AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

SDK doc updates (#1183)

Browse Source 
* correcting usage example

* rccl trace

* Adding Navi power state limitation

* Addressed feedback

* kernel-rename

* kokkos trace

* more information on kookos tracing

* Corecting tool library hardcoding

* summary domains

* Updating domain stats file

* updating images

* rocprofv3 default behavior update

* Removing README from API documentation

* Added missing description in Topics

* Fixed wrong rendering of README in API document

* Fixing Topics in API docs

* Removing API doc for details/rccl.h

* Addressed review comments
This commit is contained in:
Gopesh Bhardwaj
2024-11-22 12:05:11 +05:30
committed by GitHub
parent 8d2ce4b475
commit 7ea9ced493
13 changed files with 234 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files
source/docs/conceptual/comparing-with-legacy-tools.rst
+7 -1
View File
@@ -383,4 +383,10 @@ ROCprofiler-SDK introduces a new command-line tool, `rocprofv3`, which is a more
Timing Difference Between rocprofv3 and rocprofv1/v2
========================================================
Rocprofv3 has improved the accuracy of timing information by reducing the tool overhead required to collect data and reducing the interference to the timing of the kernel being measured. The result of this work is a reduction in variance of kernel times received for the same kernel execution and more accurate timing in general. These changes have not been backported (and will not be backported) to rocprofv1/v2, so there can be substantial (20%) differences in execution time reported by v1/v2 vs v3 for a single kernel execution. Over a large number of samples of the same kernel, the difference in average execution time is in the low single digit percentage time with a much tighter variance of results on rocprofv3. We have included testing in the test suite to verify the timing information outputted by rocprofv3 to ensure that the values we are returning are accurate.
``rocprofv3`` has improved the accuracy of timing information by reducing the tool overhead required to collect data and reducing the interference to the timing of the kernel being measured. The result of this work is a reduction in variance of kernel times received for the same kernel execution and more accurate timing in general. These changes have not been backported (and will not be backported) to rocprofv1/v2, so there can be substantial (20%) differences in execution time reported by v1/v2 vs v3 for a single kernel execution. Over a large number of samples of the same kernel, the difference in average execution time is in the low single digit percentage time with a much tighter variance of results on rocprofv3. We have included testing in the test suite to verify the timing information outputted by rocprofv3 to ensure that the values we are returning are accurate.
========================================================
Default run of rocprofv3 and rocprofv1/v2
========================================================
``rocprofv3`` has a different default behavior than rocprofv1/v2 when being run without any option. The default behavior of rocprofv3 is to collect all available agents on the system and to output it in ``csv`` format. The default behavior of rocprofv1/v2 was to output the `kernel traces` in CSV format. In rocprofv3, kernel traces can be obtained by using ``--kernel-trace`` option.