Optimize
Optimize models with sparsity regularization and quantization aware training
SPU-001
Tiny enough to fit into the smallest devices and efficient enough to last all day running powerful AI applications on-device in real time!
Co-Processor
- 1 MB of on-chip SRAM (10 MB effective with sparsity)
- 22nm process
- Sub-mW for speech, audio, and 1-D data
- SPI interface
Hardware IP
SoC Integration
- 512 kB of on-chip SRAM per core (5 MB effective with sparsity)
- Sharable memory
- 22nm process | portable to any process node
- Fractional or multi-core configurations
- AXI interface
4-core Configuration
AI-ADAM100
SPU + MCU for on-device voice cleanup and flexible commands in a single affordable package
Specifications:
SPU-001 Evaluation
Evaluation Board (EVB)
PCB Board with SPU-001 WLCSP for integrating with external host systems
- Full access to SPU pins on PMOD header
- 16 Mb onboard SPI Flash
- 1.8-3.3V operation
- Headers for easy power measurement
Evaluation Kit 2 (EVK2)
EVB + Tympan open source hearing aid platform and Tympan connector board
- Host processor Cortex-M7 with FPU
- 2x onboard MEMS microphones
- 3.5mm audio jacks for input/output
- Micro SD card reader
Software Development Kit
Our Software Development Kit (SDK) allows developers to get started with minimal barriers. The SDK provides advanced sparse optimization tools, a hardware simulator, and the Femtocrux compiler
Deploy models to the SPU with or without fine-tuning. Achieve high efficiency for dense models, and even higher efficiency for sparse models
Available Ready-to-Deploy Models
AI Noise Cancellation
- Real-time Face-to-Face
- Next-gen Hearing-aid
- Conference / Phone calls
- Extreme Noise
AI Voice Interface
- Wakeword Detection
- Local Commands
- Sentence Intent Understanding
- Voice Cleanup
