The M1076 Mythic Analog Matrix Processor (AMP™) delivers up to 25 TOPS in a single chip for high-end edge AI applications. This device integrates 76 AMP tiles to store up to 80M weight parameters and execute matrix multiplication operations without any external memory. This allows the M1076 to deliver the AI compute performance of a desktop GPU while consuming up to 1/10th the power (3 to 4 W typ.), all in a single chip in a 19 mm x 15.5 mm BGA package.
The Mythic AMP is a truly unique AI compute platform that enables effective surveillance systems, smart appliances, powerful machine vision systems, commercial drones, brilliant robotics, and more. With unmatched performance and power efficiency, the Mythic platform enables AI designers to deploy in form factors that were previously out of reach, and it supports standard platforms, frameworks, and DNNs for great ease of use; Mythic’s Optimization Suite facilitates this by converting the neural network to an 8-bit representation, while preserving accuracy in the analog compute domain.
The Mythic Analog Matrix Processor is based on a unique tile-based AI compute architecture that features three fundamental hardware technologies – Compute-in-Memory, Dataflow Architecture, and Analog Computing. Aside from the M1076 chip alone, Mythic provides some commonly used formats to integrate their capabilities into your computing system:
-> The Mythic MP10304 Quad-AMP PCIe card enables high-performance, power-efficient AI inference for edge devices and servers. This half-height, half-length PCIe card simplifies the integration effort into platforms where space is a constraint, and features four M1076s, delivering up to 100 TOPS of AI performance and supporting up to 320 million weights for complex AI workloads at less than 25 W of power.
-> For smaller workloads, Mythic also offers their MM1076 M.2 M key card (22 x 80 mm) and the even more compact ME1076 M.2 A+E key cards (22 x 30 mm), each with a single M1076 chip on board in 4- and 2-lane PCIe 2.1 formats for 2 and 1 GB/s bandwidth respectively.
If you would like to put Mythic’s technology to the test in your end products, apply for a sample, specifying your project requirements and preferred sample format. Mythic’s technology looks like an efficient new way of approaching AI workloads, and it is a plug-in-and-play solution, so let us know what you think after you’ve tried it!
(All images sourced from Mythic)