Mathews01

Leveraging its robust computing power and exceptional video processing capabilities, the Mathews01 is ideal for smart city security monitoring, enabling real-time facial recognition and vehicle detection; it supports intelligent transportation by analyzing traffic conditions and monitoring violations in real time; in cloud computing, it provides efficient inference computing power for cloud services; it helps short-video platforms with intelligent transcoding and content moderation; and in industrial vision, it assists with product quality inspection and defect detection.

Detailed information

  • Processor: Integrated NWE heterogeneous GPGPU with dedicated AI acceleration cores

  • Memory: 16GB HBM2E high-bandwidth memory (3.6TB/s peak bandwidth)

  • Performance:

     1、160 TOPS (INT8) / 80 TFLOPS (FP16) compute density

     2、4th-gen tensor cores optimized for sparse neural networks

  • Video Processing:

     1、128x 1080p60 encoding / 120x decoding channels

     2、Full codec support: HEVC (H.265), H.264, AV1, AVS2, VP9

     3、8K60 decode/encode with VVC hardware acceleration

  • Software Stack:

    1、Proprietary NebulaFlow SDK v3.5 with kernel-level optimizations

    2、Native support for TensorFlow, PyTorch, ONNX Runtime

    3、Containerized deployment via Kubernetes and Docker


Leveraging its robust computing power and exceptional video processing capabilities, the Mathews01 is ideal for smart city security monitoring, enabling real-time facial recognition and vehicle detection; it supports intelligent transportation by analyzing traffic conditions and monitoring violations in real time; in cloud computing, it provides efficient inference computing power for cloud services; it helps short-video platforms with intelligent transcoding and content moderation; and in industrial vision, it assists with product quality inspection and defect detection.

About Us

We are a leading provider of artificial intelligence (AI) chip acceleration solutions spanning cloud and edge computing environments. Through innovative computing architecture design, we deliver deep support for the development and deployment of end-to-end sparsified neural networks, building universal AI computing platforms that combine ultra-high computational throughput with ultra-low power efficiency.

Contact Us

barrack@mathews-tech.com.cn