Case Study: Today’s Demanding Algorithms Complicate Processor Selection Decisions

Submitted by BDTI on Thu, 06/29/2017 - 01:00

As system designers race to make IoT and edge devices more capable, they are incorporating increasingly complex and demanding algorithms. Cameras and microphones are now the eyes and ears of systems that help us drive our cars, maintain the safety of our homes, diagnose health issues, and much more. Processor vendors, seeking to meet escalating requirements of processing sensor data at the edge, are designing new heterogeneous devices that integrate CPU cores, DSPs, GPUs, and other specialized processing engines for tasks such as image processing and deep neural networks.

To select the best heterogeneous processor SoC for a new product design, designers must weigh numerous factors, including performance, energy consumption, scalability, and development tools. Evaluating these factors can be complex: How well will your algorithms map to the device’s architecture? How much effort will be required to map them efficiently? How robust are the tools? Is there solid support for key libraries and frameworks?

The challenges in selecting the right processor are compounded by the limited experience most product designers have in evaluating processors. This is where a tested methodology can help. For more than 25 years, over the course of dozens of projects, BDTI has developed unique insights into the key questions to ask and a framework for weighing tradeoffs when evaluating processors. Several recent engagements demonstrate the value of this experience.

In one project, an automotive subsystem manufacturer engaged BDTI to recommend a low-cost processor for a new vision-based product. BDTI analyzed a series of use cases to determine a set of weights to apply to key processor criteria including performance, cost, energy efficiency and ease of development. BDTI then evaluated candidates using the customer’s algorithms and delivered its recommendations. In another project, a medical device manufacturer engaged BDTI to advise it on selecting a processor for a significant upgrade of one of its systems. In this case, an analysis of requirements identified low latency and scalability (to support a family of products) as the most critical criteria.

In these projects, BDTI’s deep knowledge of processors and algorithms enabled it to quickly recommend the right processor for the customer’s product. If you're selecting a processor chip, processor core, or SoC for a computer vision or deep learning application, BDTI is your one-stop shop for advice, evaluation, and recommendations. Don’t rely on data sheets and vendor claims alone—take advantage of BDTI's unique experience in evaluating processors. Contact Jeremy Giddings for a confidential consultation by phone at +1 (925) 954-1411 or via BDTI’s website.

Add new comment

Log in to post comments