AI After Hours | Profiling with GitHub Copilot
Share

Post Content

 

 In AI After Hours, we’ll explore tips and tricks, engineering, and customer research behind upcoming experiences in Visual Studio that are enabled by GitHub Copilot.

GitHub Copilot is a coding assistant powered by Artificial Intelligence (AI), which can run in various environments and help you be more efficient in your daily coding tasks.

== Introduction ==
0:00 – Making profiling results actionable with GitHub Copilot.
0:20 – Meet Aaron Yim, Harshada Hole, and Nik Karpinsky.

== Demo ==
0:54 – “Ask Copilot” for top insights from profiling results.
2:40 – “Ask Copilot” to describe and suggest fixes to Top Insights.
4:20 – “Analyze method with Copilot” for methods not in top insights / hot path.

== Discussion ==
5:40 – GitHub Copilot uses line-by-line performance to focus its suggestions; web search and other AI are not.
7:52 – Context package includes source code concatenated with all performance information.
8:40 – More context is not always better.
9:38 – Including caller/callee info could inform architectural improvements in the future.
10:15 – Team finds that feature is not very disruptive but can be very helpful and informative.
11:30 – “Ask Copilot” for top insights and “Analyze method” not yet available, expected for Visual Studio 2022 version 17.11.
12:30 – Please provide feedback via “Report a Problem” button in the top right of Visual Studio.
13:06 – Success is empowering every developer to use the performance profiler.
15:45 – Team is interested in performance benchmarking to help guard against regressions. Read More Microsoft Visual Studio 

By