Mac power measurement

This directory contains the tools necessary to run different browsers and usage scenarios and measuring their performance impact. For Chromium it's also possible to generate profiles of CPU use. See benchmark.py for running the full suite. See collapse_profile.py for generating profiles.

Setting Up

Python Virtual Environment

These scripts use python Virtual Environments.

Create the venv. Only needs to be done once.

python3 -m venv ./env

Or using a specific python binary, e.g. from depot_tools

~/src/chromium/depot_tools/python-bin/python3 -m venv ./env

Activate the venv.

source ./env/bin/activate

Once the venv is activated, python refers to python3. Upgrade pip and install all python dependencies.

python -m pip install -U pip
python -m pip install -r requirements.txt

To deactivate venv.

deactivate

Chromium build

Chromium needs to be built with the following args.gn

is_debug = false
is_component_build = false
symbol_level = 0
blink_symbol_level = 0
is_official_build = true
is_chrome_branded = true

If profiling it needs to be built with these additional args.gn

enable_profiling = true
enable_dsyms = false
enable_stripping = false

If tracing it needs to be built with these additional args.gn

extended_tracing_enabled = true

In all cases the build needs to be copied to the “Applications” folder. If desired it's possible to build using components builds and dsyms but it adds the complexity of having to copy over the additional files to the test machine when building from a different machine.

Getting around sudo password

To disable asking password for sudo commands (required by powermetrics). Run sudo visudo and add the last line to User specification (replacing <user>):

# root and users in group wheel can run anything on any machine as any user
root ALL = (ALL) ALL
%admin ALL = (ALL) ALL
<user> ALL = (ALL) NOPASSWD:ALL

power_sampler

A compiled binary of power_sampler is required to run power measurements. It can be compiled using this command: autoninja -C out/Release tools/mac/power:power_sampler

DTrace

Running benchmark.py in profile mode uses dtrace to analyse the chromium processes. By default dtrace does not work well with SIP. Disabling SIP as a whole is not recommended and instead should be done only for dtrace using these steps:

• Reboot in recovery mode
• Start a shell
• Execute csrutil enable --without dtrace --without debug
• Reboot

benchmark.py

A tool that allow you to run different browsers under specific usage scenarios and:

• Measure their impact of system resource use.
• Profile the code that runs and/or is causing wake-ups. (chromium only)

./benchmark.py --scenarios idle_on_wiki:chrome idle_on_wiki:safari --power_sampler=/path/to/power_sampler
./benchmark.py --profile_mode cpu_time --scenarios idle_on_wiki:chromium --power_sampler=/path/to/power_sampler

Comparing benchmark executions

To compare the results of two benchmark runs, first generate a summary.csv file for each execution using the analyze.py tool. Assuming the baseline benchmark was output to /tools/mac/power/output/<baseline_timestamp>/<scenario_name> and the alternative was output to /tools/mac/power/output/<alternative_timestamp>/<scenario_name>:

./analyze.py --data_dir=/tools/mac/power/output/<baseline_timestamp>
./analyze.py --data_dir=/tools/mac/power/output/<alternative_timestamp>

Will create a summary.csv file in each of these directories. These can be passed to compare.py:

./compare.py --output_dir=/path/to/output/folder/ --baseline_dir=/tools/mac/power/output/<baseline_timestamp>/<scenario_name> --alternative_dir=/tools/mac/power/output/<alternative_timestamp>/<scenario_name>

Will create a /path/to/output/folder/comparison_summary.csv file containing information about the difference in metrics between both executions.

export_dtrace.py

A tool that converts the DTrace results created by benchmark.py into a format suitable for FlameGraph generation and analysis.

./export_dtrace.py --stack_dir ./output/idle_on_wiki_chromium_dtraces_cpu_time --output ./output/idle_on_wiki_cpu_profile.pb

This command will produce a file at ./output/idle_on_wiki_cpu_profile.pb.

The script can produce a pprof profile that can be used with pprof or a collapsed profile that can be used with tools such as FlameGraph and SpeedScope.

Known signatures are added to the pprof profile as labels. These can be appended to the flamegraph with

pprof -proto -tagroot signature <profile>

Usage scenario scripts

This directory contains the necessary files to make Chromium based browsers and Safari execute usage scenarios that represent real world usage.

The scripts differ slightly from similar tools like telemetry tests in that they work for Safari.

Scenarios

Scenarios are a set of operations to be applied on a set of URLs.

For example:

• Navigate to google.com
• Open the first link
• Wait 60 seconds

It's interesting to gather power metrics, profiles and traces for specific scenarios to understand their performance characteristics.

Currently supported scenarios are:

• idle
• meet
• idle_on_wiki
• idle_on_youtube
• navigation_top_sites
• navigation_heavy_sites
• zero_window

Tests

Unit tests can be run using run_tests.py.