Data processing method and data processing device based on PCM compression coding

Abstract

The invention discloses a data processing method and a data processing device based on PCM compression coding. The data processing method based on PCM compression coding comprises the steps that saturation operation is conducted on an input divisor x, and r1 and r2 meeting the PCM compression coding input range are obtained; according to a PCM compression coding rule, compression coding is carriedout on the r2 to obtain a code s1, and a shift value s2 is set according to the s1; the lookup table is inquired by utilizing the coded value s1 to obtain p1 and p2; y is obtained through multiply-accumulate operation according to r1, p1 and p2; y is subjected to displacement processing, y2 is obtained by moving the y leftwards by 32 bits, and y3 is obtained by moving the y2 rightwards by s2 bits; and a dividend z is input, and the dividend z is multiplied by y3 to obtain a division operation result. By the adoption of the technical scheme, rapid non-uniform segmentation compression can be conducted on the input number, shift compression processing is conducted according to the PCM compression coding value, errors caused by linear fitting parameter quantization are reduced accordingly, and rapid and accurate data processing is achieved through PCM compression coding, table look-up, multiply-accumulate operation and shift processing.

Description

technical field [0001] The invention belongs to the field of data processing, and in particular relates to a data processing method and a data processing device based on PCM compression coding. Background technique [0002] Traditional division implementation algorithms can be divided into two types: cyclic shift and subtraction, and divisor multiplication by reciprocal. [0003] The cyclic shift subtraction method mainly uses the cyclic shift subtraction method to simulate the hand calculation process and convert division into a series of subtractions. The divisor inversion multiplication method is a method to quickly realize division with the help of a multiplier, but as a fitting algorithm, the implementation of the piecewise fitting of the divisor will affect the number of cycles and precision required for the division. Usually, the smaller the number of segments is, the fewer operation cycles are required but the lower the precision of division is; the larger the numbe...

AI Technical Summary

Problems solved by technology

The divisor inversion multiplication method is a method to quickly realize division with the help of a multiplier, but as a fitting algorithm, the implementation of the piecewise fitting of the divisor will affect the number of cycles and precision required for the division. Usually, the smaller the number of divisions, the fewer the number of operation cycles required but the lower the precision of division, the larger the number of divisions, the higher the precision of division but the greater the number of cycles required

[0004] However, both algorithms have certain problems. If the cyclic shift and subtraction method is implemented using basic subtraction operations, more cycles are required, or a high-base SRT algorithm circuit needs to be specially designed to achieve hardware acceleration.

The method of multiplying the reciprocal of the divisor lies in the process of using polynomial fitting to approximate the reciprocal operation. The fitting accuracy of different input intervals is quite different, and the operation of segmenting according to the input is more complicated, often resulting in the need for multiple comparison operations. To complete the judgment of segment interval

Images