The M1076 Analog Matrix Processor (Mythic AMP) comes in several form factors to address a wider range of designs: a standalone processor; an ultra-compact PCIe M.2 card; and a PCIe card with up to 16 AMPs. It can support up to 25 TOPS of AI compute in a 3W power envelope, consuming up to 10x less power than a typical SoC or GPU solution, says the company.
In a 16-chip configuration, the M1076 AMP PCIe card delivers up to 400 TOPs of AI compute while consuming only 75W. The M1076 AMP has been optimized to scale from edge endpoints to server applications, addressing multiple vertical markets including smart cities, industrial applications, enterprise applications, and consumer devices.
"Mythic's groundbreaking inference solution takes AI processing and energy-efficiency to new heights," says Tim Vehling, senior vice president, product & business development at Mythic. "With its ability to scale from a compact single-chip to a powerful 16-chip PCIe card solution, the M1076 AMP makes it easier for developers to integrate powerful AI applications in a wider range of edge devices that are constrained by size, power, and thermal management challenges."
The M1076 AMP, says the company, offers an unparalleled combination of performance and power-efficiency that is integrated into an ultra-compact 22mm x 30mm PCIe M.2 A+E Key card for space-constrained embedded edge AI applications. For edge AI systems that require more demanding workloads – including many streams, multiple large, deep neural networks, and higher resolutions and frame rates – a PCIe card form-factor with 16 Mythic AMPs supporting up to 400 TOPS and 1.28 billion weights in a 75W power profile can be utilized.
The M1076 AMP is offered as being ideal for video analytics workloads including object detection, classification, and depth estimation for industrial machine vision, autonomous drones, surveillance cameras, and network video recorders (NVRs) applications. The M1076 AMP can also support cutting-edge AR/VR applications with low latency human body pose estimation, which