Akida M.2 AKD1000 Card accelerates CNN-based neural network models using BrainChip's ultra energy-efficient, and purely digital, event-based processing architecture. The Akida M.2 AKD1000 can be paired with existing Single-Board Computers to enable edge AI applications. Its low power consumption supports the creation of very compact, low-power, portable and intelligent devices for remote sensors in Healthcare, Enterprise, Industrial, Smart Office, Store and AIoT applications.
What's in the box
- AKD1000 M.2 Card E Key (M.2 2260 form factor)
Specifications
- Form factor: M.2 2260, E Key
- Host interface: PCIe PHY 2-lane
- Memory interface: LPDDR4 via DMA
- CPU: 32-bit ARM Cortex-M4
- NPU: 20 × Akida 1.0 neuron mesh
- Peak performance: 1.5 TOPs (INT8)
- On-chip memory: 8MB near-compute SRAM
- Clock frequency: 300MHz
- Power: ~1W typical
- Operating temperature: 0–70°C
- Thermal: Fanless, no heatsink required
Designed for
- Drop-In AI Acceleration
- Healthcare & Industrial Sensors
- Smart Office & AIoT Applications
- Compact Low-Power Device Design
Quick Start
- Power off host system
- Slide the M.2 card into an M.2 2260 E Key slot at an angle
- Press flat and secure with the retention screw
- Power on and install the Akida driver
