Qualcomm introduces a new Vision Intelligence Platform

As the IoT space continues to grow, one key direction is that of smart computing at the edge. Targeting this market is what Qualcomm is attempting to address with the introduction of a new Vision Intelligence Platform. The new platform targets IoT AI vision applications – things such as home surveillance cameras, smart home locks, robotic vacuum cleaners, and other applications that rely heavily on processed information from a camera. Essentially, Qualcomm is combining the processing power of their Snapdragon family of mobile processors, along with their advanced image signal processor, GPU, and Hexagon DSP to create powerful vision IoT chips.

Models

Two initial models are being introduced.

Vision Intelligence Platform
Model CPU GPU DSP
QCS605 2 × 2.5 GHz Kryo 300 Gold
6 × 2.5 GHz Kryo 300 Silver
Adreno 615 Hexagon 685
QCS603 2 × 1.6 GHz Kryo 300 Gold
2 × 1.7 GHz Kryo 300 Silver
Adreno 615 Hexagon 685

Apart from the core configuration and frequency, the two models are identical. Both parts support 16-bit LPDDR4X memory up to 1866 MHz and are fabricated on Samsung’s 10 nm Low Power Early (10LPE) process which would be Samsung’s first generation 10nm.

For imaging capabilities, both models support up to 32 MP (single) or 16 MP (dual camera) and integrate two 14-bit Spectra 270 ISPs. Those chips integrate Qualcomm’s AI Engine which consists of the Adreno 615 GPU, Hexagon 685 Vector Processor with two Vector eXtensions (HVX), and accompanying API and libraries for a reported performance of 2.1 TOPs at 1 W.

For wireless connectivity, both chips have Bluetooth 5.1 and WiFi (802.11a/b/g/n/ac) support.

Both models are currently sampling.

Spotted an error? Help us fix it! Simply select the problematic text and press Ctrl+Enter to notify us.

Leave a Reply

avatar
  Subscribe  
Notify of