d65de0a203
* Replace O(n^2²) nested loop with O(1) dictionary lookup when associating metric values with metrics. Pre-group values by (metric_id, kernel_name) to eliminate redundant iteration over entire values dataframe for each metric-kernel combination. * This optimization significantly improves database write performance for workloads with large numbers of metrics and kernels.
This directory contains all the source code files for the ROCm Compute Profiler project.
The top-level rocprof-compute executable is the entry point and defines the flow for subsequent execution modes.
To build ROCm Compute Profiler as a developer (from source) please add this directory to PATH.
You will find utilities and logic specific to ROCm Compute Profiler's different execution modes in their
respective subdirectories.