Visible to Intel only — GUID: GUID-11FBD195-12CA-426F-9AEB-2574252BCC6A
Visible to Intel only — GUID: GUID-11FBD195-12CA-426F-9AEB-2574252BCC6A
Recommended Workflows
The workflows described here capture the most important steps you complete to profile several types of workloads. Additionally, see examples in technical recipes that demonstrate popular scenarios for these workloads. Troubleshooting topics can help you avert common problems in your performance analysis.
To profile this workload | Step 1: Start here | Step 2: Continue here | Step 3: See examples | Common Problems |
---|---|---|---|---|
HPC applications |
Depending on your results, continue your analysis by examining one of these issues: |
|||
Cloud Service Provider (CSP) applications |
Learn how to install Intel® VTune™ Profiler in a cloud environment. Also learn how to profile hardware without using sampling drivers. |
Run these analysis types: |
||
GPU applications |
First, make sure your system is set up to run GPU analyses. Then, complete a GPU Offload analysis to identify if your application is GPU/CPU-bound. |
Continue profiling with the GPU Compute/Media Hotspots analysis. |
|
No GPU Utilization Data Is Collected |
Gaming, Media, and Content Creation applications |
Follow the Game Tuning with Intel® workflow. |
Run a Hotspots analysis to identify performance bottlenecks. If you encounter any threading issues, resolve them by running a Threading analysis next. |
|
|
AI applications |
If your application runs on Neural Processing Units (NPUs), start with the NPU Exploration analysis. For Python* applications, see Python* Code Analysis options. Profiling your application may also benefit from the addition of Python bindings to the Instrumentation and Tracing Technology API (ITT API). |
Learn how you use Intel® VTune™ Profiler to profile: . |