Intel® Advisor User Guide

ID 766448
Date 12/16/2022
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents

Exploring the Parallelism in a Concurrent Node

This example explores the parallelism inherent in a node with unlimited concurrency. The node used is a function_node. A source_node is connected to the function_node, as shown below, and eight items are pushed through to the function_node from the source_node. The function_node has a weight of 1s (weight = 1e6). To ensure there are only timing results from the function_node, the source_node has a comparatively negligible weight of 1e-6s (weight=1). The concurrency specification used is 1, 2, 4, 8.





The results are the following:

  • For a serial execution with only 1 thread, the total time is 8s as the same thread evaluates the tasks one after the other.

  • With 2 threads and 2 overlapping tasks, the total time is 4s.

  • With 4 threads and 8 tasks, the total time is 2s.

  • With 8 threads and 8 tasks, all tasks overlap, giving a total time of 1s.