# 概述

Qualcomm^®^ Linux AI 栈允许开发人员在 Qualcomm 硬件加速器（例如神经网络处理单元 (NPU)、图形处理器 (GPU) 和中央处理器 (CPU)）上最佳地部署预训练的深度学习模型。Qualcomm AI 软件产品包含软件开发工具包 (SDK)、API、示例程序、开发工具以及 GStreamer 和 TFLite 等第三方框架支持，以简化应用程序开发。

![../_images/top_level_ai_hw.png](data:image/png;base64,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)
**Qualcomm Linux AI 堆栈的顶层 AI 硬件和软件模块**

Qualcomm Linux AI 栈的关键组件包括：

- **AI 应用** - 基于 Gstreamer 的示例程序，可以根据需要使用或定制。
- **Gstreamer 插件** - Qualcomm Linux 软件提供基于 Gstreamer 的机器学习插件，用于通过 TFLite、Qualcomm^®^ Neural Processing Engine SDK 等加速 AI 推理，以及用于预处理和后处理的 Gstreamer 插件。
- **Qualcomm AI 栈**包含两个用于加速 AI 工作负载的 SDK。**Qualcomm Neural Processing Engine SDK** 和 **AI Engine Direct** 提供工具、库等，以在多个硬件加速器上最佳地加速 AI 模型。
- Qualcomm SoC 提供三种用于 AI 负载的**硬件核心**。

    - **神经处理单元 (NPU)** - 也称为 Qualcomm® Hexagon™ Tensor Processor (HTP) 或 DSP/HMX，适合以低功耗、高性能执行 AI 工作负载。为优化性能，需要把预训练模型量化到任意一种支持的精度。
    - **图形处理单元 (GPU)** - Qualcomm® Adreno™ GPU 适合执行中等功率、中等性能的 AI 工作负载。AI 工作负载可以通过 OpenCL 内核进行加速。GPU 还可用于加速模型预处理/后处理。
    - **中央处理器 (CPU)** - 在 CPU 上进行 AI 推理可以用来与其他硬件加速器进行模型准确性/性能的基准测试。CPU 还可以用于运行模型的预处理/后处理。

Last Published: Aug 06, 2024

[Next Topic
架构](https://docs.qualcomm.com/bundle/publicresource/80-70014-15Y/topics/architecture.md)