Qualcomm introduced the Snapdragon XR1 chip, designed for virtual and augmented reality headsets. The company believes that manufacturers will be able to use the platform in the devices of the main segment.
XR1 is based on a heterogeneous architecture, including a multi-core ARM CPU, vector and graphics processors. The neural network hardware acceleration unit supports TensorFlow, ONNX, Caffe, Caffe2, and Android Neural Networks API. The company believes that it will save energy when using AI for image processing on the device.
The processor supports OpenGL, OpenCL and Vulkan API, image output with 4K resolution at up to 60 frames per second, as well as work with two screens. It tracks with high precision movement in six degrees of freedom. XR1 displays image changes after movement with a delay of less than 20 milliseconds.
To improve the sound quality, the developers used the 3D Audio Suite technology, Qualcomm’s Aqstic Audio Technologies and aptX. Also used are the functions of Head Related Transfer, which simulate the direction of sound from virtual objects.
The company does not disclose such characteristics of the chip as the processor clock frequency and power consumption. According to representatives of Qualcomm, the company Meta, VIVE, Vuzix and Picoare are preparing to release devices with Snapdragon XR1.
In February 2018, Qualcomm announced the release of the Qualcomm AI Engine for artificial intelligence systems in mobile devices on a number of Snapdragon processors. These include Snapdragon 845, 835, 820 and 660.