搜索

PythonReactSEOPDF
GitHub Skill

unity-performance-check

Use before committing code or after major changes to verify performance. Catches expensive operations before they cause frame drops.

Sharks820☆ 0 仓库 Stars0 仓库 Forks

安装与使用

请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求,再添加到工作流。

查看来源 ↗使用指南

技能说明

Unity Performance Check

Overview

Verify code performance BEFORE it causes problems in production. Frame drops ruin games.

When to Use

  • Before committing new code
  • After implementing Update/FixedUpdate logic
  • When adding new systems
  • Before major milestones
  • When users report lag

The Quick Check (30 seconds)

Run these searches on changed files:

// RED FLAGS - Fix these immediately:
GetComponent         // In Update? Cache it
Find(                // In Update? Never do this
FindObjectOfType     // Anywhere? Avoid if possible
new List             // In Update? Preallocate
new Dictionary       // In Update? Preallocate
.ToString()          // In Update? Cache or avoid
string +             // In Update? StringBuilder
foreach              // On non-List? May allocate

The Full Check (5 minutes)

Use the unity-performance-profiler agent:

Task: Launch unity-performance-profiler agent with:
"Analyze [file/folder] for performance issues"

Performance Rules

Update Loop (60fps = 16.6ms budget)
Operation Cost Rule
GetComponent ~1ms Cache in Start
Find/FindObjectOfType ~5-50ms Never use
Instantiate ~2-10ms Use object pool
new List/Dictionary ~0.1ms + GC Preallocate
String concat ~0.5ms + GC StringBuilder
LINQ query ~1-5ms + GC Use loops
Physics.Raycast ~0.1ms Batch or limit
Memory (GC spikes cause stutters)
Cause Solution
new in Update Preallocate collections
String operations Cache strings, use StringBuilder
LINQ methods Convert to explicit loops
Boxing (int to object) Use generics
Closures in lambdas Cache delegates
foreach on IEnumerable Use for loops on arrays/lists
VeilBreakers-Specific
System Performance Concern Solution
Combat Damage calc per frame Batch damage events
Brand lookup Switch statement Dictionary lookup
Synergy calc Recalc every frame Cache on party change
UI updates Rebuild every frame Dirty flag pattern

Checklist

Before committing, verify:

  • No GetComponent in Update/FixedUpdate
  • No Find/FindObjectOfType anywhere
  • No allocations in hot paths
  • Collections preallocated
  • Strings not concatenated in loops
  • Events properly unsubscribed
  • Object pooling for spawned objects

If Performance is Bad

  1. Profile first - Don't guess, measure
  2. Find the hotspot - Usually 1-2 methods
  3. Fix the algorithm - O(n²) → O(n) > micro-optimizations
  4. Cache results - Don't recalculate what hasn't changed
  5. Batch operations - Process multiple items together
  6. Use jobs/burst - For heavy math (advanced)

使用示例

查看来源仓库中的使用示例与配置说明。

查看来源 ↗

来源历史

收录时间
2026-01-20
索引更新时间
2026-01-26
Branch
main

收录与同步日期属于本站记录。源码修改历史请查看来源仓库。

GitHub ↗

chat Comments (0)

chat_bubble_outline

No comments yet. Be the first to share your thoughts!