Skip to content
AI.info

The Pulse

VeriSilicon unveils processing subsystem for AI and AR glasses

VeriSilicon has introduced a wearable subsystem combining processing components for AI and augmented reality glasses. The offering includes a new Nano-XR series and technologies intended to reduce memory use and data transfers.

VeriSilicon unveils processing subsystem for AI and AR glasses

VeriSilicon has introduced a wearable processing subsystem for AI and augmented reality glasses, combining camera, AI, audio, display and graphics capabilities. The offering combines a new glasses-focused Nano-XR series of processing intellectual property with an architecture designed to reduce data transfers and external memory use, the company said.

Source prnewswire.com

In the Nov. 17 release, VeriSilicon said customers can select and configure processing components for applications ranging from basic low-power glasses to more advanced AR devices. The company is also working with customers on dedicated smart-glasses chip designs using the new components.

Components for glasses

The Nano-XR series extends VeriSilicon's existing Nano IP portfolio for AI and AR glasses. Its components cover image capture, infrared sensing, display processing and graphics.

For image capture, the company says ISPNanoUltra-XR delivers smartphone-level photo and video quality while optimizing power consumption and silicon area. Another component, ISPPico-XR, is designed for infrared sensors and supports always-on functions including eye tracking and gesture recognition.

DCNanoUltra-XR is intended for display processing in 3DoF and 6DoF applications. VeriSilicon describes it as low latency and says it includes power-gating strategies and support for multiple high-speed display interfaces.

For graphics, GCNanoOptimaV-XR builds on the earlier GCNanoUltraV. The company says the new component offers improved projection transformation, higher graphics performance and a smaller memory footprint for glasses applications.

Configurable processing

The subsystem can also draw on VeriSilicon's wider portfolio of graphics, neural-network, video, digital-signal, image-signal and display processing IP. Those components have configurable options for different power and performance requirements.

The company says its DSP components support voice processing, noise reduction, audio codecs and Cat.1/Cat.4-related applications in wearable devices. The NPU components support workloads including low-power wake-up, image enhancement and general-purpose AI computing.

Customers can choose components based on their application scenarios and computing requirements. The offering supports RTOS, Linux and Android.

Moving data between blocks

At the subsystem level, VeriSilicon describes an architecture that manages data flows between processing blocks while minimizing unnecessary transfers and external memory access. Two named technologies address compression and communication between those blocks.

DECNano, a compression technology, provides visually lossless compression and reduces external PSRAM and DDR capacity and bandwidth requirements, according to the company. The FLEXA interface supports segment-level data exchange between processing components, using on-chip SRAM to buffer intermediate data.

VeriSilicon says the approach further reduces external memory access and end-to-end latency. Weijin Dai, its chief strategy officer, executive vice president and general manager of the IP division, said the architecture and compression technology also reduce power consumption and improve the efficiency of data movement.

Dai said VeriSilicon is actively collaborating with customers on “dedicated smart glasses chip designs incorporating Nano-XR IPs.” Those designs are intended for more advanced AI and AR applications.

Source

  1. PR Newswire

Related stories

Explore

More articles