An approximate computation comparison circuit

Abstract

The application discloses an approximate calculation comparison circuit, which is applied to the technical field of computers and integrated circuits and aims at the problem of large area cost existing in the design of existing comparison circuits; the approximate calculation comparison function is realized by charging corresponding capacitors based on charging paths controlled by two inputs and then by competing and comparing. The method is simple, efficient and easy to integrate. It can be seen from the invention content that the application is realized based on a transistor level and the circuit is simple. Compared with the prior art, most of the area cost can be saved.

Description

Technical Field

[0001] This invention belongs to the field of computer and integrated circuit technology, and specifically relates to an approximate calculation and low-power design technology. Background Technology

[0002] As we all know, approximation methods are an important mathematical calculation method. Approximate calculation generally refers to mathematical calculation methods that use approximate numbers as the objects of calculation. An approximate value (an exact number) that approximately represents the value of a quantity is generally called an approximate number. This is frequently used in our lives and engineering practices. In fact, many numbers encountered in real-world problems are approximate numbers; therefore, approximate calculation methods play a vital role in our lives, project research and development, and engineering practices.

[0003] In certain specific application areas, approximation plays a crucial role. In some areas, approximation can achieve a significant reduction in computational complexity while sacrificing a little accuracy. For example, in image processing, our eyes are not very sensitive to image precision, and approximation can be used in many applications. Image processing, as a computationally intensive application, can significantly reduce computational complexity and overhead by using approximation.

[0004] With the development of computer and integrated circuit technologies, integrating a large number of computing functions onto a single chip has become a trend. Under this trend, power consumption has become a significant bottleneck for integrated circuits. Approximate computation, however, can significantly reduce computational complexity by sacrificing a little accuracy, thereby reducing power consumption. As an important tool for low-power design, approximate computation has attracted the attention of scholars, and many researchers have conducted related research.

[0005] In existing approximation calculations, the design of the comparator circuit is generally based on traditional digital circuit methods. The inputs are A and B, and the output is Y. It is typically processed using the operation A + (~B) + 1, i.e., implemented with an adder. The approximation calculation usually involves discarding the lower-order bits of A or B. This circuit is relatively complex and consumes considerable resources. If the encoding of A and B is non-weighted, A and B need to be calculated separately through addition. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention proposes a novel approximate calculation comparison circuit at the transistor level, starting from integrated circuit design.

[0007] The technical solution adopted in this invention is: an approximate calculation comparison circuit, including: charging transistors M1N-M10, charging transistors M2N-M20, transistors M31 and M32, capacitors C1 and C2, and inverters INV1 and INV2;

[0008] The charging transistors M1N-M10 are A-controlled charging transistors. The gate of the A-controlled charging transistor is connected to the control terminal A, the drain of the A-controlled charging transistor is connected to VDD, and the source of the A-controlled charging transistor is connected to the drain of the switching transistor M31.

[0009] The charging transistors M2N-M20 are B-controlled charging transistors. The gate of the B-controlled charging transistor is connected to the control terminal B, the drain of the B-controlled charging transistor is connected to VDD, and the source of the B-controlled charging transistor is connected to the drain of the switching transistor M32.

[0010] The capacitor C1 is connected to the source of transistor M31, the input of inverter INV1, and the output of inverter INV2.

[0011] The capacitor C2 is connected to the source of transistor M32, the input of inverter INV2, and the output of inverter INV1.

[0012] The beneficial effects of this invention are as follows: The approximate calculation comparison circuit of this invention charges corresponding capacitors based on charging paths controlled by two inputs, and then performs a competitive comparison to achieve the approximate calculation comparison function. This method is simple, efficient, and easy to integrate. As can be seen from the invention description, this invention is implemented at the transistor level, resulting in a simple circuit. Compared with existing methods, it can save a significant amount of area overhead. Attached Figure Description

[0013] Figure 1 This is a circuit diagram for approximate calculation comparison according to the present invention. Detailed Implementation

[0014] To facilitate understanding of the technical content of this invention by those skilled in the art, the following description, in conjunction with the accompanying drawings, further illustrates the invention.

[0015] like Figure 1 As shown, the present invention provides an approximate calculation comparison circuit that charges corresponding capacitors based on charging paths controlled by A and B, and then performs a competitive comparison to achieve the approximate calculation comparison function. Specifically, it consists of five parts: charging transistors M1N-M10, charging transistors M2N-M20, transistors M31 and M32, capacitors C1 and C2, and inverters INV1 and INV2.

[0016] The charging transistors M1N-M10 are A-controlled charging transistors, with their respective gates connected to the corresponding control terminals of A, their respective drains connected to VDD, and their respective sources connected to the drain of the switching transistor M31.

[0017] The charging transistors M2N-M20 are B-controlled charging transistors, with their respective gates connected to the corresponding control terminals of B, their respective drains connected to VDD, and their respective sources connected to the drain of the switching transistor M32.

[0018] Transistors M31 and M32 are switch-controlled transistors, with transistor M31 controlling path A and M32 controlling path B. The gates of both transistors M31 and M32 are connected to the control signal C. The drain of transistor M31 is connected to the source of the charging transistors M1N-M10; the source of transistor M31 is connected to capacitor C1, the input of inverter INV1, and the output of inverter INV2.

[0019] The drain of transistor M32 is connected to the source of the charging transistors M2N-M20; the source of transistor M32 is connected to capacitor C2, the input of inverter INV2, and the output of inverter INV1.

[0020] Capacitor C1 is connected to the source of transistor M31, the input of inverter INV1, and the output of inverter INV2. Capacitor C2 is connected to the source of transistor M32, the input of inverter INV2, and the output of inverter INV1. C1 and C2 are capacitors of equal value.

[0021] The input of inverter INV1 is connected to the source of M31, capacitor C1, and the output of inverter INV2. The input of inverter INV2 is connected to the source of M32, capacitor C2, and the output of inverter INV1. Inverters INV1 and INV2 are the same device.

[0022] Y and Z are the outputs, Y = (A>B); Z = (B>A), which achieves an approximate comparison.

[0023] This invention proposes a novel approximate calculation comparison circuit. The circuit charges corresponding capacitors through charging paths controlled by two inputs, then performs a competitive comparison to achieve the approximate calculation comparison function. This invention is implemented at the transistor level, resulting in a simple circuit. Compared to existing methods, it is simple, efficient, and saves a significant amount of area overhead.

[0024] The following is a typical application flow of the circuit of this invention:

[0025] 1. Process selection and installation process library

[0026] Select the appropriate integrated circuit technology and install the corresponding technology library files;

[0027] 2. Circuit Design

[0028] Based on the corresponding process library files, the circuit structure provided by this invention is designed and completed.

[0029] 3. Simulation verification

[0030] Simulation tools are used to verify and analyze the circuit function to achieve the expected functional requirements;

[0031] 4. Layout Design

[0032] With the help of relevant process library files and design rules, complete the layout design and pass physical verification such as DRC and LVS;

[0033] 5. Parameter extraction and post-simulation verification

[0034] Based on the designed layout, the corresponding circuit and parasitic parameters are extracted for simulation verification to achieve the expected functional requirements;

[0035] 6. Perform circuit characterization processing

[0036] Based on the layout, extracted corresponding circuits, and parasitic parameters, the cells are characterized to generate all the necessary files for automatic placement and routing, such as lib and lef.

[0037] 7. Applications in integrated circuit design

[0038] By loading the various files generated from the feature generation into the process library file for integrated circuit design, the content of this invention can be used for integrated circuit design.

[0039] 8. Tape-out verification

[0040] Complete the integrated circuit design, and perform corresponding simulation verification and physical verification, and finally conduct tape-out verification.

[0041] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of the claims of the invention.

Claims

1. An approximate calculation comparison circuit, characterized in that, include: The first group of charging transistors (M1N-M10), the second group of charging transistors (M2N-M20), the first transistor (M31) and the second transistor (M32), the first capacitor (C1) and the second capacitor (C2), the first inverter (INV1) and the second inverter (INV2); The first group of charging transistors (M1N-M10) are A-controlled charging transistors. The gate of the A-controlled charging transistor is connected to the control terminal A, the drain of the A-controlled charging transistor is connected to VDD, and the source of the A-controlled charging transistor is connected to the drain of the first transistor (M31). The second group of charging transistors (M2N-M20) are B-controlled charging transistors. The gate of the B-controlled charging transistor is connected to the control terminal B, the drain of the B-controlled charging transistor is connected to VDD, and the source of the B-controlled charging transistor is connected to the drain of the second transistor (M32). The gates of both the first transistor (M31) and the second transistor (M32) are connected to the control signal C; The first capacitor (C1) is connected to the source of the first transistor (M31), the input of the first inverter (INV1), and the output of the second inverter (INV2); The second capacitor (C2) is connected to the source of the second transistor (M32), the input of the second inverter (INV2), and the output of the first inverter (INV1); The inputs are A and B. The first capacitor (C1) and the second capacitor (C2) are equal capacitors. The first inverter (INV1) and the second inverter (INV2) are the same devices. The sources of the first transistor (M31) and the second transistor (M32) serve as two approximate output terminals, where Y and Z are the outputs, Y = (A>B); Z = (B>A), thus achieving an approximate comparison.

Images