檢查效能
透過集合功能整理內容
你可以依據偏好儲存及分類內容。
檢查效能有助於瞭解應用程式的運作情形，確保運作方式符合期望。
Android 提供多種可以檢查應用程式效能的工具。一開始，建議您一次專注處理一個層面。包括下列項目：
- 應用程式啟動
- 轉譯速度緩慢 (卡頓)
- 畫面轉換和導覽事件
- 長時間執行工作
- 背景作業，例如 I/O 和網路
另外，您也可以檢查應用程式工作流程的關鍵使用者旅程，這有助於全面瞭解效能不符預期的地方。
檢查效能的方法主要分為兩種：手動和自動化。檢查新區域時，您很可能會先從手動偵錯開始。
手動檢查
決定應用程式要檢查的層面後，就可以運用多種工具辨別實際情況。
檢查搭載 Android 9 以上版本的裝置時，最全面的工具是 Perfetto。Perfetto 可以提供最詳細的追蹤資訊，而您也可以透過強大的篩選器，視需求調整資料精細程度。如要進一步瞭解如何透過 Android 裝置擷取追蹤記錄，請參閱「快速入門導覽課程：在 Android 裝置上擷取追蹤記錄」指南。
Android Studio 內建的 Android 分析器也能提供有關應用程式效能的寶貴深入分析，協助您限制應用程式的資料精細程度，或是在 Android 9 以下版本裝置上執行。
詳情請參閱「系統追蹤總覽」，或觀看「效能偵錯」系列影片。
自動化測試
除了手動檢查之外，您也可以設定自動化測試，收集和匯總效能資料。這有助於瞭解使用者實際看到的畫面，並找出可能發生迴歸的時間點。如果想進一步瞭解如何為應用程式設定自動化效能測試，請參閱「為應用程式進行基準測試」。
您可以透過多種工具檢查及監控效能，進而改善應用程式。
透過基準測試程式庫瞭解本機效能
瞭解在實際工作環境中的效能表現
使用 Android Studio 在本機剖析
進階剖析工具：Perfetto 追蹤功能
其他資源
這個頁面中的內容和程式碼範例均受《內容授權》中的授權所規範。Java 與 OpenJDK 是 Oracle 和/或其關係企業的商標或註冊商標。
上次更新時間：2025-07-27 (世界標準時間)。
[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["缺少我需要的資訊","missingTheInformationINeed","thumb-down"],["過於複雜/步驟過多","tooComplicatedTooManySteps","thumb-down"],["過時","outOfDate","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["示例/程式碼問題","samplesCodeIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間：2025-07-27 (世界標準時間)。"],[],[],null,["# Inspect performance to help you understand what is happening in your app and\nensure it meets your expectations.\n\nAndroid provides several tools you can use to inspect your app's performance.\nWhen getting started, we recommend you focus on one area at a time during\ninspection. These areas can include the following:\n\n- App startup\n- Slow rendering (jank)\n- Screen transitions and navigation events\n- Long running work\n- Operations in the background, such as I/O and networking\n\nAlternatively, you can inspect critical user journeys of your app's workflow.\nThis can help you gain a holistic understanding of where performance and\nexpectations don't align.\n\nThere are two main approaches when inspecting performance, manual and automated.\nIt's likely that you start with manual debugging when inspecting a new area.\n\nManual inspection\n-----------------\n\nAfter deciding which area of your app to inspect, you can use a variety of tools\nto identify what exactly is happening.\n\nThe most comprehensive tool to inspect performance on devices running Android 9\nand higher is [Perfetto](https://perfetto.dev/). Perfetto provides the highest possible\ndetail of tracing information. By using powerful filters, you can adjust the\nlevel of detail for your needs. For more information about how to capture traces\nfrom Android devices, see the [Quickstart: Record traces on\nAndroid](https://perfetto.dev/docs/quickstart/android-tracing) guide.\n\nThe [Android profilers](/studio/profile) built into Android Studio can also provide valuable\ninsights into your app's performance, where you can limit the level of detail to\nyour app, or when running on devices earlier than Android\n9.\n\nFor more information, see [Overview of system tracing](/topic/performance/tracing) or watch the in-depth\nseries on [performance debugging](https://www.youtube.com/playlist?list=PLWz5rJ2EKKc-xjSI-rWn9SViXivBhQUnp).\n\nAutomated testing\n-----------------\n\nIn addition to manual inspection, you can set up automated tests to collect and\naggregate performance data. This helps you understand what users are actually\nseeing and identify when regressions might occur. For more information about\nsetting up automated performance tests for your app, see [Benchmark your\napp](/topic/performance/benchmarking/benchmarking-overview).\n\nApp startup performance\n-----------------------\n\nThere are multiple tools you can use to inspect and monitor performance to help\nimprove your app.\n\n### Understand performance locally with Benchmark libraries\n\n- The [Macrobenchmark library](/topic/performance/benchmarking/macrobenchmark-overview) helps you measure larger end-user interactions, such as startup, interacting with the UI, and animations.\n- The [Microbenchmark library](/topic/performance/benchmarking/microbenchmark-overview) helps analyze performance of more granular, app-specific situations.\n\n### Understand performance in production\n\n- [Android vitals](/topic/performance/vitals) can help improve your app's performance by alerting you when various performance metrics exceed predetermined thresholds.\n- The [Firebase performance SDK](https://firebase.google.com/docs/perf-mon/get-started-android) collects various metrics about your app's performance. For example, you can use the SDK to measure the time between when the user opens the app and when the app becomes responsive, helping identify potential startup bottlenecks.\n\n### Profile locally with Android Studio\n\n- Use [Android Studio](/studio/profile) to record and view system traces or stack sampling traces.\n- [Record traces](/studio/profile/record-traces) using Android Studio. For additional information, see the [Performance Debugging video series](https://www.youtube.com/playlist?list=PLWz5rJ2EKKc-xjSI-rWn9SViXivBhQUnp).\n- Use [Simpleperf](https://android.googlesource.com/platform/system/extras/+/master/simpleperf/doc/README.md), a native stack sampling tool for Android, to profile both Android apps and native processes running on Android. It can profile both Java and C++ code on Android.\n\n### Advanced profiling tools: Perfetto tracing\n\n- [Perfetto](https://perfetto.dev/): a platform-wide tracing tool available on Android 10 (API level 29) and higher. For more information, see the [overview of Perfetto traces](https://perfetto.dev/docs/).\n- [Run Perfetto using `adb`](/studio/command-line/perfetto): describes how to run the `perfetto` command-line tool to capture traces.\n- [Recording a trace through the cmdline](https://perfetto.dev/docs/quickstart/android-tracing#recording-a-trace-through-the-cmdline): describes how to build and run the `perfetto` command-line tool to capture traces.\n- [Perfetto web-based trace viewer](https://perfetto.dev/docs/quickstart/android-tracing#recording-a-trace-through-the-perfetto-ui): opens Perfetto traces and displays a complete report. You can also open [Systrace](/topic/performance/tracing) traces in this viewer using the legacy UI option.\n\nAdditional resources\n--------------------\n\n- [Performance debugging - MAD skills series](https://www.youtube.com/playlist?list=PLWz5rJ2EKKc-xjSI-rWn9SViXivBhQUnp)\n- [Profile your app performance](/studio/profile)\n- [Write a Macrobenchmark](/topic/performance/benchmarking/macrobenchmark-overview)\n- [Microbenchmark](/topic/performance/benchmarking/microbenchmark-overview)"]]
