Method and device for processing floating point numbers

Abstract

The embodiment of the present invention discloses a floating-point number processing method and device. The method includes: obtaining the first target floating-point number X and the second target floating-point number Y; obtaining the algorithm of X and Y; judging whether X and Y are not zero ;If both X and Y are not zero and the order codes of X and Y are not equal, then for order X and Y; the mantissas of X and Y after the order will be segmented respectively, and each segment of data will be converted to single precision Floating-point numbers; respectively calculate the results of the corresponding segment data of the mantissas of X and Y; generate the final mantissa according to the calculated results of each segment of data; then generate the calculation results; and normalize the calculation results. By applying the embodiment of the present invention, the mantissa of the high-precision floating-point number is segmented, and each segment of the mantissa is converted into a single-precision floating-point number for calculation, so that low-end GPU products that do not have the ability to calculate double-precision floating-point numbers have high-precision floating-point numbers. The calculation ability of points has improved the calculation ability of low-end GPU products for high-precision floating-point numbers.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a floating-point number processing method and device. Background technique [0002] GPU (Graphics Processing Unit, graphics processing unit), also known as graphics processor, display core, visual processor, and display chip, is a graphics processing unit that is specially used in personal computers, workstations, game consoles, and some mobile devices (such as tablets, smartphones, etc.) etc.) Microprocessors that work on image operations. [0003] At present, the floating-point number calculation performance of the GPU is much higher than that of the CPU, so the GPU is also mainly used for floating-point number calculations. [0004] However, in consideration of cost and market division, GPU manufacturers will retain the hardware computing unit of double-precision floating-point numbers on high-end GPU products, and reduce the hardware computing unit of double-precision floa...

AI Technical Summary

Problems solved by technology

[0004] However, due to cost and market segmentation considerations, GPU manufacturers will retain double-precision floating-point hardware computing units on high-end GPU products, and reduce double-precision floating-point hardware computing units on mid- and low-end GPU products. Low-end and mid-end GPU products have poor double-precision floating-point calculation capabilities, or even do not have double-precision floating-point calculation capabilities.

Images