NeuroCore NPU
128 TOPS, sparse acceleration.
All (6)
AI Accelerators (1)
Interface IP (1)
Security IP (1)
DSP Cores (1)
GPU Cores (1)
CPU Cores (1)
Overview
High-efficiency NPU with 128 TOPS for edge and cloud AI inference.
Advantages
- 128 TOPS performance
- Sparse acceleration
- Low power
- INT4/INT8 support
Key Features
- 128 TOPS
- Sparse engine
- INT4/INT8/FP16
- Transformer support
Deliverables
RTL source
SDK
Model zoo