A method of neural network computation, electronic device, and storage medium
By reversing the data computation order and using a three-stage pipeline mode, the problem of low efficiency in transposing discontinuous data in neural networks is solved, thereby improving computational efficiency and reducing storage space usage.
CN115222027BActive Publication Date: 2026-07-03CAMBRICON SINGGO (NANJING) TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CAMBRICON SINGGO (NANJING) TECH CO LTD
- Filing Date
- 2021-05-11
- Publication Date
- 2026-07-03
Smart Images
Figure CN115222027B_ABST
Abstract
This invention relates to a method, apparatus, electronic device, and storage medium for neural network computation. The neural network computation apparatus of this invention includes a loading module for loading data; a computation module for operator computation; and a storage module for storing data. The computation module interacts with the storage module and the loading module to jointly complete the computational operations specified by the user.
Need to check novelty before this filing date? Find Prior Art