-->

Overview

The CUDA Profiling Tools Interface (CUPTI) enables the creation of profiling and tracing tools that target CUDA applications. CUPTI provides the following APIs: the Activity API, the Callback API, the Event API, the Metric API and the Profiler API. Using these APIs, you can develop profiling tools that give insight into the CPU and GPU behavior of CUDA applications. CUPTI is delivered as a dynamic library on all platforms supported by CUDA.

What's New

CUPTI contains below changes as part of the CUDA Toolkit 11.4 release.
  • Profiling APIs support profiling of the CUDA kernel nodes launched by a CUDA Graph. Auto range profiling with kernel replay mode and user range profiling with user replay and application replay modes are supported. Other combinations of range profiling and replay modes are not supported.
  • Added support for tracing and profiling on NVIDIA virtual GPUs (vGPUs) on an upcoming GRID/vGPU release.
  • Added sample profiling_injection to show how to build injection library using the Profiling API.
  • Added sample concurrent_profiling to show how to retain the kernel concurrency across streams and devices using the Profiling API.