CSEM, a leader in low-power RFIC design and embedded systems, announced today at GLOBALFOUNDRIES (GF) annual Global Technology Conference (GTC) that CSEM is developing ultra-low power Bluetooth Low Energy® (BLE), CMOS radar mmWave IP and embedded machine-learning accelerators on GLOBALFOUNDRIES (GF) 22nm FD-SOI (22FDX®) platform.

The combination of GF’s 22FDX and CSEM’s “IcyTRX” dual mode Bluetooth Low Energy and Bluetooth classic RF is expected to achieve 50% lower power and 50% lower area than their existing GF 55 nm BLE RF IP solution. Moreover, CSEM’s patented machine learning accelerator, called Visage, is expected to achieve more than 60% power savings and 80% area savings with 22FDX compared to 55nm, allowing for true AI edge-processing with a dedicated RISC-V processor.

“Collaborating with partners such as GF, we continuously enhance our IP for the latest processes,” said Alain-Serge Porret, Head of Wireless and Integrated Systems at CSEM. “The availability of these best-in-class IPs will enable GF’s clients to design always-on smart sensors and streaming audio devices with ground-breaking battery life.”

“CSEM will help GF to further broaden our IP portfolio to continue providing our clients with a differentiated 22FDX solution for IoT connectivity, image recognition and motion-sensing devices," said Mark Ireland, vice president of Ecosystem Partnerships at GF. "We are pleased to welcome CSEM as a partner in our FDXcelerator program. Together, we enhance our IP capabilities and services enabling clients with low-power and cost-efficient AI edge processing solutions."

The CSEM RF and mmWave IPs using GF’s 22FDX will be available for licensing to qualified clients starting 2Q 2020.