docs(10_Wiki): 위키 전체 재구성 — Topic_* 폴더를 4개 카테고리로 통합 + 대규모 중복 제거

Topic_Agent/Topic_Blog/Topics/Topics_Biz/Topics_Meeting/Topics_Rag의 마크다운 지식 문서를 Topic_General/Topic_Programming/Topic_Graphic/Topic_Business 4개 카테고리로 재분류. - 중복 제거: frontmatter의 status:duplicate/merged + duplicate_of/redirect_to 필드로 자기 자신을 중복으로 선언한 리다이렉트 stub 1032개 제거, 완전 동일 내용 파일 472개 제거, 동일 파일명·다른 내용 충돌 시 더 큰(완전한) 버전만 유지(162개 제거) — 총 1639개 중복 제거. - 분류: 폴더 단위로 명확한 항목(AI_and_ML/Coding/Architecture 등 → Programming, Comfyui/Visual_Effects → Graphic, Topics_Biz/Topics_Meeting/사업 등 → Business, Poetic_Blog_Writing/창의성/Game_Design 등 → General)은 폴더 우선순위로, 나머지 혼재 폴더(Topic_Agent/Topic_Blog/Topics 루트/Thinking & Reasoning/Other/UI_UX_Assets)는 title/tags 키워드 스코어링으로 파일 단위 분류(불명확한 경우 General로 폴백). 원본 폴더명은 "From_*" 서브폴더로 보존해 추적 가능성 유지. - 최종 배치: Programming 2784 / General 1608 / Graphic 285 / Business 249 = 4926개 문서. - 에이전트 운영 상태(.astra/.agent/.obsidian/sessions/memory/_company/docs/lessons/_shared/src)는 지식 콘텐츠가 아니므로 재분류 대상에서 제외하고 원위치 유지. - Topics/Topic_email(상위 보호 폴더 Topic_email과 파일명 100% 중복) 삭제 — 보호 폴더 자체는 미변경. - 완전히 비게 된 Topic_Agent/Topic_Blog/Topics_Biz/Topics_Rag 폴더 제거.
2026-07-05 00:33:48 +09:00
parent 1cfd3bbb56
commit 9148c358d0
6455 changed files with 1 additions and 86875 deletions
@@ -0,0 +1,190 @@
+---
+id: wiki-2026-0508-buffer-allocation
+title: Buffer Allocation
+category: 10_Wiki/Topics
+status: verified
+canonical_id: self
+aliases: [ArrayBuffer Allocation, Typed Array Pool]
+duplicate_of: none
+source_trust_level: A
+confidence_score: 0.9
+verification_status: applied
+tags: [performance, memory, buffer, typed-array, gc]
+raw_sources: []
+last_reinforced: 2026-05-10
+github_commit: pending
+tech_stack:
+  language: JavaScript/TypeScript
+  framework: Browser/Node/WebGPU
+---
+
+# Buffer Allocation
+
+## 매 한 줄
+> **"매 buffer 매 reuse 매 GC 의 X"**. Buffer allocation은 binary data buffer (ArrayBuffer / Typed Array)의 effective lifecycle 관리 — naive `new Uint8Array(N)` per operation은 GC churn을 유발해 frame budget을 깬다. 2026 WebGPU / WebCodecs / canvas heavy app 의 must-skill.
+
+## 매 핵심
+
+### 매 cost source
+- **Allocation**: V8/SpiderMonkey backing-store malloc + zero-fill.
+- **GC**: large allocation → Old-gen → expensive sweep.
+- **Cache miss**: 매 fresh memory 의 cold cache.
+- **Fragmentation**: many sizes → heap 의 fragmented.
+
+### 매 strategy
+- **Pool / freelist**: 매 size class 매 reuse.
+- **Subarray view**: single big buffer + slice views.
+- **Pre-allocation**: peak size 부터 allocate.
+- **SharedArrayBuffer**: cross-thread, no copy.
+
+### 매 응용
+1. Audio / video frame buffer.
+2. WebGL / WebGPU vertex / index buffer.
+3. WebSocket binary message parser.
+
+## 💻 패턴
+
+### Simple buffer pool
+```ts
+class BufferPool {
+  private pool: Uint8Array[] = [];
+  constructor(private size: number, private max = 32) {}
+
+  acquire(): Uint8Array {
+    return this.pool.pop() ?? new Uint8Array(this.size);
+  }
+
+  release(buf: Uint8Array) {
+    if (this.pool.length < this.max) {
+      buf.fill(0);
+      this.pool.push(buf);
+    }
+  }
+}
+
+const pool = new BufferPool(4096);
+const b = pool.acquire();
+// use ...
+pool.release(b);
+```
+
+### Size-class pool
+```ts
+class SizeClassPool {
+  private classes = new Map<number, Uint8Array[]>();
+
+  private classOf(n: number): number {
+    return Math.pow(2, Math.ceil(Math.log2(Math.max(n, 16))));
+  }
+
+  acquire(n: number): Uint8Array {
+    const cls = this.classOf(n);
+    const list = this.classes.get(cls);
+    return (list?.pop() ?? new Uint8Array(cls)).subarray(0, n);
+  }
+
+  release(buf: Uint8Array) {
+    const cls = buf.buffer.byteLength;
+    if (!this.classes.has(cls)) this.classes.set(cls, []);
+    this.classes.get(cls)!.push(new Uint8Array(buf.buffer));
+  }
+}
+```
+
+### Single backing buffer + views
+```ts
+const big = new ArrayBuffer(1024 * 1024); // 1MB
+const headers = new Uint32Array(big, 0, 256);          // 1KB header
+const body = new Uint8Array(big, 1024, 1024 * 1023);   // remainder
+```
+
+### WebSocket binary parsing — no copy
+```ts
+ws.binaryType = 'arraybuffer';
+ws.onmessage = (e: MessageEvent<ArrayBuffer>) => {
+  const dv = new DataView(e.data);
+  const type = dv.getUint8(0);
+  const length = dv.getUint32(1, true);
+  const payload = new Uint8Array(e.data, 5, length);
+  handle(type, payload);
+};
+```
+
+### SharedArrayBuffer ring buffer (worker comms)
+```ts
+// main
+const sab = new SharedArrayBuffer(1024);
+const view = new Int32Array(sab);
+worker.postMessage(sab);
+
+// worker — Atomics 매 lock-free synchronization
+self.onmessage = (e) => {
+  const view = new Int32Array(e.data);
+  Atomics.store(view, 0, 42);
+  Atomics.notify(view, 0, 1);
+};
+```
+
+### Reusable canvas pixel buffer
+```ts
+class FrameRecycler {
+  private buffers: ImageData[] = [];
+  acquire(w: number, h: number) {
+    return this.buffers.find(b => b.width === w && b.height === h)
+      ?? new ImageData(w, h);
+  }
+  release(b: ImageData) {
+    if (this.buffers.length < 4) this.buffers.push(b);
+  }
+}
+```
+
+### Detecting allocation hot spots
+```ts
+// 매 dev-only — 매 wrap allocator 매 trace
+const _orig = Uint8Array;
+let count = 0;
+(globalThis as any).Uint8Array = function (...args: any[]) {
+  count++;
+  if (count % 1000 === 0) console.warn('Uint8Array allocations:', count);
+  return new _orig(...args);
+};
+```
+
+## 매 결정 기준
+| 상황 | Approach |
+|---|---|
+| Per-frame temp buffer | pool with size class |
+| Streaming binary protocol | preallocated parser buffer |
+| Cross-thread share | SharedArrayBuffer + Atomics |
+| GPU upload | persistent GPU buffer + map/unmap |
+| Rare large alloc | direct allocate |
+
+**기본값**: measure GC pressure first; pool only when allocator shows up in profile.
+
+## 🔗 Graph
+- 부모: [[Memory Management]] · [[Performance]]
+- 변형: [[Object Pool]]
+- 응용: [[WebGPU]]
+- Adjacent: [[SharedArrayBuffer]] · [[TypedArray]] · [[Garbage Collection]]
+
+## 🤖 LLM 활용
+**언제**: pool implementation scaffold, view layout calculation, ring buffer design.
+**언제 X**: 매 GC actual measurement — Chrome Memory profiler.
+
+## ❌ 안티패턴
+- **Premature pool**: 매 micro-opt — measure first.
+- **Pool 무제한**: 매 leak — max size cap.
+- **Subarray after detach**: transferred buffer — invalid.
+- **Concurrent writes without Atomics**: SharedArrayBuffer 매 race.
+- **Forget to zero-fill on release**: 매 stale data leak across owners.
+
+## 🧪 검증 / 중복
+- Verified (V8 blog, MDN ArrayBuffer, WebGPU spec).
+- 신뢰도 A.
+
+## 🕓 Changelog
+| 날짜 | 변경 |
+|---|---|
+| 2026-05-08 | Phase 1 |
+| 2026-05-10 | Manual cleanup — buffer pool + view layout pattern |