User Guide

Contents

Threading
Accuracy Levels in Command Line

For each perspective,
Intel® Advisor
has several levels of collection accuracy. Each accuracy level is a set of analyses and properties that control what data is collected and the level of collection details. The higher accuracy value you choose, the higher runtime overhead is added.
In CLI, each accuracy level corresponds to a set of commands with specific options that you should run one by one to get a desired result.
For the
Threading
perspective, you are recommended to run the accuracy levels one by one to get a Threading report.
The following accuracy levels are available:
Comparison / Accuracy Level
Low
Medium
Overhead
1.1x
5 - 8x
Goal
Find candidates for parallelization
Model threading parallelism and check for loop-carried dependencies
Analyses
Survey
Survey + Characterization (Trip Counts) + Suitability + Dependencies
Result
Basic Survey report
Survey report extended with trip count data
Dependencies report
Suitability report with parallel performance modeled for annotated loops
You can generate commands for a desired accuracy level from the
Intel Advisor
GUI. See Generate Command Lines from GUI for details.
There is a variety of techniques available to minimize data collection, result size, and execution overhead. Check Minimize Analysis Overhead.
Consider the following command examples.
Info
: In the commands below, make sure to replace the
myApplication
with your application executable path and name
before
executing a command. If your application requires additional command line options, add them
after
the executable name.

Low Accuracy

First, run the Threading perspective with low accuracy to find candidates for parallelizing based on Survey analysis results.
Run the analysis as follows:
advisor --collect=survey --project-dir=./advi_results -- ./myApplication
You can view the generated results in the Intel Advisor GUI or in the CLI. The loops/functions with high total time are the best candidates for parallelization. Annotate the loops/functions of interest to model parallelism.

Medium Accuracy

Prerequisite
: Annotate loops/functions to model parallelization for. Rebuild the application.
Run the commands as follows:
  1. Run the Survey analysis:
    advisor --collect=survey --project-dir=./advi_results -- ./myApplication
  2. Collect trip count data:
    advisor --collect=tripcounts --project-dir=./advi_results -- ./myApplication
  3. Run the Suitability analysis to model threading parallelism for the annotated loops:
    advisor --collect=suitability --project-dir=./advi_results -- ./myApplication
  4. Run the Dependencies analysis for the annotated loops:
    advisor --collect=dependencies --project-dir=./advi_results -- ./myApplication
You can view the generated results in the Intel Advisor GUI or in the CLI.

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.