# Architecture & Performance

Audio Dispatcher was built with strict Data-Oriented Design (DOD) principles to ensure it scales to AAA workloads.

#### 1. Double Buffering (No Main Thread Stalls)

In a naive implementation, the Main Thread must wait for ECS Jobs to finish before it can read the audio queue (`Dependency.Complete()`). This causes massive frame drops.

Audio Dispatcher uses **Double Buffering**:

* **Frame 1:** Your Jobs write to `Queue A`. The Main Thread reads from `Queue B` (which is empty).
* **Frame 2:** Your Jobs write to `Queue B`. The Main Thread reads from `Queue A`. Because the Main Thread is always reading the queue from the *previous* frame, it never has to wait for worker threads. The 1-frame audio delay (\~16ms) is imperceptible to players.

#### 2. 32-Byte Cache Alignment

The `AudioEvent` struct uses `[StructLayout(LayoutKind.Explicit)]` to overlap the `Entity` (used for loops) and `float3 Position` (used for one-shots) in memory.

This compresses the struct to exactly **32 bytes**. This is a magic number in CPU architecture: exactly two events fit perfectly into a standard 64-byte L1 CPU cache line, resulting in lightning-fast queue processing.

#### 3. Priority-Based Voice Stealing

What happens if you have a pool of 32 sounds, but 40 explosions happen at once? Instead of a naive Round-Robin approach, the Burst-compiled `AudioBrainJob` evaluates the priority (Volume) of all currently playing sounds. It will automatically find the quietest sound and overwrite it with the new explosion.

#### 4. TimeScale Independent

The system calculates sound lifetimes using `AudioSettings.dspTime` instead of `Time.time`. If you use slow-motion effects (`Time.timeScale = 0.1f`), your sounds will not be cut off prematurely.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://sniveler-code.gitbook.io/dots/audio-dispatcher-audio-source/architecture-and-performance.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.