British chip designer ARM has announced Project Trillium, which is a suite of machine learning IP including new processors that aims to deliver enhanced machine learning and neural network functionality. ARM says that the technologies are aimed at the mobile market and will enable a new class of machine learning-equipped devices with advanced computing capabilities, including object detection. It is essentially launching an ML processor for general AI and an object detection processor that specialises in detecting objects.
The highlights of the announcement are the processors – ARM ML and ARM OD. The ARM ML claims to have been built from the ground-up, specifically for machine learning. ARM says the processor delivers more than 4.6 trillion operations per second for mobile computing. It apparently delivers two to four times the real-world throughput. Meanwhile, the ARM OD is about object detection.
In a blog, ARM said, “In combination, the ARM ML and OD processors perform even better, delivering a high-performance, power-efficient people detection, and recognition solution. Users will enjoy high-resolution, real-time, detailed face recognition on their smart devices delivered in a battery-friendly way.”
Additionally, ARM NN software component acts as a go-between for neural network frameworks such as TensorFlow, Caffe, and Android NN and the full range of ARM Cortex CPUs, ARM Mali GPUs, and ML processors.
Similar to all ARM processors, the company will not be making the processors itself. Rather, it will license the designs to third-party manufacturers. The new suite of ARM ML IP will be available for early preview in April of this year, the company says. It will be open for general availability in mid-2018.
ARM has noted that Project Trillium is not the commercial brand name for ARM machine learning technology, rather only a codename that will be replaced later.