Ceg4131 models

1

Parallel Computer Models

CEG 4131 Computer Architecture III

Miodrag Bolic

2

Overview

• Flynn’s taxonomy• Classification based on the memory arrangement• Classification based on communication• Classification based on the kind of parallelism

– Data-parallel – Function-parallel

3

Flynn’s Taxonomy

– The most universally excepted method of classifying computer systems

– Published in the Proceedings of the IEEE in 1966

– Any computer can be placed in one of 4 broad categories

» SISD: Single instruction stream, single data stream

» SIMD: Single instruction stream, multiple data streams

» MIMD: Multiple instruction streams, multiple data streams

» MISD: Multiple instruction streams, single data stream

4

SISD

Processing element (PE)

Main memory (M)

Instructions

Data

Control Unit PE MemoryPE

IS

IS DS

5

Applications:• Image processing• Matrix manipulations• Sorting

SIMD

6

SIMD Architectures

• Fine-grained– Image processing application– Large number of PEs– Minimum complexity PEs– Programming language is a simple extension of a sequential

language

• Coarse-grained– Each PE is of higher complexity and it is usually built with

commercial devices– Each PE has local memory

7

MIMD

8

MISD

Applications:• Classification • Robot vision

9

Flynn taxonomy

– Advantages of Flynn

» Universally accepted

» Compact Notation

» Easy to classify a system (?)

– Disadvantages of Flynn

» Very coarse-grain differentiation among machine systems

» Comparison of different systems is limited

» Interconnections, I/O, memory not considered in the scheme

10

Classification based on memory arrangement

PE1 PEn

Processors

Interconnectionnetwork

Shared memory

Shared memory - multiprocessors

I/O1

I/OnPE1

Interconnectionnetwork

M1

P1

PEn

Mn

Pn

Message passing - multicomputers

11

Shared-memory multiprocessors

• Uniform Memory Access (UMA)• Non-Uniform Memory Access (NUMA)• Cache-only Memory Architecture (COMA)

• Memory is common to all the processors.• Processors easily communicate by means of

shared variables.

12

The UMA Model

• Tightly-coupled systems (high degree of resource sharing)

• Suitable for general-purpose and time-sharing applications by multiple users.

P1

$

Interconnection network

$

Pn

Mem Mem

13

Symmetric and asymmetric multiprocessors

• Symmetric: - all processors have equal access to all peripheral devices.- all processors are identical.

• Asymmetric: - one processor (master) executes the operating system- other processors may be of different types and may be dedicated to special tasks.

14

The NUMA Model

• The access time varies with the location of the memory word.• Shared memory is distributed to local memories.• All local memories form a global address space accessible by

all processors

P1

$

Interconnection network

$

Pn

Mem Mem

Distributed Memory (NUMA)

Access time: Cache, Local memory, Remote memory

COMA - Cache-only Memory Architecture

15

Distributed memory multicomputers

• Multiple computers- nodes

• Message-passing network

• Local memories are private with its own program and data

• No memory contention so that the number of processors is very large

• The processors are connected by communication lines, and the precise way in which the lines are connected is called the topology of the multicomputer.

• A typical program consists of subtasks residing in all the memories. PE

Interconnectionnetwork

M

PE

M

PE

M

PE

M

PE

M

PE

M

16

Classification based on type of interconnections

• Static networks

• Dynamic networks

17

Interconnection Network [1]

• Mode of Operation (Synchronous vs. Asynchronous)

• Control Strategy (Centralized vs. Decentralized)

• Switching Techniques (Packet switching vs. Circuit switching)

• Topology (Static Vs. Dynamic)

18

Classification based on the kind of parallelism[3]Parallel

architecturesPAs

Data-parallel architectures Function-parallel architectures

Instruction-level

PAs

Thread-level

PAs

Process-levelPAs

ILPS MIMDs

Vectorarchitecture

Associative

architecturearchitectureand neural SIMDs Systolic Pipelined

processorsProcessors)

processorsVLIWs Superscalar Ditributedmemory

(multi-computer)

Sharedmemory(multi-MIMD

DPs

19

References

1. Advanced Computer Architecture and Parallel Processing, by Hesham El-Rewini and Mostafa Abd-El-Barr, John Wiley and Sons, 2005.

2. Advanced Computer Architecture Parallelism, Scalability, Programmability, by  K. Hwang, McGraw-Hill 1993.

3. Advanced Computer Architectures – A Design Space Approach by Desco Sima, Terence Fountain and Peter Kascuk, Pearson, 1997.

20

Speedup

• S = Speed(new) / Speed(old)

• S = Work/time(new) / Work/time(old)

• S = time(old) / time(new)

• S = time(before improvement) /

time(after improvement)

21

Speedup

• Time (one CPU): T(1)

• Time (n CPUs): T(n)

• Speedup: S

• S = T(1)/T(n)

22

Amdahl’s Law

The performance improvement to be gained from using some faster mode of execution is limited by the fraction of the time the faster mode can be used

23

20 hours

200 miles

A B

Walk 4 miles /hourBike 10 miles / hourCar-1 50 miles / hourCar-2 120 miles / hourCar-3 600 miles /hour

must walk

Example

24

20 hours

200 miles

A B

Walk 4 miles /hour 50 + 20 = 70 hours S = 1Bike 10 miles / hour 20 + 20 = 40 hours S = 1.8Car-1 50 miles / hour 4 + 20 = 24 hours S = 2.9Car-2 120 miles / hour 1.67 + 20 = 21.67 hours S = 3.2Car-3 600 miles /hour 0.33 + 20 = 20.33 hours S = 3.4

must walk

Example

25

Amdahl’s Law (1967)

: The fraction of the program that is naturally serial

• (1- ): The fraction of the program that is naturally parallel

26

S = T(1)/T(N)

T(N) = T(1) + T(1)(1- )

N

S = 1

+ (1- )

N

=N

N + (1- )

27

Amdahl’s Law