Difference between revisions of "OpenMP in Small Bites/Scoping"

From HPC Wiki
OpenMP in Small Bites/Scoping
Jump to navigation Jump to search
(Created page with "{{Infobox OpenMP in Small Bites}}")
 
 
(39 intermediate revisions by 3 users not shown)
Line 1: Line 1:
−
{{Infobox OpenMP in Small Bites}}
+
[[Category:Tutorials|Data Scoping (OpenMP)]]<nowiki />
 +
{{DISPLAYTITLE:Data Scoping (OpenMP)}}<nowiki />
 +
{{Syllabus OpenMP in Small Bites}}<nowiki />
 +
__TOC__
 +
 
 +
This video provides an overview on managing one of the most challenging parts of OpenMP: Data Scoping. It discusses the differences between <code>private</code>, <code>firstprivate</code>, <code>lastprivate</code> and <code>shared</code> variables and shows how to implement a scalable reduction.
 +
 
 +
=== Video === <!--T:5-->
 +
 
 +
<youtube width="600" height="340" right>3iU7rd7jLL0</youtube>
 +
 
 +
([[Media:hpc.nrw_03_Introduction-Scoping.pdf | Slides as pdf]])
 +
 
 +
=== Quiz === <!--T:5--> 
 +
 
 +
{{hidden begin
 +
|title = 1. What is the default data scoping of variable in a parallel region when this variable is declared before a parallel region?
 +
}}
 +
<quiz display=simple>
 +
{
 +
|type="()"}
 +
+ Click and submit to see the answer
 +
|| shared
 +
</quiz>
 +
{{hidden end}}
 +
 
 +
{{hidden begin
 +
|title = 2. What is the data scoping of the variables <code>a</code>, <code>b</code>, <code>c</code> and <code>d</code> in the following code snippet in the parallel region? What is printed when executing the code? <br>
 +
<code>int a = 0;<br /> int b = 23;<br /> int c = -3;<br /> # pragma omp parallel num_threads(2) private(a) reduction(+:c)<br /> { </code> <div style="margin-left: 1em;"><code>int d = omp_get_thread_num();<br /> a = 42 + d;<br /> # pragma omp critical<br /> b = 1;<br /> c += a + b;</code></div> <code>} <br />c = c / 2;<br /> printf("a=%d, b=%d, c=%d\n", a, b, c) </code><br />
 +
}}
 +
<quiz display=simple>
 +
{
 +
|type="()"}
 +
+ Click and submit to see the answer
 +
||{{Note|'''a: private, b: shared, c: reduction, d: private '''<br />'''output: <code>a:0, b:1, c:42</code>'''}} 
 +
</quiz>
 +
{{hidden end}}
 +
 
 +
{{hidden begin 
 +
|title = 3. The following code snippet is wrong due to a missing data sharing attribute. How can you fix it? <br>
 +
<code> int i, int s = 0; <br /> # pragma omp parallel for <br /> for (i = 1; i < 100; i++){</code> <div style="margin-left: 1em;"><code> s = s + a[i];</code></div> <code>}  </code> <br />
 +
}}
 +
<quiz display=simple>
 +
{
 +
|type="()"}
 +
+ Click and submit to see the answer
 +
|| {{Note|'''Add <code>reduction(+:s)</code> clause to the parallel for construct'''}}
 +
</quiz>
 +
{{hidden end}}

Latest revision as of 09:54, 18 January 2021

Tutorial
Title: OpenMP in Small Bites
Provider: HPC.NRW

Contact: tutorials@hpc.nrw
Type: Multi-part video
Topic Area: Programming Paradigms
License: CC-BY-SA
Syllabus

1. Overview
2. Worksharing
3. Data Scoping
4. False Sharing
5. Tasking
6. Tasking and Data Scoping
7. Tasking and Synchronization
8. Loops and Tasks
9. Tasking Example: Sudoku Solver
10. Task Scheduling
11. Non-Uniform Memory Access

This video provides an overview on managing one of the most challenging parts of OpenMP: Data Scoping. It discusses the differences between private, firstprivate, lastprivate and shared variables and shows how to implement a scalable reduction.

Video

( Slides as pdf)

Quiz

Collapse
1. What is the default data scoping of variable in a parallel region when this variable is declared before a parallel region?

Click and submit to see the answer

Collapse
2. What is the data scoping of the variables a, b, c and d in the following code snippet in the parallel region? What is printed when executing the code?
int a = 0;
int b = 23;
int c = -3;
# pragma omp parallel num_threads(2) private(a) reduction(+:c)
{
int d = omp_get_thread_num();
a = 42 + d;
# pragma omp critical
b = 1;
c += a + b;
}
c = c / 2;
printf("a=%d, b=%d, c=%d\n", a, b, c)

Click and submit to see the answer

Collapse
3. The following code snippet is wrong due to a missing data sharing attribute. How can you fix it?
int i, int s = 0;
# pragma omp parallel for
for (i = 1; i < 100; i++){
s = s + a[i];
}

Click and submit to see the answer

Retrieved from "https://hpc-wiki.info/hpc/index.php?title=OpenMP_in_Small_Bites/Scoping&oldid=4560"