Formatting issue in code block and TOC fixed for PC Sampling (#731)

* Formatting issue in code block and TOC fixed

* Performance model reverted
Este commit está contenido en:
Pratik Basyal
2025-06-06 16:16:55 -04:00
cometido por GitHub
padre 25f21715bc
commit ed05c00103
Se han modificado 5 ficheros con 16 adiciones y 16 borrados
+1 -1
Ver fichero
@@ -45,4 +45,4 @@ key hardware blocks, each detailed in the following sections:
* :doc:`command-processor`
* :doc:`system-speed-of-light`
* :doc:`system-speed-of-light`
+11 -10
Ver fichero
@@ -19,30 +19,31 @@ particularly useful for understanding stalls during the kernel execution. The PC
Profiling options
---------------------
For using profiling options for PC sampling the configuration needed are:
- ``--pc-sampling-method``: Should be either ``stochastic`` or ``host_trap``
- ``--pc-sampling-interval``: For stochastic sampling, the interval is in cycles. The finest granularity is 1 cycle. For host_trap sampling, the interval is in microsecond (DEFAULT: 1048576). The interval should be the power of 2. You are recommended to try to starting from 1048576, and lowering until reaching 65536.
* ``--pc-sampling-method``: Should be either ``stochastic`` or ``host_trap``
* ``--pc-sampling-interval``: For stochastic sampling, the interval is in cycles. The finest granularity is 1 cycle. For host_trap sampling, the interval is in microsecond (DEFAULT: 1048576). The interval should be the power of 2. You are recommended to try to starting from 1048576, and lowering until reaching 65536.
**Sample command:**
.. code-block:: bash
rocprof-compute profile -n pc_test -b 21 --no-roof --pc-sampling-method stochastic --pc-sampling-interval 1048576 -VVV -- target_app
.. code-block:: shell
$ rocprof-compute profile -n pc_test -b 21 --no-roof --pc-sampling-method stochastic --pc-sampling-interval 1048576 -VVV -- target_app
-----------------------
Analysis options
-----------------------
For using analysis options for PC sampling the configuration needed are:
- ``--pc-sampling-sorting-type``: ``offset`` or ``count``. The default option is ``offset``. ``offset`` is an assembly instruction offset in the code object.
* ``--pc-sampling-sorting-type``: ``offset`` or ``count``. The default option is ``offset``. ``offset`` is an assembly instruction offset in the code object.
**Sample command:**
.. code-block:: bash
rocprof-compute analyze -p workloads/pc_test/MI300A_A1/ -b 21 -k 0 --pc-sampling-sorting-type offset
.. code-block:: shell
$ rocprof-compute analyze -p workloads/pc_test/MI300A_A1/ -b 21 -k 0 --pc-sampling-sorting-type offset
.. note::
- PC sampling feature is currently in BETA version. To enable PC sampling,
you have to explicitly enable it with block index 21.
- To associate PC sampling info back to HIP source code, you need to build the profiling target app with ``-g`` to keep the symbols. Otherwise, PC sampling info would be only associated with assembly lines.
* PC sampling feature is currently in BETA version. To enable PC sampling, you have to explicitly enable it with block index 21.
* To associate PC sampling info back to HIP source code, you need to build the profiling target app with ``-g`` to keep the symbols. Otherwise, PC sampling info would be only associated with assembly lines.
+1 -1
Ver fichero
@@ -252,7 +252,7 @@ Filtering options
.. _profiling-hw-component-filtering:
Hardware report block filtering
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
You can profile specific hardware report blocks to speed up the profiling process.
In ROCm Compute Profiler, the term hardware report block refers to a section of the
+1 -1
Ver fichero
@@ -19,8 +19,8 @@ subtrees:
entries:
- file: how-to/use.rst
- file: how-to/pc_sampling.rst
- file: how-to/profile/mode.rst
title: Use PC sampling
- file: how-to/profile/mode.rst
- file: how-to/analyze/mode.rst
entries:
- file: how-to/analyze/cli.rst
+2 -3
Ver fichero
@@ -15,7 +15,7 @@ ROCm Compute Profiler primarily targets use with
:doc:`accelerators in the MI300, MI200, and MI100 families <rocm:conceptual/gpu-arch>`.
Development is in progress to support Radeon™ (RDNA) GPUs.
ROCm Compute Profiler is built on top of :doc:`ROCProfiler <rocprofiler:rocprofv1>` to
ROCm Compute Profiler is built on top of :doc:`ROCProfiler <rocprofiler:index>` to
monitor hardware performance counters.
.. _high-level-design:
@@ -88,8 +88,7 @@ high level.
* :ref:`Memory Chart Analysis panel <grafana-panel-memory-chart-analysis>`
* :ref:`Roofline Analysis panel <grafana-panel-roofline-analysis>`
(*Supported on MI200 and above architectures only)
* :ref:`Roofline Analysis panel <grafana-panel-roofline-analysis>` (Supported on MI200 and above architectures only)
* :ref:`Command Processor (CP) panel <grafana-panel-cp>`