Ara-2 M.2 (M-Key) Module for Generative AI Workloads
- This page contains information on a preproduction product. Specifications and information herein are subject to change without notice. For additional information contact support or your sales representative.
The Ara-2 M.2 (M-Key) Module offers up to 40 effective TOPS (eTOPS) for high-performance, power-efficient AI inference. Designed to handle the compute demands of Large Language Models (LLMs) and transformer-based models, this compact module is ideal for edge AI applications. It runs multiple AI models simultaneously without performance penalties.
Product Details
Features
AI model frameworks supported
- TensorFlow
- PyTorch
- ONNX
Performance
- Up to 40 eTOPS*
- LLaMA-7B: 12 output tokens/sec
- MobileNetV1 SSD: 974IPS (1.03 ms latency)
Security
- Secure Boot
- Root-of-trust processor
Memory (LPDDR4)
- 8 GB or 16 GB (stores all user models)
Operating system support (Runtime)
- Linux
- Windows
Host interface
- 4-lane PCIe Gen 4
Dimensions (W x L)
- 22 mm x 80 mm
Power consumption (typical)
- 3 W (typical workload)
- 8 W (full performance)
Thermal Management (typical)
- Active cooling (heat sink with fan)