09f_cpe431_chap7_6

# 09f_cpe431_chap7_6 - UAH Chapter 7 (4th Ed.) CPE 431/531...

This preview shows pages 1–3. Sign up to view the full content.

1 1 Electrical and Computer Engineering CPE 431/531 Chapter 7 – Multicores, Multiprocessors and Clusters Dr. Rhonda Kay Gaede UAH Electrical and Computer Engineering Page 2 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.1 Introduction - Motivation Why multiprocessors? ____________________ ____________________ ____________________ Electrical and Computer Engineering Page 3 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.2 The Difficulty of Creating Parallel Processing Programs The difficulty with parallelism is not the _________, it’s the ________. Why is it difficult to write parallel processing programs that are fast? ____________ ____ ___________ ___________ ____ ____________ _________ ________ ___ Electrical and Computer Engineering Page 4 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.2 The Difficulty of Creating Parallel Processing Programs - Speedup Challenge Suppose you want to achieve a speedup of 90 times faster with 100 processors. What percentage of the original computation can be sequential? Electrical and Computer Engineering Page 5 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.2 The Difficulty of Creating Parallel Processing Programs - Bigger Problem Suppose you want to perform two sums: one is a sum of 10 scalar variables and one is a matrix sum of a pair of two- dimensional arrays, size 10 by 10. What speedup do you get with 10 versus 100 processors ? Electrical and Computer Engineering Page 6 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.2 The Difficulty of Creating Parallel Processing Programs - Bigger Problem Next, calculate the speed-ups assuming the matrices grow to 100 by 100 Strong scaling Weak scaling

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
2 Electrical and Computer Engineering Page 7 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.2 The Difficulty of Creating Parallel Processing Programs – Balancing Load To achieve the speed-up of 91 on the previous larger problem with 100 processors, we assumed the load was perfectly balanced (each processor did 1 % of the work). Instead, show the impact on speed-up if one processor’s load is higher than all the rest. Calculate at 2% and 5%. Electrical and Computer Engineering Page 8 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.3 Shared Memory Multiprocessors A shared memory multiprocessor (SMP) is one that offers the programmer a _____ ________ ______ _____ across all processors Processor communicate through ______ __________ in memory. SMPs come in two flavors – ______ – ______ Processors need to coordinate when sharing data, this process is called ______________, processors must acquire a _____ Electrical and Computer Engineering Page 9 of 23 UAH CPE 431/531 Chapter 7 (4 th Ed.) 7.3 Shared Memory Multiprocessors - Parallel Program Suppose we want to sum 100,000 numbers on an SMP with UMA. Let’s assume we have 100 processors
This is the end of the preview. Sign up to access the rest of the document.

## This note was uploaded on 07/22/2011 for the course CPE 431 taught by Professor Staff during the Fall '09 term at University of Alabama - Huntsville.

### Page1 / 4

09f_cpe431_chap7_6 - UAH Chapter 7 (4th Ed.) CPE 431/531...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online