Accelerators Analysis Group
The
Accelerators
group introduces analysis types that monitor CPU, GPU and FPGA usage for your application/system.- Use the GPU Offload analysis to profile applications that use a Graphics Processing Unit (GPU) for rendering, video processing, and computations. This analysis type helps you identify whether your application is CPU or GPU bound.
- For GPU-bound applications, use the GPU Compute/Media Hotspots (preview) analysis type to see the GPU kernel execution per code line. Identify performance issues caused by memory latency or inefficient kernel algorithms.
- With the CPU/FPGA Interaction analysis, you can explore FPGA utilization for each FPGA accelerator and identify the most time-consuming FPGA computing tasks.
A
PREVIEW FEATURE
may or may not appear in a future production release. While a preview feature is available for your use, feedback about its usefulness will determine its availability in future releases. Data collected with a preview feature is not guaranteed to be compatible with future releases.
Prerequisites:
- Install the sampling driver for hardware event-based sampling collection types. For Linux* and Android* targets, if the sampling driver is not installed,VTunecan work on Perf* (driverless collection).Profiler
- To enable system-wide and uncore event collection, use root or sudo to set/proc/sys/kernel/perf_event_paranoidto0.$ echo 0>/proc/sys/kernel/perf_event_paranoid