Date: July 16, 2025
+Application: CodeOptimizationsSampleApp
+ +
+ A significant CPU performance issue was identified in the application. The issue was located in the ReviewValidation.StringValidation method,
+ specifically in the repeated creation of filtered lists from a large collection of disallowed words. The fix involved implementing a caching mechanism to
+ avoid repeated filtering and list creation operations, resulting in significant CPU usage reduction.
+
+ The Azure MCP Profiler identified a hotspot in the CPU usage profile of the application. The callstack showed that a significant amount of CPU time
+ was being spent in the ToList() method called from ReviewValidation.StringValidation. This method was filtering and creating
+ a new list of disallowed words on every call, despite the source data remaining constant.
+
| Issue | +CPU Impact | +Function | +Parent Function | +
|---|---|---|---|
| High CPU Usage | +25.39% | +Enumerable.ToList | +ReviewValidation.StringValidation | +
+ The StringValidation method was filtering a collection of 30,000 disallowed words on every invocation, creating a new list each time.
+ This was inefficient because:
+
DisallowedWords) remains constant throughout the application lifecycleCultureInfo, which has a limited set of possible valuesBackgroundReviewValidation class+ The solution was to implement a caching mechanism for the filtered word lists. A static dictionary was added to store pre-filtered lists for each + culture that has been encountered. This ensures the filtering and list creation operations are performed only once per culture. +
+ +public static string StringValidation(string data, char replacementChar, CultureInfo culture)
+{
+ List<string> wordList = DisallowedWords
+ .Where(word => culture.Equals(CultureInfo.InvariantCulture) || culture.Equals(word.Culture))
+ .Select(word => word.Text).ToList();
+
+ foreach (string word in wordList)
+ {
+ data = data.Replace(word, replacementChar.ToString(), ignoreCase: true, culture);
+ }
+ return data;
+}// Cache for filtered disallowed word lists by culture to avoid repeated filtering
+private static readonly Dictionary<CultureInfo, List<string>> _cultureWordCache = new Dictionary<CultureInfo, List<string>>();
+
+public static string StringValidation(string data, char replacementChar, CultureInfo culture)
+{
+ // Get or create the filtered word list for the specific culture
+ if (!_cultureWordCache.TryGetValue(culture, out var wordList))
+ {
+ // Filter once and cache the result
+ wordList = DisallowedWords
+ .Where(word => culture.Equals(CultureInfo.InvariantCulture) || culture.Equals(word.Culture))
+ .Select(word => word.Text)
+ .ToList();
+
+ _cultureWordCache[culture] = wordList;
+ }
+
+ foreach (string word in wordList)
+ {
+ data = data.Replace(word, replacementChar.ToString(), ignoreCase: true, culture);
+ }
+ return data;
+}+ The implemented solution should provide the following benefits: +
++ While the current fix addresses the immediate performance bottleneck, here are some additional recommendations for further optimization: +
+Report generated on July 16, 2025
+ + diff --git a/prompts/fix-top-cpu-issue-by-padawan.prompt.md b/prompts/fix-top-cpu-issue-by-padawan.prompt.md new file mode 100644 index 0000000..2e4fcf3 --- /dev/null +++ b/prompts/fix-top-cpu-issue-by-padawan.prompt.md @@ -0,0 +1,21 @@ +--- +mode: 'agent' +tools: ['codebase', 'usages', 'vscodeAPI', 'problems', 'changes', 'testFailure', 'terminalSelection', 'terminalLastCommand', 'openSimpleBrowser', 'fetch', 'findTestFiles', 'searchResults', 'githubRepo', 'extensions', 'runTests', 'editFiles', 'runNotebooks', 'search', 'new', 'runCommands', 'runTasks', 'applicationinsights'] +--- + +You are a developer working on a ASP.NET Core project. The source code is located in the `src` directory of the app. + +The GitHub repo has an agent to work on improving code. It needs as much details for the issue as possible, like issue descriptions, callstacks and so on. + +Can you query the top CPU related code optimization insights, and create a github issue so that it can be fixed by the github coding agent? + +Here are some additional context to help you with the task: + +## Inputs + +Query the insights from the Azure MCP Profiler using the following parameters: + +- Subscription: Service Profiler Development +- App Id: 94adcdae-2d1b-4c68-9c1f-42d876be4c2c +- Time Range: Last 24 hours + diff --git a/prompts/fix-top-cpu-issue.prompt.md b/prompts/fix-top-cpu-issue.prompt.md new file mode 100644 index 0000000..ee7f5fb --- /dev/null +++ b/prompts/fix-top-cpu-issue.prompt.md @@ -0,0 +1,27 @@ +--- +mode: 'agent' +tools: ['editFiles', 'runNotebooks', 'search', 'new', 'runCommands', 'runTasks', 'usages', 'vscodeAPI', 'problems', 'changes', 'testFailure', 'openSimpleBrowser', 'fetch', 'githubRepo', 'extensions', 'runTests', 'applicationinsights'] +--- + +You are a developer working on a ASP.NET Core project. The source code is located in the `src` directory of the app. + +You are tasked to figure out the top CPU usage issue in the app and fix it. + +You will find the code optimization insights in the Azure MCP applicationinsights codeoptimizations. You will use the `applicationinsights` tool to list all code optimization insights for the app. + +Once you have the insights, you will analyze the code and identify the top CPU usage issue. You will then use the `editFiles` tool to make the necessary changes to the code. Note that the code change should based on the callstack provided in the insights. + +Also, concentrate on the performance fix. Do not change the functionality of the code, and do not try to fix any other issues that are not related to CPU usage. + +Generate a report at the end of the task and save it to a file named `cpu-issue-report.html` in the root directory of the app. + +Here are some additional context to help you with the task: + +## Inputs + +Query the insights from the Azure MCP Profiler using the following parameters: + +- Subscription: Service Profiler Development +- Resource Group: eshop-demo +- Time Range: Last 24 hours + diff --git a/prompts/list-codeopt.prompt.md b/prompts/list-codeopt.prompt.md new file mode 100644 index 0000000..fd1130e --- /dev/null +++ b/prompts/list-codeopt.prompt.md @@ -0,0 +1,20 @@ +--- +mode: 'agent' +tools: ['changes', 'codebase', 'editFiles', 'extensions', 'fetch', 'findTestFiles', 'githubRepo', 'new', 'openSimpleBrowser', 'problems', 'readCellOutput', 'runCommands', 'runNotebooks', 'runTasks', 'runTests', 'search', 'searchResults', 'terminalLastCommand', 'terminalSelection', 'testFailure', 'updateUserPreferences', 'usages', 'vscodeAPI', 'Azure MCP Debug Build v0.0.2', 'azmcp-profiler-list-insights'] +--- + +Read through the following instructions thoroughly first. + +## Inputs + +Subscription: Service Profiler Development +App Id: 94adcdae-2d1b-4c68-9c1f-42d876be4c2c +Time Range: Last 24 hours + +## Actions + +List all my code optimization insights for the app. + +## Notes + +The source code in the callstack locates in the `src` directory of the app. diff --git a/src/Reviews/ReviewValidation.cs b/src/Reviews/ReviewValidation.cs index 18e3a01..0226a21 100644 --- a/src/Reviews/ReviewValidation.cs +++ b/src/Reviews/ReviewValidation.cs @@ -8,11 +8,22 @@ internal record LocalizedWord(string Text, CultureInfo Culture); private static IEnumerable