Analyze the annotated program to check its predicted parallel performance.
The Suitability tool Image may be NSFW.
Clik here to view. runs your serial application's executable and measures its execution to predict the approximate maximum performance of its parallel site(s) and overall program maximum gain.
To speed-up your program, you need to choose the best places to add parallelism, by locating which parallel sites contribute the most to the overall program's run-time gain. Because of the overhead of parallel execution - such as coordinating threads - certain parallel sites and tasks may not contribute to the overall program's run-time gain. After you mark such parallel sites or tasks, either modify or eliminate their annotations.
If you have not used Intel Advisor previously, considering reading the appropriate Intel Advisor getting started Tutorial while using the C/C++ nqueens_Advisor sample or the Fortran nqueens sample. If you are not familiar with parallel concepts like data race, please read the help sections in Key Concepts.
Using the Suitability Report
The Suitability tool performs extensive analysis of your running serial program's executable (target) to show you the approximate performance characteristics of your annotated parallel sites and tasks. It also lets you modify certain assumptions - such as the number of CPUs on the target system and whether you will fix certain overhead items - and view their impact of the approximate performance based on your annotated sites and tasks.
Build a Release build target executable and run the Suitability tool. Once data collection completes, data appears in the Suitability Report window.
One of the first modeling parameters you should choose is the Target System. Use the Target System to select the type of hardware configuration to be analyzed: CPU, Intel Xeon Phi, or Offload to Intel Xeon Phi. The latter two apply to the Intel® Xeon Phi™ processor system and can help you determine which hardware configuration would be an appropriate choice for your target program.
For each Target System you choose, use the various modeling parameters to estimate how you can improve your program's predicted performance.
To view source-level details about a specific site or task, double-click its data row (or right click a row and select View Source from the context menu) to display the Suitability Source window. To return to the Suitability Report window, click the Suitability Report button.
To display and edit sources in your code editor, right-click a line and select Edit Source from the context menu. When using Visual Studio, the Visual Studio code editor appears with the file open at the corresponding location. When using the Intel Advisor GUI, the file type association (or Open With dialog box) determines the editor used. Later, to return to the Suitability Report or Suitability Source window, click the Result tab.
The Suitability Report , Suitability Source, and other Intel Advisor windows appear within the Result tab. There is one Result tab for each startup project.
For help using the Suitability Report and Suitability Source windows, see the help topic Viewing Data from the Suitability Tool.
Next Steps
After you examine the data in the Suitability Report and Suitability Source windows, choose one of the following:
- If the Site Gain values for the selected sites show a performance gain of 1.0 or less, the overhead costs of parallel thread execution exceeds the potential performance gains. Modify or remove the annotations for the task(s) and the enclosing site.
- If the Site Gain values for the selected sites show a performance gain greater than 1.0, look at the site's contribution to the Maximum Program Gain for All Sites. For sites that do not contribute significantly to the Maximum Program Gain for All Sites, modify or remove the annotations for the task(s) and its enclosing site. For sites that only contribute slightly to the Maximum Program Gain for All Sites, examine more closely the annotations and the assumptions about fixing the various overhead costs of parallel thread execution. In some cases, you may be able to adjust the annotations to improve the performance gain.
- When the Maximum Program Gain for All Sites for the program and the Site Gain values for all the sites show a worthwhile performance gain, proceed to 4. Check Correctness to check your remaining annotated sites for data sharing problems.
Note
If you modify your program's annotations or related sources, rebuild your program and run the Suitability tool again.