Raspberry Pi Compiling testing.

I’ve been playing around with a Raspberry Pi along with the BitTorrent tracker I work on and decided to break out some benchmarks.

When compiling, I tested with a Pi Model B running a recent build of raspbian. I really only tested two compiling methods, both with “time make -j4” as the command.
One was with the Makefile configured as normal to only build on the Pi.

The time results for that are:
real 13m46.096s
user 13m19.210s
sys 0m13.040s

I also tested with a gentoo virtual machine configured with the cross tool chain for compiling for the pi and the Makefile configured to use distcc to compile from the pi to the virtual machine. The virtual machine was running on an i7 4770 machine with 4 cores configured. Additionally, distcc would also attempt to use the pi’s cpu for some compile work.

The results for that are:
real 0m37.467s
user 0m17.860s
sys 0m5.240s

As you can see, using distcc coupled with big machines can allow you to compile the big projects a LOT faster.

Update April 8, 2015:
I have obtained a Raspberry Pi 2 kit and tested compiling with it.

The results for it compiling on its own are:
real 1m48.933s
user 5m40.360s
sys 0m6.810s

And with distcc supplement matching above:
real 0m32.153s
user 0m44.860s
sys 0m2.730s

And finally, with distcc supplement with -j8 instead of -j4 (this one confused me a bit because it’s getting supplement from the pi for compiling):
real 0m53.763s
user 1m33.920s
sys 0m4.270s

Additionally, I plan to test with other SoC-class systems such as the Banana Pi (running the AllWinner A20 SoC) and Solid-Run’s MicroSoM family (such as the i4pro with an appropriate carrier) in the future.

It looks like having all the heavy lifting done on external machines gives the best performance for compiling rather than having the local pi do the work.

Your mileage may vary.