Neural network scheduling method and device

The invention discloses a neural network scheduling method and device, relates to the field of neural networks, and can improve the utilization rate of on-chip storage capacity and improve the running performance of hardware. The method comprises: determining a first batch size corresponding to each...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: XIONG LEJIN, LI SHUCHENG, YUAN HONGHUI, NIKITA, CZERNEGA
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses a neural network scheduling method and device, relates to the field of neural networks, and can improve the utilization rate of on-chip storage capacity and improve the running performance of hardware. The method comprises: determining a first batch size corresponding to each layer in a neural network (S701); based on the first batch size, segmenting the neural network into neural networks comprising at least one first layer group (S702); based on the segmentation result of the first layer group, segmenting the neural network into a neural network comprising at least one second layer group (S703); and scheduling the neural network based on the segmentation result of the second layer group (S704). 一种神经网络调度方法及装置,涉及神经网络领域,能够提升片上存储容量的利用率,提升硬件的运行性能。该方法包括:确定神经网络中每一层分别对应的第一批尺寸(S701);基于第一批尺寸,将神经网络切分为包含至少一个第一层组的神经网络(S702);基于第一层组的切分结果,将神经网络切分为包含至少一个第二层组的神经网络(S703);基于第二层组的切分结果,调度神经网络(S704)。