The RAM use and startup times of Fifth 0.1 and some less-heard-of competitors were measured. Xubuntu 14.04 x86_64 was used as the platform.
|RAM use (mb)||Cold startup (ms)||Warm startup (ms)|
|Arora 0.11.0 (QtWebkit 4)||1731.7||4745||345|
|Midori 0.5.8 (WebkitGTK)||376.7||1250||645|
|Otter 0.9.04 #45 (QtWebkit 5)||1445.7||1185||585|
We can see that the cold startup speed of Fifth is in need of improvement, with it only managing to defeat Arora. This is mainly due to webkit being plain huge, as well as the embedding of ICU (another huge library) in the portable binary.
In RAM use, Fifth leads the pack with no competition. Otter and Arora both used an ungodly amount of RAM for these four tabs. This is further exacerbated by the fact the Fifth portable binary has webkit linked statically, while the other webkit browsers link it dynamically (!).
Moving to startup speeds, Otter is quite commendable, defeating even Firefox and Opera handily in cold startup, both browsers having had heavy optimization for this use case. In warm starts, Otter and Fifth are neck-to-neck, sharing the second place to Arora.
MethodFor the RAM test, each browser had these websites opened in tabs: bbc.com, phoronix.com, tomshardware.com, and gmail.com. After each site had fully loaded, the browser's RAM use was measured with
pmap -d PID, under the writable/private column, which counts the application's RAM use excluding shared libraries. For multi-process browsers, the sum of all processes was used.
For the startup tests, each browser was set to open a blank page (about:blank homepage). To
get a repeatable measurement, the starts were recorded with VNC over localhost. This gives a
theoretical 14.2 kHz accuracy (as calculated by the inverse RTT from
However, to account for CPU peaks and other delays, these values have been rounded up to the
nearest 5 ms.
Clean profiles were used in all cases. As the platform was a clean Xubuntu, no browser had flash, adblock, or other such plugins/extensions - all defaults.
Firefox and Arora were from the repo, the others from each upstream binary package.