derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
105 lines
4.0 KiB
ReStructuredText
105 lines
4.0 KiB
ReStructuredText
.. meta::
|
|
:description: Documentation of the usage of streaming performance monitor(SPM) with rocprofv3 command-line tool
|
|
:keywords: Sampling counters, streaming performance monitors, rocprofv3, rocprofv3 tool usage, Using rocprofv3, ROCprofiler-SDK command line tool, SPM
|
|
|
|
.. _using-spm:
|
|
|
|
==================
|
|
Using SPM
|
|
==================
|
|
|
|
SPM (Streaming Performance Monitor) sampling service for GPU profiling is a profiling technique to periodically sample performance counters with GPU timestamp.
|
|
|
|
Here are the benefits of using SPM to sample counters:
|
|
|
|
- Identify performance bottlenecks
|
|
- Understand kernel execution behavior
|
|
- fine-grained, time-resolved performance data.
|
|
|
|
To try out the SPM, you can use the command-line tool ``rocprofv3`` or the ROCprofiler-SDK library.
|
|
|
|
SPM availability and configuration
|
|
===========================================
|
|
|
|
To check counters that can be sampled, use:
|
|
|
|
.. code-block:: bash
|
|
|
|
rocprofv3 -L
|
|
|
|
Or
|
|
|
|
.. code-block:: bash
|
|
|
|
rocprofv3 --list-avail
|
|
|
|
The output lists if ``rocprofv3`` supports SPM
|
|
|
|
.. code-block:: bash
|
|
|
|
Counter_Name : TCC_MISS
|
|
Description : Number of cache misses. UC reads count as misses.
|
|
Block : TCC
|
|
SPM : Supported
|
|
Dimensions : DIMENSION_INSTANCE[0:15] DIMENSION_XCC[0:7]
|
|
|
|
The preceding output shows that the TCC_MISS counter can be sampled.
|
|
|
|
.. note::
|
|
For proper functioning, SPM requires AMD GPU Driver version **6.19.14.31400000** or later.
|
|
Before using SPM, verify the loaded ``amdgpu`` kernel module version.
|
|
|
|
To check the driver version, use:
|
|
|
|
.. code-block:: bash
|
|
|
|
cat /sys/module/amdgpu/version
|
|
|
|
# Example output:
|
|
# 6.19.14.31400000
|
|
|
|
You can also check this driver version using ``amd-smi version`` command on DKMS-built systems.
|
|
|
|
Use the following command to use SPM:
|
|
|
|
.. code-block:: bash
|
|
|
|
rocprofv3 --spm-beta-enabled --spm SQ_WAVES --spm-sample-interval-unit sclk_cycles --spm-sample-interval 1200 --output-format json -- <application_path>
|
|
|
|
The preceding command enables SPM for SQ_WAVES and sample interval with unit as sclk cycle counts. Replace ``<application_path>`` with the path to the application you want to profile.
|
|
This generates a JSON results file prefixed with the process ID.
|
|
|
|
.. code-block:: bash
|
|
|
|
rocprofv3 --spm-beta-enabled --spm SQ_WAVES --spm-sample-interval-unit sclk_cycles --spm-sample-interval 1200 -- <application_path>
|
|
|
|
The preceding command enables SPM for SQ_WAVES and sample interval with unit as sclk cycle counts. Replace ``<application_path>`` with the path to the application you want to profile.
|
|
This generates a rocpd results file prefixed with the process ID.
|
|
|
|
SPM is emitted directly only to the JSON and ``rocpd`` (default) output formats. Only CSV output is supported from the rocpd database via ``rocpd convert``.
|
|
|
|
Input parameters
|
|
===================
|
|
|
|
Here are the input parameters used to configure SPM
|
|
|
|
- ``Metrics List``: The list of counters that you want to sample with SPM. Currently, basic counters are supported.
|
|
- ``SPM Sample Interval``: Specifies the sampling interval for SPM counter collection. It is used with spm-sample-interval-unit to define how frequently counters are sampled.
|
|
- ``SPM Sample Interval Unit``: Specifies the unit for the SPM sample interval. Used with --spm-sample-interval to define the sampling interval. Currently, sclk_cycles unit is supported and the sample interval used with this unit is rounded to nearest multiple of 32.
|
|
|
|
|
|
.. code-block:: bash
|
|
|
|
rocprofv3 --spm-beta-enabled --spm SQ_WAVES --spm-sample-interval-unit sclk_cycles --spm-sample-interval 1200 --output-format json -- <application_path>
|
|
rocprofv3 --spm-beta-enabled --spm SQ_WAVES --spm-sample-interval-unit sclk_cycles --spm-sample-interval 1200 -- <application_path>
|
|
|
|
CSV format conversion
|
|
=====================
|
|
|
|
.. code-block:: bash
|
|
|
|
rocpd convert -i <input-file>.db --output-format csv
|
|
|
|
The CSV conversion process generates a output file ``rocpd-output-data/out_spm_counter_collection_trace.csv`` path relative to the current working directory.
|
|
|