[Meta] Setting input.value when input has focus is 1.1x slower than Chrome/Safari
Categories
(Core :: DOM: Core & HTML, defect, P3)
Tracking
()
Performance Impact | ? |
People
(Reporter: jandem, Assigned: masayuki)
References
(Depends on 4 open bugs, Blocks 3 open bugs)
Details
(Keywords: meta, perf)
Attachments
(4 files, 1 obsolete file)
(deleted),
text/html
|
Details | |
(deleted),
patch
|
Details | Diff | Splinter Review | |
(deleted),
text/html
|
Details | |
(deleted),
patch
|
Details | Diff | Splinter Review |
Reporter | ||
Updated•8 years ago
|
Reporter | ||
Updated•8 years ago
|
Reporter | ||
Comment 1•8 years ago
|
||
Comment 2•8 years ago
|
||
Comment 3•8 years ago
|
||
Comment 4•8 years ago
|
||
Reporter | ||
Comment 5•8 years ago
|
||
Comment 6•8 years ago
|
||
Updated•8 years ago
|
Comment 7•8 years ago
|
||
Comment 8•8 years ago
|
||
Comment 9•8 years ago
|
||
Comment 10•8 years ago
|
||
Reporter | ||
Comment 11•8 years ago
|
||
Comment 12•8 years ago
|
||
Comment 13•8 years ago
|
||
Updated•8 years ago
|
Comment 14•8 years ago
|
||
Updated•8 years ago
|
Reporter | ||
Comment 16•8 years ago
|
||
Comment 17•8 years ago
|
||
Updated•8 years ago
|
Comment 18•8 years ago
|
||
Comment 19•8 years ago
|
||
Assignee | ||
Comment 20•8 years ago
|
||
Assignee | ||
Comment 21•8 years ago
|
||
Comment 22•8 years ago
|
||
Comment 23•8 years ago
|
||
Comment 24•8 years ago
|
||
Comment 25•8 years ago
|
||
Comment 26•8 years ago
|
||
Comment 27•8 years ago
|
||
Assignee | ||
Comment 28•8 years ago
|
||
Comment 29•8 years ago
|
||
Reporter | ||
Comment 30•8 years ago
|
||
Comment 31•8 years ago
|
||
Reporter | ||
Comment 32•8 years ago
|
||
Comment 33•8 years ago
|
||
Comment 34•8 years ago
|
||
Comment 35•8 years ago
|
||
Comment 36•8 years ago
|
||
Comment 37•8 years ago
|
||
Comment 38•8 years ago
|
||
Comment 39•8 years ago
|
||
Updated•8 years ago
|
Updated•8 years ago
|
Comment 40•7 years ago
|
||
Comment 41•7 years ago
|
||
Comment 42•7 years ago
|
||
Comment 43•7 years ago
|
||
Comment 44•7 years ago
|
||
Comment 45•7 years ago
|
||
Comment 46•7 years ago
|
||
Comment 47•7 years ago
|
||
Updated•7 years ago
|
Assignee | ||
Comment 48•7 years ago
|
||
Assignee | ||
Comment 49•7 years ago
|
||
Comment 50•7 years ago
|
||
Updated•7 years ago
|
Updated•7 years ago
|
Updated•7 years ago
|
Comment 51•7 years ago
|
||
Updated•7 years ago
|
Updated•7 years ago
|
Updated•6 years ago
|
Updated•6 years ago
|
Comment 52•6 years ago
|
||
Assignee | ||
Comment 53•6 years ago
|
||
Updated•6 years ago
|
Assignee | ||
Comment 54•5 years ago
|
||
Chrome's score is 15ms ~ 18ms, Nightly's score is 19ms ~ 20ms. Our performance of this is really close to Chrome now.
Reporter | ||
Comment 55•5 years ago
|
||
(In reply to Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(no pain of the broken bone, but the corset blocks me to concentrate) from comment #54)
Chrome's score is 15ms ~ 18ms, Nightly's score is 19ms ~ 20ms. Our performance of this is really close to Chrome now.
Great work! It's amazing how far we've come when comparing this to my numbers in comment 0.
Comment 56•5 years ago
|
||
Reflecting current numbers from comment 54.
Updated•5 years ago
|
Assignee | ||
Comment 57•5 years ago
|
||
Because of landing a lot of optimization for bug 1597679, today's Nightly build's score of attachment 8848015 [details] becomes really close to Chrome, but we need some more optimization for wining.
Assignee | ||
Comment 58•5 years ago
|
||
On my Win10 environment (Ryzen Threadripper 3960X), Nightly's score is 15-18ms on attachment 8848015 [details] and Chrome 79 score is 18-20ms. So, now, Nightly is faster than Chrome at least on my environment. But I've not checked on other environments (Win10 + Intel CPU, macOS, Linux and Android) yet.
Comment 59•5 years ago
|
||
On Mac I get pretty noisy numbers, but generally 22-24ms on Chrome and 26-30ms on today's nightly, with outliers down to 18 and up to 30 on both...
(This is a mid-2017 15" MBP with the 3.1 GHz i7-7920HQ, in case that matters.)
Assignee | ||
Comment 60•5 years ago
|
||
Thank you, Boris. I also tested on Win10 on Intel CPU. Then, Nightly is faster than Chrome.
If we're slower only on macOS, I guess that the difference could be caused by the heap allocation cost. I usually see it as bottle neck only on macOS when I'm working on Quantum Flow.
Assignee | ||
Comment 61•5 years ago
|
||
On Linux, Nightly: 13-15ms, Chrome: 15-18ms.
But on Android, Nightly (Firefox Preview): 78-88ms, Chrome: 50-60ms. The profile is here: https://perfht.ml/2Qk31vc
Comment 62•5 years ago
|
||
Interesting.
Andrew, you might be familiar with compiler optimization settings we use on Android.
Wasn't there some plan to change optimization level?
(If one wants to find the relevant to this bug part of the profile, filter using HTMLInputElement.value)
Comment 63•5 years ago
|
||
(In reply to Olli Pettay [:smaug] from comment #62)
Interesting.
Andrew, you might be familiar with compiler optimization settings we use on Android.
Wasn't there some plan to change optimization level?(If one wants to find the relevant to this bug part of the profile, filter using HTMLInputElement.value)
https://bugzilla.mozilla.org/show_bug.cgi?id=1591725 for optimization. As historical reason, we still use -Os
and -Oz
due to package size of Fennec. Other platforms uses -O2
/ -O3
.
Comment 64•5 years ago
|
||
Yes, as Makoto mentioned, libxul.so on Android is still -Oz
for aggressive size optimizations.
Os
and -O2
are being considered, but we're trying to find a way to measure the impact on retention, etc.
Those options are roughly 10% faster than -Oz
, so that would help reduce or remove the gap.
Also, PGO just landed for MacOS (I believe Dec 19), so that should help as well.
Assignee | ||
Comment 65•5 years ago
|
||
Probably by the fix of bug 1591725, Fenix Nighty score is now 57-78ms on my tablet. But Chome runs 50ms-60ms. So, still a little bit slower than Chrome.
Assignee | ||
Comment 67•3 years ago
|
||
On my Galaxy Tab S5e, Fenix Nightly equals or faster then Chrome. Perhaps, stop deleting Text
node from TextEditor
and the reorganizing editor class inheritance improved the score.
Assignee | ||
Comment 68•3 years ago
|
||
Hmm, on my Pixel 3a, Fenix Nightly takes about 60~70ms, but Chrome takes about 50ms. Although on the desktop, now, Firefox Nightly (8ms) is really faster than Chrome (12ms) though. This meta bug is too long. So, I'll fix this for Desktop and file a new bug for Fenix.
Updated•3 years ago
|
Description
•