Correct rocprofv3 usage instructions (#2925)
* Correct rocprofv3 usage * Apply suggestion from @SwRaw * Apply suggestion from @SwRaw * Update .gitignore
This commit is contained in:
@@ -1 +1,2 @@
|
||||
.cline_storage
|
||||
/projects/hip/_build
|
||||
|
||||
@@ -27,10 +27,10 @@ Follow this systematic approach to optimize GPU performance:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
rocprofv3 --stats ./your_application
|
||||
rocprofv3 --stats --<tracing_option> -- <application_path>
|
||||
|
||||
Collect metrics on kernel execution time, memory bandwidth, occupancy, and
|
||||
CU utilization.
|
||||
CU utilization. For more details on using ``rocprofv3`` for application tracing and profiling, see :doc:`rocprofv3 documentation <rocprofiler-sdk:how-to/using-rocprofv3>`.
|
||||
|
||||
2. **Analyze metrics to identify bottlenecks**
|
||||
|
||||
|
||||
Reference in New Issue
Block a user