Deep Learning Accelerator
Overview
NEUCHIPS is the first company with a purpose-built recommender system. RecAccelTM N3000 is built on the most advanced process technology, designed to be scalable, and has an optimized hardware and software stack that delivers superior performance while being energy efficient and offering low TCO.
Key Features
- NEUCHIPS Asymmetric Quantization
- NEUCHIPS Advanced Symmetric Quantization
- Patented memory architecture
- Integrated Interface: PCIe Gen3/4/5
- Minimizing TCO
Benefits
- Characteristics
- Embedding-specific memory architecture, allocation and access scheme.
- Application-specific processing pipeline.
- Scalable Multiply-And-Accumulator (MAC) array
- Accelerator Instructions
- Read weight
- Read activation
- Write activation without charge
- Configuration
Block Diagram
Technical Specifications
Foundry, Node
TSMC 7nm
Related IPs
- High performance-efficient deep learning accelerator for edge and end-point inference
- Unified Deep Learning Processor
- Compact 32-bit MCU core for deeply embedded applications and accelerator control
- Deeply Embedded AI Accelerator for Microcontrollers and End-Point IoT Devices
- 2D Vector Graphics Accelerator / GPU (Graphics Processing Unit)
- Bit Block Transfer 2D video accelerator