile of the performance while it is executing. The advantage of this approach is that existing executables can be profiled without the need to recompile the code. To manage the reporting accuracy the sampling frequency is configurable.
Read more about gprofng:
For more free videos on Oracle Linux, check out this Oracle Linux training track which also includes an extensive list of free hands-on labs.