#15979: Switch to google benchmark for pgm dispatch tests #16547
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
tests/tt_metal/tt_metal/perf_microbenchmark/dispatch/test_pgm_dispatch.cpp
Outdated
Show resolved
Hide resolved
jbaumanTT
added a commit
that referenced
this pull request
Jan 9, 2025
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs. Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the `--benchmark_filter=<regex>` command-line argument. One-off test cases can be run by passing `--custom` and the command line arguments as before.
jbaumanTT
force-pushed
the
jbauman/googlebenchmark2
branch
from
43eafb6 to
c1c334e
Compare
jbaumanTT
requested review from
abhullar-tt,
pgkeller,
aliuTT,
tt-aho,
tt-dma,
tt-asaigal,
ubcheema and
a team
as code owners
pgkeller
reviewed
Jan 9, 2025
jbaumanTT
added a commit
that referenced
this pull request
Jan 9, 2025
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs. Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the `--benchmark_filter=<regex>` command-line argument. One-off test cases can be run by passing `--custom` and the command line arguments as before.
jbaumanTT
force-pushed
the
jbauman/googlebenchmark2
branch
from
c1c334e to
a754edf
Compare
jbaumanTT
added a commit
that referenced
this pull request
Jan 9, 2025
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs. Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the `--benchmark_filter=<regex>` command-line argument. One-off test cases can be run by passing `--custom` and the command line arguments as before.
jbaumanTT
force-pushed
the
jbauman/googlebenchmark2
branch
from
a754edf to
99a4fc7
Compare
Ok, added back filt_pgm_dispatch.pl. I've added a new json_to_csv.py that can be used to dump to a CSV (one line per test). The workflow is to run google benchmark framework also has a native way to output CSVs, but it's deprecated and requires some extra work. This patch also uploads the json file from the CI bots, so it's pretty easy to download that and convert it. |
tt-rkim
reviewed
Jan 9, 2025
tests/tt_metal/tt_metal/perf_microbenchmark/dispatch/compare_pgm_dispatch_perf_ci.py
Outdated
Show resolved
Hide resolved
jbaumanTT
added a commit
that referenced
this pull request
Jan 10, 2025
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs. Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the `--benchmark_filter=<regex>` command-line argument. One-off test cases can be run by passing `--custom` and the command line arguments as before.
jbaumanTT
force-pushed
the
jbauman/googlebenchmark2
branch
from
99a4fc7 to
61c15a4
Compare
jbaumanTT
added a commit
that referenced
this pull request
Jan 10, 2025
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs. Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the `--benchmark_filter=<regex>` command-line argument. One-off test cases can be run by passing `--custom` and the command line arguments as before.
jbaumanTT
force-pushed
the
jbauman/googlebenchmark2
branch
from
61c15a4 to
36ae490
Compare
tt-rkim
approved these changes
Jan 14, 2025
tests/tt_metal/tt_metal/perf_microbenchmark/dispatch/compare_pgm_dispatch_perf_ci.py
Outdated
Show resolved
Hide resolved
tests/tt_metal/tt_metal/perf_microbenchmark/dispatch/json_to_csv.py
Outdated
Show resolved
Hide resolved
|
@TT-billteng LOOK SEE LOOK SEE |
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs. Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the `--benchmark_filter=<regex>` command-line argument. One-off test cases can be run by passing `--custom` and the command line arguments as before.
jbaumanTT
force-pushed
the
jbauman/googlebenchmark2
branch
from
36ae490 to
0c88372
Compare
pgkeller
approved these changes
Jan 15, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Ticket
#15979
Problem description
The current output format works well with a smaller number of tests, but with a large number of tests it's hard to connect the resulting number to the test that created it. This particular hinders storing test results in a database and comparing them across runs.
What's changed
Switch to using the google benchmark framework, which outputs a JSON file containing the results of all tests. By default, running the binary runs all the benchmarks from sweep_pgm_dispatch.sh. The set of tests to be run can be filtered using the
--benchmark_filter=<regex>command-line argument.One-off test cases can be run by passing
--customand the command line arguments as before.Checklist