Qualcomm has launched the Snapdragon 670 chip to improve the performance of AI on mid-range mobile devices.
On Wednesday, the US chipmaker revealed the new mobile platform, the latest addition to the mid-range Snapdragon 600 tier.
The Snapdragon 710 mobile platform, launched in May for high-tier smartphone designs, has already made a push towards artificial intelligence (AI) and neural networking.
However, the refresh of the Snapdragon 600 family to include enhanced AI, reminiscent of the higher processor range, may entice vendors of less expensive devices to also explore the possibilities of such next-generation technologies.
Qualcomm says the 670 processors have been "engineered to deliver cutting-edge technologies to mainstream smartphone buyers who want to make the most of what their Snapdragon-powered device has to offer."
The Qualcomm Snapdragon 670 features various architectures including the Qualcomm Kryo CPU, Qualcomm Spectra ISP, Qualcomm AI Engine, and Snapdragon X12 LTE modem.
The Kryo 360 CPU has been selected to power the Snapdragon 670, which utilizes two cores of up to 2.0GHz and six "efficiency" cores capable of up to 1.7GHz. Qualcomm says that users of the revamped chip can expect a performance boost of up to 15 percent.
According to the firm, the particular focus of the processors is to target AI-driven applications such as smart camera focus and settings. The 670 uses the same digital signal processing (DSP), the Qualcomm Hexagon 685, which was previously utilized in higher-tier mobile platforms.
In addition, Qualcomm has included the Spectra 250 ISP, which enables a number of features for camera use such as noise reduction and active depth sensing.
Qualcomm has also integrated the company's third-generation AI engine in the processor, which the firm claims improves AI performance up to 1.8 times in comparison to previous generations.
"Our AI Engine offers an extensive array of frameworks giving developers a variety of choices to create better apps and immersive multimedia," Qualcomm added. "This includes a wide range of tools, including the Qualcomm Neural Processing SDK and Hexagon Neural Network (NN)."
Caffe, Caffe2, TensorFlow, TensorFlowLite, and Open Neural Network Exchange (ONNX) are supported by the engine.
The X12 LTE modem has also been included which has been designed for improved speeds in congested areas and stronger signals in rural areas and places where coverage is poor.
The chip manufacturer says that the addition of different architectures should give vendors the freedom also to experiment with different mobile sizes from previous generations.
The Snapdragon 670 Mobile Platform is currently available to vendors and devices utilizing the chip are expected to land in the commercial market later this year.