Artificial intelligence accelerator, device, chip and data processing method

Abstract

The embodiment of the present application discloses an artificial intelligence accelerator, device, chip, and data processing method. The artificial intelligence accelerator includes an instruction analysis unit, an instruction scheduling unit, and a set of processing engines; the instruction analysis unit is used to read the target long instruction, and the target long instruction The instruction includes a first engine identification set, a first configuration information set, a second engine identification set and a second configuration information set; the instruction scheduling unit is used to extract a first processing engine subset from the processing engine set according to the first engine identification set , extract the second processing engine subset from the processing engine set according to the second engine identification set, call the first processing engine subset to execute the first service processing set according to the first configuration information set; when there is a processing engine in the first processing engine subset When the execution is completed, the second processing engine subset is invoked according to the second configuration information set to execute the second service processing set. By adopting the application, the processing time of the neural network model can be shortened.

Description

technical field [0001] This application relates to the field of Internet technology, in particular to the field of artificial intelligence technology, and in particular to an artificial intelligence accelerator, an artificial intelligence acceleration device, an artificial intelligence acceleration chip, and a data processing method. Background technique [0002] Artificial intelligence represented by convolutional neural networks has been successfully applied in the fields of image recognition, object classification and pattern recognition. However, with the continuous optimization of deep learning algorithms, the depth of the model has expanded from a few layers to hundreds of layers, followed by a synchronous expansion of the amount of computation. Further, in certain fields (for example, the medical field or high-frequency video field), the resolution of the picture has reached 2k*2k, or even 5k*5k, and the increase of the picture resolution further increases the amount ...

AI Technical Summary

Problems solved by technology

Furthermore, in specific fields (for example, the medical field or high-frequency video field), the resolution of the picture has reached 2k*2k, or even 5k*5k, and the increase of the picture resolution further increases the amount of calculation

Images