Contact YouTube LinkedIn

iSYSTEM Trace Solutions: Sampling-based Profiling

 

A new winIDEA feature enables you to statistically analyze:

  • CPU utilization
  • OS tasks
  • ISR load distribution
  • Monitor data signals over time via a debug port.

 

Carry out timing analysis via the debug interface

The Sampling-based Profiling allows you to carry out timing analysis via the debug interface of microcontrollers and processors with limited trace options or hardware trace resources. Via Program Counter (PC) and Data Sampling, you can statistically measure and monitor CPU utilization, load distribution and data signals over time to test your application.

The new feature is simple to use with any kind of microcontroller or processor and streamlines software development and test especially for safety applications in various markets like automotive, healthcare and industrial automation.

 

Fig. 1: The concept of Sampling-based Profiling in winIDEA

 

Sampling-based Profiling is part of the winIDEA Analyzer. With this function, the developer obtains data by sampling specific memory locations or the CPU Program Counter via the debug interface – without the need for any trace hardware. The profiler also analyzes the retrieved data and visualizes the results.

 

PC Sampling and Data Sampling made easier

iSYSTEM has implemented support for two main use cases: 

  • PC Sampling
  • Data Sampling

The Profiler samples data on program execution and calculates the CPU Utilization in percent for each function to calculate the function load. OS Running Task and ISR Load Profiling gives insights to OS Task and ISR Load distribution. In both cases, the developer will run samples until the statistics stabilize. This provides a very good indication about the load situation in the system and helps the developer to judge if the application is running as expected or not. The third option of Sampling-based Profiling – which is not really related to Timing Analysis and less practical – is to build a kind of an oscilloscope for monitoring data signals over time.

 

Fig. 2: The three use cases of sampling-based profiling

 

Perfect tool for statistical analysis

Sampling-based Profiling does not replace real hardware trace as it cannot guarantee that all relevant events are captured. However, it is the perfect tool for statistical analysis or for sampling slowly changing variables in a system and helps developers to further streamline their testing effort.

“Some of us call it ‘Poor Man’s Profiler’, but nevertheless, Sampling-based Profiling is an extremely helpful statistical analysis method based on sampling, and a valuable extension of the iSYSTEM Software Analyzer,” says Erol Simsek, CEO of iSYSTEM. “Even if the processor has no trace interface for pure timing and load analysis, this new feature enables our customers to quickly get an idea of the run-time behavior of their software.”

 

Further Links and Resources

Don't Miss Out!

Get updates on new products, webinars and other events.

Please register my email address for the iSYSTEM Newsletter. You will receive an email to confirm the registration. Your data will only be used to personalize the newsletter and will not be passed on to third parties. You can unsubscribe from the newsletter at any time or revoke your consent by e-mail to info@isystem.com. Further information on the performance measurement included in the consent, the use of the service provider CleverReach, logging of your registration and your rights of revocation can be found in our iSYSTEM privacy statement.