AI's immense potential is being constrained
Innovation is hampered
Today’s digital AI processors are tremendously expensive to develop and overly complex to use, restricting innovation to only the largest technology companies.
Inference at the edge is a challenge
Inference is what enables a neural network to perform in real-life situations, as it encounters new data. Edge inference on today’s digital processors is a massive technical challenge. Edge devices running AI must meet demanding requirements for size, performance, and power consumption. Legacy digital architectures simply can’t keep up.
Edge performance suffers
Edge devices running AI must meet demanding requirements for size, performance, and power consumption. Legacy architectures simply can’t keep up.
MYTHIC - ADVANCED AI INFERENCE FOR EDGE DEVICES
A leap forward in performance and affordability
Mythic Analog Matrix Processors (Mythic AMP™) offer huge advantages in power, performance, and cost. They lower the barriers to innovation, making it vastly easier and more cost-effective to create powerful Edge AI solutions.
A unique analog compute architecture
Mythic AMPs leverage analog computing by performing the calculations required for inference of deep neural networks inside a dense flash-memory array. This represents a significant advantage over typical digital architectures.
Designed for the most demanding workloads
With Mythic’s integrated development environment, AI developers can quickly deploy even the most sophisticated deep neural networks, confident that they will perform effectively – from the data center to the edge device.
-
Smarter cities and spaces
Real time understanding makes cities and spaces safer and more sustainable
-
Personalized devices and services
Consumer electronics and services customized to each user’s preferences, personality, and needs
-
Safer and simpler vehicles
Multimodal perception forms the bedrock of safe Autonomous Vehicle and ADAS stacks
-
Intelligent machines and robots
Precise factories and farms maximize resource efficiency based on changing environments