Browse Prior Art Database

Floating Point Performance Enhancement Tool

IP.com Disclosure Number: IPCOM000036174D
Original Publication Date: 1989-Sep-01
Included in the Prior Art Database: 2005-Jan-28
Document File: 1 page(s) / 11K

Publishing Venue

IBM

Related People

Urquhart, RJ: AUTHOR

Abstract

Disclosed is a tool (FPPET-Floating Point Performance Enhancement Tool) that was used to improve the floating point performance of the IBM RT PC. Comments are added to an RT assembly listing. These comments are processed by the FPPET which expands them into RT instructions that directly access high function AFPA (Advanced Floating Point Accelerator) capabilities.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Floating Point Performance Enhancement Tool

Disclosed is a tool (FPPET-Floating Point Performance Enhancement Tool) that was used to improve the floating point performance of the IBM RT PC. Comments are added to an RT assembly listing. These comments are processed by the FPPET which expands them into RT instructions that directly access high function AFPA (Advanced Floating Point Accelerator) capabilities.

Enhancements to the AFPA in the area of vector processing are not used by the non-vectorizing RT compilers; thus, an alternate path to these functions was required. The process is described below. First statements are added to an assembler language file. These statements begin with the characters #FP. Following the #FP is a list of positional and key word parameters that describe the AFPA architecture with its vector extensions. Using this format AFPA instructions can be described. After all the #FP statements are added and any other assembler level changes made, the FPPET is executed with the modified assembler file as input. Each line that does not begin with #FP is written to the output unchanged. When a #FP statement is encountered, it also is written to the output file and becomes a comment. Also additional RT code is generated based on the parameters that follow the #FP characters. In this manner vector floating point operations are generated and the maximum capability of the AFPA realized.

1