High-speed I/O processors sport 64-bit time-predictable and energy efficient cores

Kalray announced that it will launch the third-generation of its high-performance MPPA processor: the MPPA3 Coolidge. With the success of its second-generation MPPA2-256 Bostan processor, selected by numerous automotive and aeronautic OEMs and Tier1’s, Kalray has taken its unique manycore technology to the next level in terms of performance.

The third-generation of the MPPA will be available for critical embedded applications, like autonomous vehicles, aeronautics, medical devices, robotics and drones. The chip will bring unique data fusion capabilities and will be especially efficient for compute-heavy processing, like deep learning and computer vision.

The MPPA3 Coolidge will be produced with 16nm FinFet technology and is designed for ASIL B safety functionality. It will be comprised of 80 or 160, Kalray 64-bit time predictable and energy efficient cores and, in addition to the cores themselves, will boast 80 or 160 complimentary co-processors used for the acceleration of computer vision and deep learning. The Coolidge will run at 1.2 GHz of core frequency using less than 20W of power. This superior ratio will be the key to making Coolidge work for high performance embedded applications. Coolidge performances are expected to reach up to 5 Tera FLOPS and 9 TOPS.

The MPPA Coolidge will be compatible with multiple interfaces – PCIe, Ethernet, DDR, CAN, USB and others. It will also contain security and safety blocks to keep the processor running safely and protect systems against cyber-attack.

The MPPA Coolidge will be commercially available in mid-2018. Kalray is currently working with various customers to put the Coolidge at the heart of newly developed critical embedded applications.

In addition to its industry-leading power efficiency, the architecture of Coolidge will offer numerous other major advantages for customers: its massive parallelization will allow multiple applications to run in parallel at high performances, and mixing application criticality and its predictable architecture will allow applications to be predictable.