IBM GC28-1982-02 manual Latency, Network, Loop, Menu, Passing

latency .

The

time

interval

between

the

instantMPMD

at(MultiplewhichProgram

Multiple

Data)

.

A

parallel

an instruction control unit initiates a call forprogrdatamming model

in

which

different,

but

related,

transmission, and the instant at which the actualprograms

are run on different sets of data.

transfer of data (or receipt of data at the remote

end)

begins.

Latency

is

related

to

the

hardwareMPI .

Message

Passing

Interface;

a

standardized

AP

characteristics

of the

system

and

to

the

for

implementing the message passing model.

different

layers

of software that are involved

in

initiating

the

task

of

packing

and

transmitting

the

data.

N

Licensed

Program

Product

(LPP)

.

A

collection

of

network .

An interconnected group of nodes, lines, a

software

packages,

sold

as

a

product,

that

customers

A

network

provides

the

ability

to

transmit

pay

for

to

license.

It

can

consist

of

terminals.

packages

and

to and receive data from other systems and users

filesets a customer would install. These packages and

filesets

bear

a

copyright

and

are

offered

under

the

a

network,

the

point

where

one

or

terms

and

conditions

of

a

licensing

node

.

(1)

In

agreement.

See

also

filesetandpackage.

functional

units interconnect transmission lines. A

computer

location defined in a network.

(2)

In

ter

LoadLeveler

.

A

job

management

system

that

the

IBM

RS/6000

SP,

a

single

location

or

workstation

i

works

network. An

SP

node

is

a

physical

entity

(a

with

POE to

allow

users

to

run

jobs

and

a

match

processor).

processing needs with system resources, in order to

better

utilize

the

system.

node

ID .

A

string

of

unique

characters

that

identi

local

variable

.

A

variable

that

is

defined

the

node

on

a

network.

and

used

only

in

one

specified

portion

of

a

computer

program.

.

An

operation,

such

as

nonblocking

operation

loop

unrolling

.

A

program

transformation

which

sending

or

receiving

a

message,

which

returns

immediately

whether or not the operation was

makes

multiple

copies

of

the

body

of

a

loop,

placing the

copies

also

within

the

body

of

the

loop.

completed. For example, a nonblocking receive will not

The

loop

trip

but

a

blocking

rece

count

and

index

are

adjusted

appropriately

wait until a message is sent,

so

the

new

receive

will

return

a

status

valu

loop

computes

the

same

values

as

the

wait.

A

nonblocking

original.

This

that indicates whether or not a message was rece

transformation makes it possible for a compiler to take

additional advantage of instruction pipelining, data

cache

effects,

and

software

pipelining.

O

See

alsoptimization.

object

code

.

The

result

of

translating

a

computer

M

program

to

a

relocatable,

low-level

form.

Object

co

contains

machine

instructions,

but

symbol

names

(such

menu .

as

array,

scalar,

and

procedure

names),

are

not

y

A

list of

options

displayed

to

the

user by

a

data

given a location in memory.

processing system, from which the user can

select

an

action

to

be

initiated.

optimization

.

A not strictly accurate but widely us

message

catalog

term

for

program

performance

improvement,

especially

.

A

file

created

using

the

AIX

for performance improvement done by a compiler or

Message

Facility

from

a

message

source

file

that

other program translation software. An optimizing

contains

application

error and

other

messages,

which

is

one

that

performs

extensive

code

compiler

can

later

be

translated

into

other

languages

without

in

order

to

obtain

an

executable

th

transformations

having

to

recompile

the

application

source

code.

faster

but gives the same answer as the ori

runs

message

passing

.

Refers

to

the

process

by

Such

code

transformations,

however,

can

make

code

which

parallel

tasks

explicitly

exchange

program

debugging and performance analysis very difficult

data.

because complex code transformations obscure the

MIMD (Multiple Instruction Multiple Data)

.

A

parallel

correspondence

between

compiled

and

original

source

code.

programming model in which different processors

perform

different

instructions

on

different

sets

of

data.

option flag . Arguments or any other additional

information

that

a

user

specifies

with

a

program n

Also

referred parametersto as

orcommand

line

options.

Glossary of Terms and Abbreviations191