iTP Active Transaction Pages (iTP ATP) Programmer's Guide

Introduction
iTP Active Transaction Pages (iTP ATP) Programmer’s Guide—522292-002
1-6
IEEE Floating-Point Support
IEEE Floating-Point Support
Depending upon your system, you have the option of choosing between IEEE floating-
point format and Compaq floating-point format in your native C and C++ programs for
performing floating-point arithmetic.
The data format is different between the two floating-point options. Data interchange
with systems other than Compaq systems is easier with IEEE floating-point format than
with Compaq floating-point format.
Differences Between IEEE and Compaq Floating-Point Formats
•
IEEE floating-point data format is an industry standard, whereas Compaq’s floating-
point format is Compaq’s implementation of floating-point arithmetic.
•
IEEE and Compaq floating-point data formats have a different range of values and
different precisions.
•
IEEE and Compaq floating-point data formats have different internal layouts (for
example, the exponents are in different bit fields).
•
IEEE floating-point format enables your application to take advantage of the greater
performance provided by the floating-point instructions available in some processor
hardware.
•
IEEE floating-point format is good for porting applications.
•
IEEE floating-point has good default handling of overflow, underflow, divide-by-
zero, and invalid operation.
•
IEEE floating-point denormalized numbers avoid computational problems that arise
from very small numbers as intermediate results in computations.
System Requirements
The ATP 2.1 IEEE floating-point support is only available when the target
hardware/software system supports IEEE-floating point format. The following list
shows the basic system requirements for IEEE floating-point support.
•
IEEE floating-point format runs on S7000 and S72000 or above processors. It does
not run on S70000 processors.
•
IEEE floating-point format requires the G06.06 or later release of native C and C++
compilers and the nld utility for linking.
The processor type and operating system release version used are determined during the
installation process, so be sure you install ATP on a CPU on which IEEE floating-point
format can run (refer to Before You Begin the Installation
on page 2-2). The installation
process automatically selects the correct libraries and builds the final executable binary.
For further information, refer to the Guardian Programmer’s Guide.