Fast and accurate performance analysis
Take advantage of hotspot analysis to find the function that are taking the longest to run. Then jump straight to the source to optimise.
Identify locks and waits
Find common causes of parallel application slow down - where waiting too long time on a lock means the cores are underutilised during the wait.
Easy to use
Pre-defined performance profiling experiments help you quickly get detailed profiling information without having to know micro architectural details.
See when threads and running, waiting and when transitions occur. Select a time range in the timeline and filter out inaccurate data - such as startup time - which masks accurate results.
Works on threaded and non-threaded code
Determines calling sequences and graphically displays the critical path, allowing you to see which functions took the most time to process or were blocked the longest.
System Wide Event Based Sampling (EBS)
Uses the on chip PMU to count performance events like cache misses, clock ticks and instructions retired.
Supports multiple languages
Works with C++, Fortran, Java, assembly and more - infact VTune supports compilers from any vendor that follow platform standards including Microsoft, GCC and Intel.
Supports multiple systems
Works on either Linux or Windows systems including Windows 7, Windows Vista, Windows XP, Windows Server, Red Hat Enterprise Linux, Fedora, SUSE and more.
Works with Microsoft Visual Studio
On Windows systems VTune can (optionally) integrate directly with Microsoft Visual Studio 2008, 2010 or 2012.
Not just for Intel processors
Much of the functionality available in VTune works on both Intel and other compatible processors. Plus results can be analysed on any compatible processor but some on-chip performance monitoring features do require a genuine Intel processor for data collection.