If data is collected using CLI, then use Import Session to import the session into GUI to analyze data in GUI. Below are the supported views to analyze GPU Profile.
Here is a screenshot of an imported GPU Profile session in GUI.
Figure 1. GPU Offloading Analysis - Hot Spots
Legend:
- Application Details: Gives an overview of the target application traced.
- GPU Kernel Launch Summary: To identify top 4 hottest kernels launched to GPU and its count, total execution time on GPU cores.
- Data Transfer Summary: To identify how much time spent in data transfer between host and device, how many times data transfer initiated.
Figure 2. GPU Offloading Analysis - Session Information
Legend:
- Profile Details 
- System Details
- Target Details
- GPU Device Details
Figure 3. GPU Offloading Analysis - Analyze
Legend:
- HIP Overview: HIP API calls summary
- HSA Overview: HSA API calls summary
Figure 4. GPU Offloading Analysis - Per Thread Timeline
Use this UI to analyze the following:
- GPU Usage, GPU Memory usage and GPU Power of your application over the profile duration.
- GPU Kernels executed over the profile duration.
- Data Transfer between host and device over the profile duration will help to identify the time spent in data copy. Identify the hottest GPU Kernel.
Use GPU Trace to get a list of hottest GPUS Kernels. All the kernels are sorted in descending order of elapsed time.