Bug 1297520 Comment 5 Edit History

Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.

Thanks!

I have done some measurements with the test snap in the `candidate/one-locale` channel on a raspberry 4 (arm64).
Cold starts and hot starts are marginally faster than the snap in `stable` (by less than a second, out of a baseline of ~12s for hot starts and ~17s for cold starts).

Where I'm seeing a very noticeable improvement is for the very first cold start, without an existing profile, nor one to import from `~/.mozilla`. In this case, the startup time is halved, from 34s down to 17s.

Note that these timings were recorded using a stopwatch, so they are quite imprecise, and it was done on arm64 because the amd64 snap wasn't ready yet (it is now published).
Thanks!

I have done some measurements with the test snap in the `candidate/one-locale` channel on a raspberry 4 (arm64).
Cold starts and hot starts are marginally faster than the snap in `stable` (by less than a second, out of a baseline of ~12s for hot starts and ~17s for cold starts).

Where I'm seeing a very noticeable improvement is for the very first cold start, without an existing profile, nor one to import from `~/.mozilla`. In this case, the startup time is halved, from 34s down to 17s.

Note that these timings were recorded using a stopwatch, so they are quite imprecise, and it was done on an arm64 board where I/O is slow because the amd64 snap wasn't ready yet (it is now published).
Thanks!

I have done some measurements with the test snap in the `candidate/one-locale` channel on a raspberry pi 4 (arm64).
Cold starts and hot starts are marginally faster than the snap in `stable` (by less than a second, out of a baseline of ~12s for hot starts and ~17s for cold starts).

Where I'm seeing a very noticeable improvement is for the very first cold start, without an existing profile, nor one to import from `~/.mozilla`. In this case, the startup time is halved, from 34s down to 17s.

Note that these timings were recorded using a stopwatch, so they are quite imprecise, and it was done on an arm64 board where I/O is slow because the amd64 snap wasn't ready yet (it is now published).

Back to Bug 1297520 Comment 5