Processor, variable-precision convolutional network calculation method and calculation equipment

The embodiment of the invention provides a processor, a variable-precision convolutional network calculation method and a variable-precision convolutional network calculation device, the processor abstracts the densest calculation in a convolutional neural network into a convolutional network instruction through an instruction set expansion mode, hardware acceleration of convolutional network calculation on the processor is realized, and the calculation efficiency is improved. And the calculation speed and efficiency of the convolutional neural network are improved. Besides, the convolutional network instruction comprises a precision parameter, convolutional network calculation of variable precision data can be supported, and the applicability of the processor is improved. 本说明书实施方式提供了一种处理器、可变精度的卷积网络计算方法和计算设备，该处理器通过指令集扩展的方式，将卷积神经网络中最密集的计算抽象为卷积网络指令，实现了对处理器进行卷积网络计算的硬件加速，提升了卷积神经网络的计算速度和效率。此外，所述卷积网络指令包括精度参数，可以支持可变精度数据的卷积网络计算，提高了处理器的适用性。