New pprof.me Features: Profiling Data Comparison and Folded Stacks format support

Visualize differential Icicle graphs from two shared profiles.

December 07, 2023

We are excited to announce two more new features on pprof.me, our free profile data sharing service that has been aiding developers in analyzing profiling data seamlessly. Today, we are introducing "Profiling Data Comparison" and "Sharing of Folded Stacks format" to allow you to share a wider format of profiling and also compare any two uploaded profiles for more insights.

Profiling Data Comparison

Until now, our platform has facilitated the sharing and visualization of individual profiling data. Moving forward, we are excited to announce the introduction of a new capability: the ability to differentially visualize two shared profiles. The comparative analysis of profiling data is essential in any performance optimization endeavour. When addressing performance bottlenecks and implementing fixes, it becomes crucial to validate the impact by comparing the program's profiling data before and after the optimization. This newly added feature precisely addresses this need, providing a valuable tool in your performance debugging journey.

To use this feature, visit https://pprof.me/compare. Enter the profile IDs of the profiles you wish to compare, then submit to initiate the visualization of the profile comparison.

Also these comparisons can be embedded in your own website. For example, the following iframe embeds the comparison of two profiles:

Sharing of Folded Stacks format

We've long had the request to support other profiling format on our sharing platform, and today we're excited to introduce folded stacks. Folded stacks is perhaps the simplest format, but nonetheless very popular.

Quick refresher, folded stacks are just semicolon delimeted frames making up a stack and each line contains one sample.

main;a;b;c 1
main;a;b;d 2
main;a;b;d;e 1

When uploading the above data, the data looks like this.

That's it!

We hope that these new features on pprof.me will help you in your performance debugging journey. We are always looking for feedback and suggestions to improve our platform. If you have any questions or feedback, please let us know on our Discord.

Discuss:
Sign up for the latest Polar Signals news