### Dispatch_Compared ### |
|
=========================================================================== |
USAGE: Dispatch_Compared [<n_benchmarks> [<n_folds> [<n_iterations>]]] |
|
All arguments are optional, though the earlier arguments must be specified to use the later ones. |
|
n_benchmarks - argument passed to dispatch_benchmark; more runs reduces fluctuations |
|
n_folds - how many times to run the work_function, to increase computation vs. overhead |
|
n_iterations - maximum # of times to iterate each API |
|
|
=========================================================================== |
DESCRIPTION: |
|
This sample code uses dispatch_benchmark to time the performance of |
a relatively compute-intensive loop using different APIs: |
- simple for loop |
- GCD: dispatch_apply |
- GCD: serial queue (private) |
- GCD: parallel queue (global) |
- GCD: multiple queues |
- POSIX threads |
|
Please note that this is NOT a "macro" benchmark that compares the real-world performance of different implementations. |
Rather, it is a *micro* benchmark that simply shows the overhead of invoking each API that number of times. |
|
Also note that a complete run with the default arguments can take several hours. |
|
=========================================================================== |
BUILD REQUIREMENTS: |
|
Mac OS X v10.6 or later |
|
=========================================================================== |
RUNTIME REQUIREMENTS: |
|
Mac OS X v10.6 or later |
|
=========================================================================== |
PACKAGING LIST: |
|
ReadMe.txt - This document |
main.c - Primary source code file |
Sample_Results.txt - Sample output plus discussion of results |
|
=========================================================================== |
CHANGES FROM PREVIOUS VERSIONS: |
|
Version 1.0 |
- First version. |
|
=========================================================================== |
Copyright (C) 2009 Apple Inc. All rights reserved. |
