Browse Prior Art Database

Method and System for Generating and Executing Test Program Populated With Floating Point Data Values for Testing a Floating Point Unit

IP.com Disclosure Number: IPCOM000192767D
Original Publication Date: 2010-Feb-02
Included in the Prior Art Database: 2010-Feb-02
Document File: 5 page(s) / 129K

Publishing Venue

IBM

Abstract

A method and system for generating and executing test cases populated with Floating Point data values for testing a Floating Point unit in a processor is disclosed. The test cases are populated with Floating Point data values based on a set of categories specified by a user.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 54% of the total text.

Page 1 of 5

Method and System for Generating and Executing Test Program Populated With Floating Point Data Values for Testing a Floating Point Unit

Disclosed is a method and system for generating and executing test cases populated with Floating Point data values for testing a Floating Point Unit in a processor.

Testing a Floating Point Unit in a processor utilizes a large range of Floating Point data values. However, the number of data values that can be initialized in a register is comparatively low. Such a scenario inhibits the initialization of the register with a large range of Floating Point data values for effective testing. Accordingly, the method and system disclosed herein generates test cases populated with Floating Point data values based on a set of categories specified by a user. Thereafter, a value from each of the set of categories is retrieved and initialized in the register.

The set of categories which signify the type of Floating Point data values that needs to be initialized in the registers is specified by the user using a Configuration File. The layout of the Configuration File is as shown in Fig. 1. The Configuration File specifies the set of categories and the number of Floating Point Data values in each of the set of categories that needs to be tested. As shown in Fig. 1, the Configuration File has a set of default categories which are utilized when the user does not specify any category. The default categories can be NAN (

N

Normal numbers, denormal numbers, zeros, infinities etc. The number of values in

each category can be zero or any positive integer. The User Patterns in the Configuration File are the set of categories that the user specifies for Floating Point Unit testing.

ot a

Numbe

r

),

1

Page 2 of 5

Figure 1

Based on the set of categories specified in the Configuration File, Buckets of Floating Point Data values are generated. There can be Buckets of different types, holding different categories of Floating Point data values. The values are generated based on the "Number of Values"

parameter specified in the Configuration File. For example, the Configuration File may have

Normal Numbers

                           500, Minimum Denormal Numbers - 256; Zeros - 1000; Infinities - 0; User Case Pattern 1

entries such as NAN - 0,

Normal numbers, Mi...