Computing circuitry
By designing an array-arranged computing circuit, utilizing ferroelectric transistors to enter a resistive state under write voltage drive, and adjusting the voltage through pre-charging and output circuits, the problems of low integration and high power consumption of TNN array units are solved, realizing a computing circuit with high integration and low power consumption, and improving the inference efficiency of deep neural networks.
Patent Information
- Application Number
- CN202311262056.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-27
- Publication Date
- 2026-05-12
- Estimated Expiration
- 2043-09-27
AI Technical Summary
Existing ternary neural network (TNN) array units have a large number of components and a large area, resulting in low integration and high power consumption, which cannot effectively improve the inference efficiency of deep neural networks.
The computational circuit, which employs an array arrangement, includes a multiplication computation circuit, a pre-charge circuit, and a computation result output circuit. It utilizes the first and second ferroelectric transistors to enter the target resistive state under the drive of the write voltage, adjusts the computation output node to the target voltage through the pre-charge circuit, and outputs the voltage through the computation result output circuit to obtain the computation result.
It improves the integration of computing circuits, reduces power consumption, is suitable for ternary neural network calculations, reduces the use of components, and thus improves computing efficiency.