perf(ecs): per-frame copies, allocations and WASM crossings in run_systems and the zippers

未關閉
#493 0 則留言 0 個 reaction 已指派 0 人 在 GitHub 檢視

維護者通常 1 天內回覆

還沒有人認領這個 Issue。

評估

難度
4/5
預估耗時
3-5 天
新手友好度
42/100
Issue 類型
功能
描述清晰度
基本清楚
活躍度
活躍
技術堆疊
cpp, wasm

研究方向

Read modules/ecs/wasm/Registry.hpp around lines 260–262 and 302–322 to trace run_systems and getZipper. First establish an e2e game performance measurement, since the issue says none has been taken. Compare the per-frame allocations and WASM crossings before and after the proposed caching and length-read changes; done means measured improvement without changing ECS behavior.

由索引模型根據 Issue 內容生成。

描述

enhancement modules:ecs
Which package or utils is this feature request for?

ecs

Feature

The hottest path breaks the repo's rule against per-frame allocation, at a cost of entities × systems. Every frame copies the systems vector and creates a shared_from_this() JS wrapper per system. Each getZipper builds a std::map, a JS array and one JS object per matching entity, calls get_js_class_name twice per component (3–6 JS property reads each) and re-reads comps["length"] on each iteration. No measurement was taken yet.

Where:

Evidence:

std::vector<emscripten::val> systems_copy = _systems;
for (emscripten::val &system : systems_copy)
    system(shared_from_this(), ctx);
Ideal solution or implementation

Copy the systems only when they change during a run, cache the class-name → array pointer lookup per constructor, and read length once. Measure before and after with the e2e game.

Alternative solutions or implementations

No response

Other context

Found during a code audit of main at be72076.

主要語言
TypeScript
星號
7
分支
2
平均合併
1 天 17 小時
30 天內合併 PR
17

環境準備

從這裡開始

  1. 先讀完整個 Issue,再讀專案的貢獻指南。
  2. 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
  3. Fork 儲存庫,在一個分支上完成修改。
  4. 送出 Pull Request,並在描述裡引用這個 Issue 編號。

NanoForge-dev/Engine 的其他 Issue

查看 NanoForge-dev/Engine 的全部 Issue

相似的 Issue

更多 TypeScript Issue

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。