Learning OpenMP (6) together -- Atomic of Thread Synchronization

I. Introduction

In OpenMP, the thread synchronization mechanism includes the mutex lock synchronization mechanism and event synchronization mechanism. This section describes the atomic method in the mutex lock synchronization mechanism.

 

Ii. mutex lock synchronization Concept

The concept of mutex synchronization is similar to the criticalsection in windows, the mutex in Windows and Linux, and the semtake semgive in VxWorks (the semaphore is full during initialization ), that is, to protect a piece of code operations, to ensure that only one thread can execute this section of code at the same time.

 

Iii. Atomic (atomic) operation syntax

# Pragma OPM atomic

X <+ or * or-or * Or/or & or | or <or> = expr

(For example, x <= 1; or x * = 2 ;)

Or

# Pragma OPM atomic

X ++ // or X --, -- X, ++ x

It can be seen that the atomic operation only applies to two situations:

1. Auto increment/Subtraction

2. x <operations listed above> = expr

 

Iv. Example

# Include <iostream> <br/> # include <OMP. h> // header file to be included in OpenMP programming </P> <p> int main () <br/>{< br/> int sum = 0; </P> <p> STD: cout <"before:" <sum <STD: Endl; </P> <p> # pragma OMP parallel for <br/> for (INT I = 0; I <20000; ++ I) <br/>{< br/> # pragma OMP atomic <br/> sum ++; <br/>}</P> <p> STD :: cout <"after:" <sum <STD: Endl; <br/> return 0; <br/>}

20000 output.

If the # pragma OMP atomic declaration is removed, the output value is uncertain.

 

Section 5

This section describes the usage of atomic and the usage of critical section.