Highlights:

  • The software stack focuses on power efficiency, like how to quantize workloads, and probability of code across many hardware targets of Qualcomm and its clients.

Mobile and Internet of Things (IoT) chip giant Qualcomm recently announced an expansion of its Artificial Intelligence (AI) developer runtime offering named Qualcomm AI Stack portfolio. The company claims the new solution will address the “unique needs of AI for each and every business line” that Qualcomm’s chips serve. It includes mobile phones, connected vehicles, and infrastructure in the IoT installations.

The software company also incorporates tools like the company’s Natural Processing SDK while allowing digital engineers to “leverage that same work across every product.”

In a media briefing, Qualcomm Vice President of product management Ziad Asghar said, “We are extending one technology roadmap across all our different business lines.”

He added, “The step function here is to be able to take your work and port it across those different business lines.”

The software stack focuses on power efficiencies, like quantizing workloads and the probability of code across many hardware targets of Qualcomm and its clients.

Asghar said, “the performance per watt that we are showing is just outstanding in terms of how much work we can do for a given amount of power.” “Even for the same hardware, as we optimize the software, we can get 30% to 40% better performance in some cases.”

Asghar said the capacity to handle many conditions is the primary virtue.

Asghar said, “Some environments are very focused on their end markets — auto or cloud.”

He also said, “This is the best of class of experience for what you could get out of our hardware, that’s the unique advantage.”

“This will allow you to get the most optimized and best experience from an efficiency perspective and an accuracy perspective, so we think this is a huge differentiation for our partners and us.”

According to Asghar, one of the goals of the technology is for customers to be able to modify the performance of machine learning for different criteria.

“For this application, latency is very critical; you have to have the ability to optimize for that, whereas for that application, latency is most important, or accuracy — you can make those trade-offs with the stack we are offering, and then leverage that works everywhere.”