Lecture 18: Synchronization: Mutex

1
Lecture 18Synchronization Mutex

• Prof. Kenneth M. Mackenzie
• Computer Systems and Networks
• CS2200, Spring 2003

Includes slides from Bill Leahy
2
Review 1/4Parallel Machines

• Two main HW types
• shared memory
• distributed
• Two main prog. models
• shared memory (threads)
• message-passing
• Note it is possible to make logically shared
  memory atop physically distributed memory

• SMP (Symmetric)
• Distributed

3
Review 2/4 Speedupmetric for performance on
latency-sensitive applications

• Time(1) / Time(P) for P processors
• note must use the best sequential algorithm for
  Time(1) -- the parallel algorithm may be
  different.

linear speedup (ideal)
speedup
typical rolls off w/some of processors
1 2 4 8 16 32 64
occasionally see superlinear...
1 2 4 8 16 32 64
processors
4
Review 3/4 SGI Origin 2000note slides/photos
from SGI web site.

• Hardware DSM ccNUMA
• 1-1024 processors w/shared memory (directory
  limit)

5
Review 4/4 Cray T3Enote slides/photos from
SGI/Cray web site

• Message-Passing non-coherent remote memory
• Up to 2048 processors
• Air- or liquid-cooled
• New product SV1

6
Administrivia

• Make sure your projects compile on acme!
• makefile must work (dont change it!)
• dont rename the file you modify
• dont use // comment syntax ... use / /
• test it with the libraries we give you.
• In other words, just change the one file.

7
Quiz 2 Tuesday!

• Cheat sheet one 8.5 by 11, doublesided sheet
  of paper
• Calculator is useful
• All topics are game well emphasize topics since
  quiz 1
• PRJ2 I/O, interrupts
• HW3 caches
• PRJ3 virtual memory
• Review session! Monday 515 ... someplace
• see the syllabus on the web page

8
Today Synchronization

• specifically mutual exclusion (mutex)
• 1. Problem
• 2. Solutions?
• four low-level solutions
• 3. use low-level to build a blocking primitive
• Generality

9
Terminology

• Process full-blown virtual machine
• register set stack
• protected area of memory
• Thread multiplexed CPU only
• registers set stack
• A process may contain multiple threads. If so,
  all threads see the same address space.