Snapdragon 820 Chip for Machine Learning Applications
Qualcomm's new Snapdragon 820 chip (used in many high-end smartphones) can be used for deep learning and neural network applications. The Snapdragon Neural Processing Engine is an SDK powered by the Zeroth Machine Intelligence Platform.
The combination can be used on the 820 to speed the performance of CNNs and deep learning on devices ranging from connected video cameras to cars and much more. The 820 incorporates a CPU, GPU and DSP, all of which can be used to run deep learning algorithms for different applications.
The 820 is a chip integrates five main components: Kryo CPU, Adreno 530 GPU, Hexagon 680 DSP, Qualcomm Spectra ISP and the X12 LTE Modem.
The Kryo CPU features a custom-built quad-core setup running at 2.2 GHz and size has shrunk to 14 nanometers and works with Qualcomm’s Symphony system manager to automatically find the right balance between system power and heat management.
The Adreno 530 GPU provides both 40 percent faster graphics and 40 percent less power consumption and makes it easier for smartphones to power devices with dense 4K displays.
The Hexagon 680 DSP (digital signal processor) and Spectra ISP (image signal processor) work together to produce images that are brighter and less noisy, especially in low-light scenarios. Spectra also features 14-bit processing for higher quality images considering each photo contains more data.
The X12 LTE modem offers Cat 12 performance for cellular data, providing download speeds up to 600 Mbps and uploads up to 150 Mbps for a 33% improvement for downloads and 200% improvement for uploads compared to the 810 chip. This makes it easier to stream content from YouTube or Netflix and provides improved broadcasting on Periscope. The modem also features support for multiple user MIMO wireless communication allowing improved Wi-Fi performance especially in places packed with many connected devices.