Byte-Oriented Operands; Monitoring Feature - IBM System/370 145 Manual

• Extended Precision Floating Point
The optional floating-point Cllrithmetic feature includes floating
point and extended precision floating-point
instruction~.
Extended
precision is provided for
USE!
in application areas in which the
precision provided by the 10Illg-form floating-point format is not
large enough.
Precision of up to 28 hexadecimal digits, equal to up to 34 decimal
digits, is provided by the eJC:tended precision data format. Extended
precision is achieved by usinlg two doublewords (16 bytes) to
represent an extended precision floating-point number instead of
using one doubleword as is done in long form representation.
Fourteen hexadecimal digits, or up to 17 decimal digits, of
precision are provided by the long floating-point format.
Seven extended precision floating-point instructions are included in
the optional floating-point arithmetic feature,. They provide
addition, subtraction, and multiplication operations for extended
precision data, using a pair of floating-point registers, and the
ability to round from long to short form or from extended to long
form.
An
extended precision divide instruction is not provided;
however, a simulator for this operation is provided in
os
(discussed
in Section 30).
BYTE-ORIENTED OPERANDS
The Model 145 supports a byte boundary alignment facility for
processor storage. The presence of the byte-oriented operand function
allows the processor storage operands of unprivileged instructions (RX
and RS formats) to appear on any byte boundary without causing a
specification program interruption. without this facility, operands
must be aligned on integral bounda.ries, that is, on storage addresses
that are integral multiples of op4~rand lengths. Byte orientation is
standard and does not apply to alignment of instructions or channel
command words
(CCW'
s') •
Use of byte alignment in a proqram degrades instruction execution
performance. However, byte orien1:ation can be used effectively in
commercial processing to eliminatE~ the padding bytes added within
records or to blocked records to E~nsure binary and floating-point field
alignment. The smaller physical record that results from the
elimination of padding bytes requires less external storage and
increases effective I/O data rates. I/O-bound commerical programs, in
which throughput is in almost dirE~ct proportion to the I/O data rate,
can achieve performance improvement by using byte alignment for binary
and floating-point data.
A program written to use byte t~undary alignment will not necessarily
run on a System/360 model that
dOE~S
not have the feature. Therefore,
programs that are to run on both t~he Model 145 and on System/360 models
without byte orientation should bE! written to adhere to integral
boundary rules.
MONITORING FEATURE
The monitoring feature is stanclard on the Model 145. This feature
provides the capability of monito:ring the occurrence of programmed
events. For example, monitoring can be used to perform measurement
functions (how many times a routin.e was executed) or for tracing
functions for the purpose of progz'am debugging (which routines were
executed) •
A Guide to the IBM System/370 Model 145 21