Firefox Compile Benchmark

[STH-Dev]

I think this is a good one to add and Ian @ Anandtech requested it also.

Basic idea:
Download FF source to RAM disk
Compile FF.

[STH-Dev]

Maybe we use FF 40.0:

dependency

apt-get install zip

working on FF compile

wget https://archive.mozilla.org/pub/firefox/releases/40.0/source/firefox-40.0.source.tar.bz2
tar -xjvf firefox-40.0.source.tar.bz2
cd mozilla-release
./mach build

This is failing but should be close

[chuckleb]

How complete a build are we testing? We can probably build a basic FF without GTK dependencies, etc.. or if we go full steam, lots of dependencies.

[STH-Dev]

I think it is more about being able to be consistent across architectures
less about exactly how much we can compile.
On Aug 27, 2015 7:43 AM, "Charles Nguyen" notifications@github.com wrote:

How complete a build are we testing? We can probably build a basic FF
without GTK dependencies, etc.. or if we go full steam, lots of
dependencies.

—
Reply to this email directly or view it on GitHub
#69 (comment).

[chuckleb]

I have a working build, tested in a clean docker. Some things to think about:

• Do the options look ok?
• Compile speed can vary due to number of cores in the make step.
• Need to point to /dev/shm though the shm in my Docker is 64M and the build directory is 5.9GB. This may be a problem for the Docker version, but better for bare-metal version.
  Allow dev/shm size to be configurable in the native driver moby/moby#4981
  I may try to integrate in later, but documented for now.

---

apt-get install python gtk+-2.0 dbus yasm libasound2-dev libxt-dev zip
export SHELL=/bin/sh make

wget https://archive.mozilla.org/pub/firefox/releases/40.0/source/firefox-40.0.source.tar.bz2
tar xvfj firefox-40.0.source.tar.bz2
mkdir -p ffbuild; cd ffbuild
../mozilla-release/configure --disable-dbus --disable-pulseaudio --disable-gstreamer --disable-necko-wifi
make -j 10

[seankelly]

What about making a tmpfs and using that?

[chuckleb]

I think that is a standard workaround, passing /dev/shm as a volume into
the docker container. Have to enable privileged mode and whatnot, but we
can do that. Makes running it more complicated for users. I think we can do
some testing on actual performance difference from tmpfs and disk.

---

Please excuse autocorrect typos, sent from a phone.
On Aug 28, 2015 8:14 PM, "Sean Kelly" notifications@github.com wrote:

What about making a tmpfs and using that?

—
Reply to this email directly or view it on GitHub
#69 (comment).