Parallel logic operation method based on multi-bit nonvolatile memory and full adder

Abstract

The invention provides a parallel logic operation method based on a multi-bit nonvolatile memory and a full adder, the input is different levels and device nonvolatile storage states, and the output is the device nonvolatile storage state obtained based on the logic operation rule. The parallel logic operation method is suitable for any nonvolatile memory which has a multi-bit storage state and can be converted between different storage states. Based on the method, 16 complete Boolean logic functions can be completed by using one memory under two-step operation, and multiple sets of logic functions can be realized in parallel in the operation process. The N-bit full adder designed on the basis only needs three storage devices to carry out (3N+2)-step operation, so that the addition operation can be realized, and the N-bit full adder has efficient operational capability.

Description

technical field [0001] The invention relates to the technical field of integration of storage and calculation, and more specifically, relates to a method for realizing parallel logic operation based on a multi-bit nonvolatile memory, and a full adder designed based on the method. Background technique [0002] In the era of big data, low energy consumption, high-efficiency storage and processing of massive data is one of the keys to promoting the development of artificial intelligence, Internet of Things and other technologies in the future. However, modern computers adopt an architecture in which data storage and processing are separated, requiring data to be transferred between different memories and processors, which brings additional delay and energy consumption, which is called the "von Neumann bottleneck". In-memory computing technology, that is, the technology of realizing data calculation in memory, is considered to be an effective way to break through this bottleneck...

AI Technical Summary

Problems solved by technology

However, the data of these traditional digital operation units is volatile, and the data needs to be obtained by accessing storage units other than the operation unit, which leads to limited bandwidth and time-consuming and energy-consuming calculations.

Recently, people have developed a method of in-memory logic operation based on non-volatile memory, but Boolean logic is still implemented based on 1-bit memory, so a large number of memory devices and complicated operation steps are required

Images