OpImplMode

OpImplMode类表示算子的精度模式，具体定义如下：

enum class OpImplMode : uint32_t {
    // ImplMode support OR operation
    IMPL_MODE_DEFAULT = 0x1,
    IMPL_MODE_HIGH_PERFORMANCE = 0x2,
    IMPL_MODE_HIGH_PRECISION = 0x4,
    IMPL_MODE_SUPER_PERFORMANCE = 0x8,
    IMPL_MODE_SUPPORT_OUT_OF_BOUND_INDEX = 0x10,
    IMPL_MODE_ENABLE_FLOAT32_EXECUTION = 0x20,
    IMPL_MODE_ENABLE_HI_FLOAT32_EXECUTION = 0x40,
    IMPL_MODE_KEEP_FP16 = 0x80,
    IMPL_MODE_RESERVED = 0xFFFFFFFF
};