AdaptiveStorm

  • Dual-core RISC-V ASIC

    • Watchdog mode for ultra-low power

    • Nullhop mode for high throughput

  • Adaptiva: co-processor for AI workloads

    • Watchdog mode: < 1mW. (<100uW in some use-cases)

    • Nullhop mode: 10 TOPs/W

  • Supported neural networks

    • Convolutional neural networks including resNet, VGG, MobileNet

    • Recurrent neural networks with multiple layers

  • Sensor data processing

    • Supports standard audio and video interfaces such as MIPI, I2C, SPI

    • Direct interface to multiple channels of raw sensor data, ex: audio and accelerometers

Intelligent workload management

Our compiler libraries optimize for energy-efficiency. Developers can use popular AI libraries such as PyTorch or Tensorflow to train neural networks. These networks will be ported to Adaptiva by our compilers, which also handles workload planning to ensure maximum energy-efficiency.

Fast and easy to use software framework

Our compilers make deploying new models on our chips are as easy as it can get.

Location
Dammstrasse 16, 6300 Zug (HQ),
Josefstrasse 219, 8005 Zurich

Social media
LinkedIn
Twitter