High-precision floating-point number simulation method based on domestic heterogeneous many-core platform
A floating-point number and high-precision technology, which is applied in various digital computer combinations, electronic digital data processing, digital computer components, etc., can solve the problems of complex implementation of high-precision floating-point calculations and low calculation efficiency, and achieve the realization of instruction bar Small number, high computational efficiency, and reduced programming difficulty
Pending Publication Date: 2022-03-22
JIANGNAN INST OF COMPUTING TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] The purpose of the present invention is to provide a high-precision floating-point number simulation method based on a domestic heterogeneous many-core platform to solve the problems of complex implementation of high-precision floating-point calculations and low calculation efficiency on a domestic heterogeneous many-core platform
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0017] Embodiment: The present invention provides a high-precision floating-point number simulation method based on a domestic heterogeneous many-core platform, which is suitable for subjects with high requirements for floating-point calculation accuracy and calculation efficiency, and uses scalar to simulate the sign bit and exponent bit of floating-point numbers And the mantissa, and accelerate based on domestic heterogeneous many-core platform instructions, realize the addition, subtraction, multiplication, division, square root, rounding, decimal, type conversion, comparison, absolute value, and inversion of high-precision floating-point numbers function, based on the following configuration:
[0018] Define the high-precision floating-point number as the structure mdouble, and the definition of each member of mdouble meets IEEE754, as follows:
[0019]
[0020] Include the following steps:
[0021] S1. Conversion and representation of initial data types: use data type...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a high-precision floating-point number simulation method based on a domestic heterogeneous many-core platform. The method comprises the following steps: S1, conversion and representation of an initial data type: realizing conversion from a source data type to a high-precision floating-point number by using a data type conversion function; s2, according to specific functions, performing corresponding operation on sign bits, exponent bits and mantissas, and performing acceleration by using instructions; and S3, performing nearby even rounding on the result in the S2, and adjusting the index according to the rounding result to obtain a high-precision floating point operation result. According to the method, the programming difficulty of a user is reduced, the number of implementation instructions of high-precision floating point operation is smaller, and the calculation efficiency is higher.
Description
technical field [0001] The invention relates to a high-precision floating-point number simulation method based on a domestic heterogeneous many-core platform, and belongs to the technical field of high-precision floating-point calculations. Background technique [0002] High-precision floating-point operations are widely used in many large-scale parallel algorithms, but the current domestic heterogeneous many-core platform only supports single-precision and double-precision floating-point operations, and there is no corresponding support for high-precision floating-point instructions. The functions of high-precision floating-point numbers are indirectly realized through arrays, scalars, and complex algorithms, which is complex and has low computational efficiency. Take the multiplication of 80-bit floating-point numbers as an example, use the integer type to represent the sign bit and exponent bit of the 80-bit floating-point number, and use the unsigned long integer type to...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06F7/483G06F9/30G06F15/16
CPCG06F7/483G06F15/161G06F9/30007
Inventor 刘齐文延华王飞董恩铭方燕飞杨小川于康李雁冰王云飞
Owner JIANGNAN INST OF COMPUTING TECH