Overview
PCIe Gen3 x16 edge AI accelerator card housing four Axelera Metis AIPU chips with 16GB LPDDR4X memory, delivering ~856 TOPS INT8 aggregate throughput. Designed for high-density edge inference with power-efficient neural processing. Supports active cooling for sustained workloads and integrates with Axelera's MetaWare SDK for model deployment.
Key Features
- 4× Axelera Metis AIPU chips delivering ~856 TOPS INT8 aggregate inference performance
- 16GB LPDDR4X on-card memory (4GB per chip) for model and activation storage
- PCIe Gen3 x16 interface for deployment in standard desktops and edge servers
- Active cooling for sustained edge inference without thermal throttling
- Compatible with MetaWare SDK supporting ONNX, PyTorch, and TensorFlow model formats
Ideal For
Engineers deploying real-time inference pipelines for computer vision, NLP, or generative AI at the edge or in compact form-factor systems.
AI Accelerator
4× Axelera Metis AIPU
Performance
856 TOPS INT8 (214 TOPS per chip)
On-Card Memory
16GB LPDDR4X (4GB per chip)
Interface
PCIe Gen3 x16
Cooling
Active cooling
TDP
75W
$1,770.95
Prices may vary. Verify on vendor site.
Quick Specs
- AI Accelerator
- 4× Axelera Metis AIPU
- Performance
- 856 TOPS INT8 (214 TOPS per chip)
- On-Card Memory
- 16GB LPDDR4X (4GB per chip)
- Interface
- PCIe Gen3 x16
- Cooling
- Active cooling
- TDP
- 75W
Tags
llm-inferencecomputer-visiongenerative-ai