QNN HTA Op Support Revision History

Introduced in QNN SDK Version

Description

2.15.0
  • Added Support of 8bit and 16 bit ElementWiseBinary Op with sub operations:

    • ElementWiseAdd, ElementWiseMultiply, ElementWiseSubtract

  • Added Support of 8bit and 16 bit ElementWiseNeuron Op with sub operations:

    • HardSwish, Relu, ReluMinMax, Sigmoid, Tanh

1.7.0
  • Added Support for 16 bit for following Operations:

    • AveragePool, MaxPool

    • ElementWiseSum, ElementWiseSub, ElementWiseProd

    • ResizeNearestNeighbor, ResizeBilinear

    • Conv2D, DepthwiseConv2D, TransposeConv2D

    • Relu, Relu6, Prelu, Tanh, Sigmoid, HardSwish

    • Concat

    • Softmax