This get started video for Intel® VTune™ Profiler provides a hands-on overview of the UI, a data collection configuration, and performance data analysis views. Follow along to set up:

• Remote or local data collection
• A location for source and binary application files for source code hot spot and bottleneck identification
• Hardware events for sampling and sampling intervals

Analysis types that can be configured include:

• Hot spot analysis to detect functions that are consuming the most processor time
• Threading analysis for locks and waits such as semaphores, mutexes, and I/O waits
• System overview to explore system resource use
• HPC performance characterization to analyze compute-intensive applications, memory access, and vectorization
• I/O analysis for disk, network, and system bus throughput and latency
• Microarchitecture exploration to identify processor instruction execution pipeline bottlenecks
• Memory consumption and access analysis
• GPU offload analysis

Each analysis type has various configuration options. Many include the option to specify user-mode sampling or hardware event-based sampling.