火炬 Releases Prototype Features To Execute 机器学习 Models On-Device Hardware Engines

0
3329
资源: //github.com/pytorch/pytorch

火炬 has recently released four new 火炬 prototype features. The first three enable mobile machine-learning developers to execute models on the full set of hardware (HW) engines making up a system-on-chip (SOC) system. This allows developers to optimize their model execution for a unique 性能, power, and system-level concurrency.

新功能包括以下使能在设备上的硬件引擎执行的功能:

  • 使用与Google Android合作开发的Android神经​​网络API(NNAPI)的DSP和NPU。
  • 通过Vulkan在Android上执行GPU
  • 通过Metal在iOS上执行GPU

There is increasing ARM usage in the 火炬 community with Raspberry Pis and Graviton(2) platforms. Hence, the new release also includes developer efficiency benefits with recently launched support for ARM64 builds for Linux.

广告Coursera Plus标语,包含约翰·霍普金斯大学,谷歌和密歇根大学的课程,突出显示数据科学职业发展的内容

谷歌 Android的NNAPI支持

火炬’与Google Android团队的合作使Android’s Neural Networks API (NNAPI) via 火炬 Mobile. On-device machine learning allows ML models to run locally on the device without transmitting data to a server. This offers lower latency and improved privacy and connectivity. The Android Neural Networks API (NNAPI) is designed for running computationally intensive processes for machine learning on Android gadgets. Thus, machine learning models can now access additional hardware blocks on the phone’的片上系统,允许开发人员在Android手机上解锁高性能执行。 NNAPI使Android应用可以在为android提供动力的最强大和最活跃的芯片上运行计算加速的神经网络,包括DSP(数字信号处理器)和NPU(专用神经处理单元)。 

The API was first introduced in Android 8 and significantly expanded in Android 10 and 11 to support a richer AI model. This integration allows developers to access NNAPI directly from 火炬 Mobile. This initial release includes fully-functional support for a core set of features and operators. 谷歌 and 脸书 will be working on expanding capabilities soon.

火炬 Mobile GPU Support

GPU推论可以在许多模型类型上提供出色的性能,尤其是那些使用高精度浮点数学运算的模型。像在高通,联发科技和苹果公司的SOC中那样,利用GPU来执行机器学习模型,都支持CPU卸载。这为非机器学习用例释放了移动CPU。设备GPU的原型协助的主要级别是通过iOS的Metal API规范和Android的Vulkan API规范。这项特征’的性能尚未优化,模型覆盖范围有限,因为它还处于不成熟阶段。该团队预计,这一情况将在2021年显着改善。

资源: //pytorch.org/blog/prototype-features-now-available-apis-for-hardware-accelerated-mobile-and-arm64-builds/
的GitHub: //github.com/pytorch/tutorials/tree/master/prototype_source

广告

发表评论

请输入您的评论!
请在这里输入您的名字

该网站使用Akismet减少垃圾邮件。 了解如何处理您的评论数据.