Configure Target Application

Intel® Advisor supports targets:

Developed to run on Windows* or Linux* operating systems using the Intel® oneAPI DPC++/C++ Compiler, Intel® C++ Compiler Classic, Intel® Fortran Compiler Classic, or GNU* gcc compiler development environment
That use C/C++, Fortran, or mixed Python* code for the portions that will run in parallel.
That use Data Parallel C++ (DPC++), OpenCL™, or OpenMP* with pragma omp target (for C++) or directive omp target (for Fortran) code

The target executable must contain source symbol table debug information, so the Intel® Advisor can provide source line correlation and viewing sources.

Important

To analyze an application with the Intel® Advisor, the application should take longer than 500 milliseconds to execute on CPU or GPU. If your application execution time is lower, it might cause inaccurate data sampling or a No data is collected error.

Before you start profiling your application and applying changes that should increase performance, you can configure the application as follows to optimize it for analyses:

Limit the number of threads used by parallel frameworks to configure the application for threading.
Choose a small, representative data set to reduce analysis overheads by reducing the amount of analyzed data.