Snapdragon 820 Deep Learning SDK Will Promise Better Privacy For The Future
In terms of performance, Snapdragon 820 is a force to be reckoned with and now, Qualcomm has introduced its deep learning SDK for the chipset to make it even better than before.
Snapdragon 820 Deep Learning SDK: Here Are Some Of Its Key Features
Given below are some of the useful features of Qualcomm’s deep learning SDK:
- Accelerated runtime for on-device execution of convolutional and recurrent neural networks on the Snapdragon 820 cores (Qualcomm Kryo CPU, Qualcomm Adreno GPU, Qualcomm Hexagon DSP)
- Support for common deep learning model frameworks, including Caffe and CudaConvNet;
- A lightweight, flexible platform designed to utilize Snapdragon heterogeneous cores to deliver optimal performance and power consumption;
- Supports companies in a broad range of industries, including healthcare, automotive, security, and imaging, to run their own proprietary trained neural network models on portable devices.
The software development kit is called Snapdragon Neural Processing Engine and it has been specifically made for Zeroth, which is a machine intelligence platform. Currently, the SDK will only be compatible with Snapdragon 820, but Qualcomm’s Snapdragon 830 will probably support it as well.
What we do not know is whether or not future chipsets filling the mid-range categories will provide support for it or not. While we will stumble upon this at a later time, we would like to inform you that this SDK will allow companies to run deep learning programs locally on your devices, and since the most ubiquitous devices running the SoC are mobile devices, companies will definitely start from there.
One of the major highlights of the new Snapdragon 820 SDK is that it will provide better privacy because the system will not be uploading significant amounts of data to the cloud; after all, since all deep learning will be done locally, minuscule amounts of information will be left online.
Another perk of Qualcomm’s SDK is that it will not just be limited to mobile devices. Qualcomm stated that a startup called Nauto, who’s forte is to produce smart dashboard cameras will also be integrating the new SDK into its devices. As a result, vehicles will provide better safety for drivers in the future.
What we really want is support for future chipsets that do not belong to the flagship category. However, will Qualcomm entertain such a possibility? It is very difficult to state this at the moment.