Mythic is an AI processor company that has developed a unique analog compute-in-memory technology for edge AI applications. The company's flagship product is the M1076 Analog Matrix Processor (Mythic AMP), which utilizes analog computing to deliver high performance at low power consumption. Mythic's technology stores neural networks completely on-chip using high-density analog memory, allowing for efficient AI inferencing without the need to constantly swap data on and off chip like digital solutions.
The M1076 chip can perform up to 25 trillion operations per second (TOPS) while consuming only about three watts of power. This represents a significant improvement in energy efficiency compared to digital processors, with Mythic claiming its solution is 10 times more cost and power-efficient. The chip has a capacity for around 80 million synapses that can be arranged in layers as needed for different neural network architectures.
Mythic offers the M1076 in various form factors, including a standalone processor, M.2 cards, and PCIe cards. The company also provides development boards, evaluation systems, and software support for integration with different host platforms and pre-trained deep neural network models.
Key customers and partnerships
Mythic has secured Lockheed Martin as a customer, with Lockheed Martin Ventures also investing in the company. The defense and aerospace giant is utilizing Mythic's technology for applications in areas like drones, autonomous systems, and augmented reality headsets.
Mythic is targeting its products for use in computer vision applications across industries including defense, public safety, industrial environments, smart cities, and consumer devices. The company sees potential applications in security cameras, drones, robots, and AR headsets that require high-performance edge AI processing.
By using this site, you agree to allow SPEEDA Edge and our partners to use cookies for analytics and personalization. Visit our privacy policy for more information about our data collection practices.