Unlocking AI/ML Performance on NXP’s i.MX 93 Applications Processor | NXP Semiconductors

デザイン・センター
トレーニング
Unlocking AI/ML Performance on NXP’s i.MX 93 Applications Processor

Unlocking AI/ML Performance on NXP’s i.MX 93 Applications Processor

アカウントが必要です

サインインこのコンテンツおよびサイトの追加機能にアクセスすることができます。アカウントをお持ちでないですか？今すぐご登録下さい。

About This Training

Every bit of performance counts when developing machine learning applications for the edge. NXP i.MX 93 applications processor, featuring the Arm® Ethos™-U65 microNPU, delivers fast, energy-efficient, ML inference processing. With NXP’s eIQ® toolkit, you can seamlessly convert, optimize, and deploy your models on the NPU with ease.

In this training, we’ll dive into the key considerations for maximizing AI/ML performance on the i.MX 93 MPU. You’ll learn how to avoid common model conversion pitfalls, navigate trade-offs in model parameters, quantization, and ML frameworks, and apply advanced techniques to further enhance performance.

Presentation

Unlocking AI/ML Performance on NXP’s i.MX 93 Applications Processor

What You Will Learn

Key features of the i.MX 93 processor and how the integrated Arm Ethos-U65 microNPU accelerates edge AI workloads
Comparing post-training quantization formats and techniques to improve model efficiency
Using the Vela compiler in the eIQ environment to optimize models for the NPU
Enhancing performance through advanced optimizations, including camera pipeline tuning and distributing operations across processor components
Real-world applications demonstrating these strategies in action

Presented by

Monica Houston

Technical Marketing Manager, Tria

Davis Sawyer

AI Product Marketing Manager, NXP Semiconductors

期間

43 min

言語

英語

スキルレベル

初級編

0閲覧数1いいね

We are looking for ways to improve.

アンケートに答える。

Additional Resources

i.MX 9 Applications Processors

GoPoint for i.MX Applications Processors

Avnet Launches MaaXBoard OSM93 Development Kit for Design Engineers Seeking an Energy-efficient, Edge AI Single-Board Computer

MaaXBoard OSM93

Stay In the Loop

Subscribe to our newsletter to stay updated with our latest developments and if you need further assistance, we are here to help.

Subscribe Contact Support