Mozilla: Firefox Quantum twice as fast as Firefox six months ago
Firefox 57 is twice as fast as Firefox versions that were released six months ago according to Mozilla, the organization behind the web browser.
Mozilla decided to give Firefox 57 a special name, Firefox Quantum, to highlight the speed gains of that version. The organization plans to release Firefox 57 on November 14th, 2017 to the Firefox release channel.
Mozilla used the web benchmark Speedometer 2.0 to benchmark Firefox 52, a version of Firefox released in March 2017, with Firefox 57 -- Firefox Quantum -- which is available in the Beta and Developer channel currently.
Speedometer 2.0 is still in development, but the results highlight that a lot of things have improved in Firefox 57 compared to versions of the browser released half a year ago.
Note: I ran the benchmark in Firefox 56 and 57. The Firefox 56 run returned a rating of "infinite" after the run, the Firefox 57 run a result that was about 20 points less than the result of Google Chrome (95 to 75).
A new blog post on the official Mozilla blog reveals how Mozilla managed to make Firefox Quantum faster. One of the core changes is the effective utilization of multi-core CPUs, for instance when it comes to laying out pages.
This improved utilization of your computer’s hardware makes Firefox Quantum dramatically faster. One example: we’ve developed a breakthrough approach to laying out pages: a super fast CSS engine written in Rust, a systems programming language that Mozilla pioneered. Firefox’s new CSS engine runs quickly, in parallel across multiple CPU cores, instead of running in one slower sequence on a single core. No other browser can do this.
Firefox includes other improvements, prioritization of foreground tabs and a multi-process architecture. Mozilla ran an initiative in the past couple of months to identify things that slow down the browser, and to find ways to eliminate these slowdowns altogether, or to speed things up at the very least.
468 of these issues were identified by Firefox engineers according to Mozilla in that time and corrected.
While speed gains are important when you compare one browser version to another, it is also important how Firefox Quantum stacks up against its main competitor Chrome.
Mozilla tested the web load performance of several important web properties, Google Search, Yelp, YouTube, Yahoo Search and others, and published a video of its findings.
Firefox does not beat Chrome in all of the load tests, but there is no instance where the browser is lagging behind severely. Interestingly enough, Firefox manages to load some Google-owned pages, Google's own login page for instance, faster than Google Chrome does.
Check out the video below to see a side by side comparison of load times. The usual reservations when it comes to benchmarks created by organizations or companies that showcase their products apply.
Interested users may download Firefox Beta, Developer or Nightly to experience the speed gains first hand.
Additional information is available on the Mozilla Hacks blog.
Closing Words
I did not measure Firefox's performance, but it seems to have improved a lot in Firefox 57 when compared to previous versions of the browser. That's a good thing and something that Firefox really needs to keep up with Google Chrome.
Now You: Have you tried Firefox 57 builds yet? What's your impression of the speed of the browser?
I found Firefox 58 to be slower than Chromium 61 on Linux, and also to consume more RAM with 10 tabs opened. Good thing they haven’t claimed Quantum to be faster than Crhome.
I think the coming invasion of Bitcoin mining scripts on hacked sites, will ABSORB whatever gains are made by Quantum.
I don’t see corn-hive on my EasyList and EasyPrivacy filters, so I’m manually adding it to the menu settings for javascript blocking on Chrome and other browsers.
Installed the v57 beta yesterday.. It’s twice as faster as chrome with everything it does!!.. No benchmarks needed.. It’s that fast. Lastpass extension currently not working(should after final release)..only thing keeping me from switching full time
I feel Mozilla should have allowed users to try this along with the older versions. It would have given them a lot of feedback to improve upon.
It should be possible to install both Beta and Release on the same computer at the same time, shouldn’t it ? I would make a profile backup before trying but I’m pretty sure I’ve seen desktops with both Nightly and Release, the question being, did they have to do any special tweak to make it happen, or did they just install both the regular way.
Either way you can use this non exhaustive list to find add-on replacements: https://docs.google.com/spreadsheets/d/1TFcEXMcKrwoIAECIVyBU0GPoSmRqZ7A0VBvqeKYVSww/edit?usp=sharing
Or go to Reddit’s Firefox sub and ask for advice.
I checked this for a day . Though fast, almost all my addons disabled except U block, click & clean.
As I cannot survive without addons, reverted back .
Sorry for posting again. Screenshot of network connections made by a clean FF57 nightly with ublock and tineye extensions only, without any tabs – just after fresh start it starts to connect to multiple domains.
https://imgur.com/a/Mjwke
Maybe its connecting to 127.0.0.1 or 0.0.0.0 and because you blocked coin-hive.com in your host file the connection appears as “coin-hive.com”?
I had the similar issue with an host used by WIndows Telemetry and Killswitch from Comodo
And as soon as Tineye is disabled, this stops even on your clean profile ? Assuming that Martin didn’t miss some obfuscated code in Tineye’s source, this is odd.
I’m going to assume that you installed Tineye from AMO and that you’ve done it multiple times, since you are talking about a clean profile.
I don’t see a point for a malware on your computer to only do this shit if Tineye is enabled, it would just do it all the time.
I guess someone else will have to install Tineye and watch firewall logs to try and reproduce the issue.
Martin – thanks for checking, but then it puzzles me why FF with tineye connects to that domain, and one without does not.
Have you checked if you a tab making requests to coin-hive? Or an program in your computer itself?
If you use uBlock Origin, you can check the requests made by a web site via the dynamic filtering pop-up. However, if you’re using it in the first place, it should block coin-hive.com with the default filter lists.
I’ve been using nightlies for months now. Much faster than before, I even dumped Palemoon, which I was an avid fan of.
Offtopic: coin mining
While watching new nightly performance in resource manager I noticed that firefox.exe makes connections to coin-hive.com. I have two extensions – ublock origin and tineye reverse search. Disabled the latter and coin-hive.com connections disappeared. Any comments? How and where should I post it, ask for clarification, warn others?
Doesn’t, the piratebay use that site for mining?
Unless it’s you, then someone else made a similar comment on AMO complaining that tineye make connections to coin-hive.com. if true then we need to spread the word on this issue.
https://addons.mozilla.org/en-US/firefox/addon/tineye-reverse-image-search/
MJ_3city’s first on top comment “Warning – possible coin mining “feature”!”
Looks like this extension makes connections to coin-hive.com. Would anyone please explain why? Seems like privacy breach. I’ve already sent info to AMO.
That was me. Let me underscore “was”, because that comment has been deleted by extension maintainer apparently. So, I asked it again. Someone posted it on reddit, but it was deleted quite fast as well (but title of this post could be misleading – hence deleted?). As no GPU load is visible then I should just forget about it, but anyway – it bothers me.
https://www.reddit.com/r/firefox/comments/72r3qc/tineye_reverse_image_search_extension_is_using_a/
I checked the source and there is no indication that it connects to the domain.
Speed is nice as long as its quest doesn’t occult the other aspects of a browser. What’s the point in driving a dragster from NY to LA? Comfort, availability of what makes a browser a surfing companion seems to me at least as important as speed. I’ve noticed a considerable difference already with Firefox 55 without Electrolysis, especially concerning the start-up delay. Otherwise pages load fast… I’m not sure I’d feel a velocity increase under a half second. I wonder sometimes if there isn’t a sort of speed hysteria, speed like energy, quantity (reversed but the idea is the same) with quality playing a secondary role. If I run whatever browser out of the box, no extension, it’s always fast. The choice is to forget the dragster and think about the comfort of a solid well equipped automobile. That’s how I see it. Finally, is the primary aim of e10 security or speed? What more security? Avoiding a browser crash? Never crashed here. An overall security background (system & browser) seems secure enough for me. I still don’t understand the terrific Webextensions and Electrolysis United Odyssey otherwise than as a go-go-go run in the mad race of browsers.
maybe also use a different speed test than the one from mozilla?
personally i don’t care really about my browser speed, the extensions i use are much more important. and right now, there’s no useful replacement for most of them.
so… sticking to ESR as long as possible and check again later, but honestly, i’m not all that optimistic.
Ok, did some tests on my older system, i7-3770k @4.5 GHz, nvidia GTX 680, 32 GB RAM
Firefox 52.4.0 ESR vs 57b3, both 32 bit, empty profile used.
Speedometer 2 Score:
FF 52 -> 36.3
FF 57 -> 70.3
Octane Score:
FF 52 -> 36329
FF 57 -> 37217
Basemark Score:
FF 52 -> 602.99
FF 57 -> 585.98
Jetstream Score:
FF 52 -> 197.86
FF 57 -> 212.44
Speedometer really is a lot faster, but not much difference for the other tests.
but honestly, i don’t know which one(s) represents a typical browser use case.
there should be a test that loads the 100 most used sites in my browsing history, or where i can specify manually some sites i want to use.
Note that JavaScript optimization has kind of hit a ceiling in all browsers, so it’s to be expected that JS benchmarks aren’t spectacular improvements.
The bottleneck nowadays is not JS raw performance as much as parsing, compiling, things that occur synchronously, layout, rendering, DOM parsing, loading times, input latency, i.e. anything but JS raw performance.
Super-fast Servo uses Firefox’s JS engine so that’s not really a bottleneck.
Thanks for reporting the numbers :)
Speedometer is supposed to be more closely related to a typical browser use case. It’s not perfect, especially if you have add-ons, e.g. uBlock Origin used to slow down speedometer a lot but that was due to a use case that would very rarely happen in real life, but was all over the place in Speedometer.
Octane is Google’s benchmark, it tests JavaScript performance. No longer maintained since April 2017. V8 replaced it.
Jetstream is Apple’s benchmark, it also tests JS.
Basemark I don’t know what it’s worth. Currently Speedometer seems to be the most recognized…
Speedometer is not from Mozilla
Firefox 57 is getting great, but still has some rough edges…
For example, while watching a stream on Twitch, the browser easily uses up to 500MB-600MB on a single proccess, compared to 250MB-300MB on Chrome. The CPU usage is also higher than Chrome.
So yeah, depending on what you’re doing, is Firefox who eats up all your RAM (and in this case, CPU), not Chrome.
Another problem is that Gecko can’t deal with complex web apps (like WhatsApp Web, Discord and some others) yet.
With that said, I really think that their focus now should be on Javascript and HTML5 video performance. Not sure if WebRender is that so important at the moment seeing that there are still major regressions that affects a lot of widely used services.
Firefox 57 is definitely worse than Chrome for Twitch. It’s a nice and smooth 60fps on Chrome whereas the Firefox version feels like it’s chugging along at maybe 20fps.
I have no problem with Discord and other apps though… Sure it’s not your setup ?
Haven’t tried Whatsapp.
Well, I exaggerated when I said that they’re neglecting… definitely not the case. I’m just a little impacient about it, that’s all. :P
I only hope that they do an uplift to 57/56 if these regressions are fixed on Nightly v58.
Taken from your first link, recent activity points to this bug that is going to also fix yours : https://bugzilla.mozilla.org/show_bug.cgi?id=1377253
There’s ongoing work on this one.
The second one has been reported 16 days ago and got someone assigned to it 14 days ago. It’s fast.
So I wouldn’t call it “neglect”… For memory leaks on Twitch and Google Maps, submitting a performance profile to Bugzilla might help: https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Reporting_a_Performance_Problem
Hope things will get fixed. Sometimes though, the bug is on the website’s end. In which case the place to report to is here: https://webcompat.com/
https://bugzilla.mozilla.org/show_bug.cgi?id=1377253
Sorry, I forgot to mention that… I can’t speak for Discord because I don’t use it, but I listed it there because I heard some people complaining about it. Guess this isn’t the case then, and I was wrong…
However, for WhatsApp Web I can confirm that there’re some performance issues with it, which are: a input delay[1] regression and a problem related to the use of the built-in zoom[2] on it.
Unfortunately, I’m watching these bugs closely for some time now and there’s no sign of fix from the devs. This really saddens me because it’s a big dent on the recent improvements that Firefox has gotten.
IMHO, it’s really a bad idea to neglect a popular web app like this one. Imagine the average joe trying to switch to Firefox when 57 hits stable, only to discover that the web app that he uses frequently performs badly on the browser.
This along with some memory leaks (mainly in Twitch and Google Maps) is the reason I can’t switch from Chrome yet (and I tried numerous times).
[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1376536
[2] https://bugzilla.mozilla.org/show_bug.cgi?id=1398697
This comment is besides the topic of the article
Imagus is now a WebExtension
I clicked the update button on the about:addons page of the Firefox web browser that I use, to my surprise the ‘Imagus’ addon lost its it’s ‘Legacy’ tag which could only mean that it is no longer a Legacy Addon
I have disabled all ‘Legacy Addons’ but I continue to keep them in the browser in the hope that one day they will be made into WebExtensions
“It is also important how Firefox Quantum stacks up against its main competitor Chrome.”
Not really. There are reasons many of us use Firefox over Chrome regardless of “browser speed”, and that is privacy. Comparing Firefox with previous versions makes a lot of sense for those of us that will never be using Chrome (or Edge).
Privacy is also improving significantly with First party isolation, fingerprinting resistance, containers, and various other items from the Tor Uplift project. As usual, this requires a couple tweaks to get real good. (And then you can ramp up and reach stupidly high levels with more tweaks)
I was just reading about Quantum, Servo and Rust last night and it all sounds very impressive. As far as page load times go I had noticed for some time now that it depends on the specific page on which is faster but Nighlty has definitely improved the percentage for FF. I don’t even worry about anymore, they are both very similar and very fast. Without the dev tools open it’s hard to perceive who crossed the finish line first. In Chrome I’ve noticed that having QUIC enabled makes a difference somtimes but then at times it feels to me like it does the opposite of help, I have it disabled.
Anyway, I hope that in the long-term the improvements Mozilla is making will help FF put a dent in the goolag’s dominance which will help everyone, including Chrome users. Wishful thinking? Probably. ;)
Just a few thoughts…
Speed can be quite subjective. While benchmarks can be a nice measure, I value the overall experience. A “snappy” interface can do a lot to improve the feel of a software.
I haven’t tried the latest Firefox versions, but historically FF always felt sluggish to me. Probably much of this was due to the UI code.
A while ago I noticed that Firefox and Chrome load pages differently when you click a link on a slow site. Firefox stayed on the page longer, waiting for the response. Chrome switched to the new page faster (even if it had not finished loading the page). This made FF appear slower.
when it comes to how snappy it feels under load, fx 57 is a huge improvement, I highly recommend trying out the developer edition if you haven’t yet.
I’m all for staying on the previous page until the next is at least readable. Better than watching an empty screen. The trick is then to get to first paint as fast as possible and prioritize the most important element, which has been a Mozilla focus throughout this project and will continuer to be.
Agreed. It’s a fine line though, when to change page for best experience. If you have a server taking several seconds just to respond (SharePoint Online, I’m looking at you!), how does the user experience the page load when after each click nothing happens for 3-5 seconds? I must say that I’m looking forward to testing the newer FF versions though!
There’s more to come though. Quantum has 5 parts, 4 major ones. Out of these 4 only two are shipped with Firefox 57. A large one, WebRender, is planned for 59. The other missing is Quantum DOM, about which I don’t know enough to tell how big it is when it comes to performance, smoothness and responsiveness matters.
Like Stylo (shipped with 57, and possibly Firefox for Android 58), WebRender almost x2 speed when you have 2 cores, x4 with 4 cores, x8 with 8 cores, on the tasks it is responsible for. (Stylo: styling, WebRender: rendering)
Like Stylo and Quantum DOM it is written in Rust, which is what makes any of this possible in the first place, but also notably improves security.
Firefox 57 also has sandboxing fully or almost fully implemented.
What I would like to know is what’s left to import from Servo ?
The less you can do, the faster you can go.
I have posted this comment on Ghacks through the years, it is still true with vastly different hardware, new version, old version, no addons/extensions, many addons/extensions, flash, no flash, Linux, Windows …take your pick.
Firefox RAM usage >>>> Chrome RAM usage.
Hey, I filed an issue about RAM usage, right here: https://bugzilla.mozilla.org/show_bug.cgi?id=1376753, can you please post a comment about your tests there?
Firefox’s developers didn’t believe me about Firefox’s RAM Usage.
Thanks.
They’re not not believing you, they need more information in order to move forward.
” Please provide a copy of about:support and a memory report, and see if you can reproduce with all of these except uBO disabled (I’m assuming the others aren’t webextensions). “
You know Firefox developers are waiting for your input in that bug, right?
@RG: I don’t disagree. However, the primary reason I left Firefox: Firefox uses a lot more CPU and is way less responsive than Chrome.
And I (personally) haven’t had any RAM issues using Chrome (I have 8GB and will buy 16 or more with next computer). I understand that for some people Chrome’s RAM is an issue. But for me, I’d rather have less CPU usage and a responsive web-browser. Firefox failed miserably at these. v57 is much improved but still not at Chrome level.
Poor Flash was too good for us ungrateful.
Will try 57 when it comes out. Long time ago I blamed flash but slowly and surely realized that’s not it
By which you mean better, and not higher :)
It is still true, however Firefox 57 has RAM a regression on mac and now consumes as much RAM as Chrome (still much less than Safary though). Mozilla developers plan to work on reducing RAM consumption further after 57 on all platforms and that includes the mac regression.
The “norm” is supposed to be that until E10S, Firefox used way less RAM than the other big browsers, and the gap, although still significant, has decreased with multi-process and stuff. Also 64-bit uses more than 32-bit.
So basically Firefox 57 should use more RAM than say Firefox 45, but still noticeably less than other browsers, with one exception on Mac only. And from 57 onward, there should be focus to fix the regression on Mac and improve decrease RAM all around so that the amount of content processes can be increased some more. So, say, Firefox 61 will use less RAM than Firefox 57 (but more than Firefox 45 due to multi-process and 64-bit).
But if it’s all backwards on your computer, nothing can be predicted.
Good ‘catch’ ;)
Seriously though, higher for me. Haven’t tried 57 yet, not doing the nightly/beta stuff anymore, when it is released will certainly test.
This just shows how bad Firefox had become. Chrome is still faster than v57 Firefox on my linux box.
If you were a fan of K-Meleon, sounds like we may be witnessing its demise. The K-Meleon site is down. Kinda sad, if that’s indeed the case.
The site still works but K-Meleon has been maintained no longer for two years
http://kmeleonbrowser.org/
the linux version is a red headed stepchild that lacks all of the graphical hardware acceleration goodies that the windows and osx versions have, though I think a fair amount of that can be blamed on the poor state of the linux graphics stack.
Chrome has better hardware acceleration on linux mainly because google’s chromeos is linux based and they’ve been able to put more resources into it.
I’m finding the opposite is true here. Firefox 57/58 is much faster than Chrome for me.
I have tried some of the FF 57 builds. I definitely liked what I used, though the re-design of the browser definitely takes some getting used to. That being said, anything that helps put a dent in Chromes dominance is a good thing by me.
“It seems to have improved a lot.” – smooth transitions from site-to-site, smooth scrolling, great news video playback on YouTube from AlJazeera – FF 57 B-3.