AVAI
Mobilint
AI Accelerator Chip

ARIES NPU

Ultra-performance AI accelerator chip delivering 80 TOPS at just 25W

AI Performance
80 TOPS
TDP
25W
NPU Cores
8
PCIe
Gen4
ARIES NPU

Overview

ARIES NPU

ARIES is an AI accelerator chip with a unique architecture that delivers high compute efficiency at low power consumption. Designed for versatility in handling diverse vision AI tasks, ARIES has been validated on over 400 deep learning models across CNN, RNN, and transformer architectures.

Run up to 32 different deep learning models concurrently on a single chip — detection, classification, and tracking in parallel. INT8 quantization via SDK qb maintains 99% of original model accuracy with no retraining required.

The programmable architecture supports state-of-the-art and custom model architectures. DNN-based architectural blocks adapt to diverse workloads while maximizing data reuse and minimizing memory access for best-in-class performance per watt.

Mobilint

Authorised Mobilint Distributor
Official sales, integration & support services

Features

Key Capabilities

Multi-Model Execution

Run up to 32 different deep learning models concurrently on a single chip — detection, classification, and tracking in parallel.

99% Accuracy Retention

INT8 quantization via SDK qb maintains 99% of original model accuracy. No retraining required.

400+ Validated Models

Tested across CNN, RNN, and transformer architectures including ResNet, YOLO, MobileNet, EfficientNet, and more.

Framework Support

SDK qb compiles models from PyTorch, TensorFlow, ONNX, TFLite, and Keras with no architecture changes.

Energy Efficient

Maximizes data reuse and minimizes memory access for best-in-class performance per watt at the edge.

Programmable Architecture

Supports state-of-the-art and custom model architectures. DNN-based architectural blocks adapt to diverse workloads.

Specifications

Technical Details

AI Performance80 TOPS (INT8)
NPU Cores8 cores @ 1.25 GHz
CPU4× U74 RISC-V
TDP25 W
Memory16 GB LPDDR4X (optional 32 GB)
Memory Bandwidth66.7 GB/s
L2 Cache2 MB
Host InterfacePCIe Gen4 x8
Concurrent ModelsUp to 32
Validated Architectures400+

Applications

Use Cases

Robotics & AMRs

Real-time object detection and path planning for autonomous mobile robots and robotic arms.

Smart Factory

Quality inspection and defect detection on the production line. Air-gapped operation keeps data on-premises.

AI Vision & Security

Multi-channel video analytics for surveillance, access control, and perimeter monitoring.

Smart City

Traffic monitoring and pedestrian analytics at the edge — no cloud infrastructure required.

Edge Datacenter

Dense inference workloads with exceptional power efficiency for edge server deployments.

Super Resolution

Real-time image and video upscaling for broadcast, medical imaging, and display enhancement.

Downloads

Documents & Datasheets

Request product documentation and our team will email you the files directly.

ARIES NPU Datasheet

datasheet

Ready to Get Started with the ARIES NPU?

Contact our team for pricing, technical questions, or to request an evaluation unit.