N Series: Feature-map for AI

    N-series IPs are collection of propietary algorithm which are used for real-time compress/decompress featuremap data.These IPs/algorithm are designed for neural network(AI) device, which facilitate temporal storage efficiency of featuremap data.AI edge devices and end products embedded NPU may benefit from N-series IPs.

    N-series IPs are featured by lossless/lossy bi-direction support, tiny hardware resources, friendly IP integration, and flexible access/store compressed bitstream.Feature support/algorithm are tailored for neural network architecture via TITC engineer team.

成果圖1-AI.png
成果圖2-AI.png
系統圖-AI.png

◇◆TITC N-Series IP

表格-AI.png