Comment by DiabloD3 | Hacker News Reader

Comment by DiabloD3 | Hacker News Reader

As someone that jumped ship from how slow Chrome has gotten and went to Nightly /w Stylo and (CPU threads - 1) e10s processes enabled about 2 monthsish ago...

Holy damn Firefox is actually fast.

Fair comparison, both having Pocket, uBlock, Evernote, OneNote, Pushbullet, and Bitwarden; neither of them having an extension the other doesn't. Above described Firefox Nightly config vs Chrome Dev.

Test machines are a workstation with a i7-4771 @ 3.9ghz, 32GB DDR3-2133, Radeon 7970 with a trio of 1080p screens (a very fast modern machine from the Haswell era) and a laptop with a i5-32120M, 8GB of DDR3-1600, Intel HD4000 iGPU feeding a 13" 2560x1600 @ 200% hidpi (an Ivy Bridge era MBPr 13" Late 2012). Both machines run Win10.

Both machines have less real world wait on Firefox than Chrome, and the interface has less latency between when I do something and it even begins processing the request. Also, under a ton of windows and tabs, Firefox seems to use less RAM and the speed gap seems to widen.

I don't care about about artificial benchmarks, btw, they never seem to measure what actually makes browsers slower for humans.

Edit: Even though I just said I don't care about benchmarks, using Speedometer 2.0-r2216: On the workstation: Firefox 58.0a1 2017-09-26 64bit 61.18 vs Chrome 63.0.3217.0 dev 64-bit 51.67; on the laptop, same versions: Firefox 28.58 vs Chrome 26.74

So, arguably, flat out benchmarkable performance is the same, with Firefox just slightly edging ahead (18% and 6% faster). It isn't enough to explain how fast Firefox feels now.

Edit 2: And now with Edge 40.15063.0.0/EdgeHTML 15.15063: Workstation, 46.72; Laptop 21.37. Firefox is 34% and 31% faster, Chrome is 25% and 11% faster.

> It isn't enough to explain how fast Firefox feels now.

That's because we've been making a dedicated effort to find and stamp out sources of UI lag! It's called Quantum Flow; see https://wiki.mozilla.org/Quantum/Flow for more.

I beg you to boost performance at least upto the level of chrome even WITH addons. That's the main issue. With addons like ublock origin and other content filters, the cpu usage reaches 50-60% and the page rendering is not that quick as compared to chrome even with addons. Chrome's UI is seriously so liquid smooth that after being with firefox for so many years for my love for privacy, I HAD to switch to chrome sacrificing my privacy. I still use heavy privacy addons but chrome still doesn't get slow noticeably.

I highly appreciate what you doing. Hoping to get back on Firefox as soon as possible.

Once again please take care of those sudden cpu and ram usage with almost necessary addons like ublock, tab suspender etc.

I never thought I'd say this but the 2 fastest bowsers on my computer are Firefox and Edge

Well that's because Chrome is busy doing random crap like this in the background - chrome://site-engagement/

I really doubt that that feature has any performance impact. It looks like a bunch of counters.

every browser needs to have some counter like that if it wants to suggest pages when you open a new tab or something like that.

i think it's cool that you can see the actual ranking.

Intended use cases are documented here: https://www.chromium.org/developers/design-documents/site-en...

If you want to trace uses of it in the code, here is a reasonable place to start, I imagine: https://cs.chromium.org/chromium/src/chrome/browser/engageme...

I guess you've never checked ~/.mozilla/firefox/YOUR-PROFILE/places.sqlite which includes a record, including date & referer, for every time you visit a site. See https://developer.mozilla.org/en-US/docs/Mozilla/Tech/Places... for documentation (or http://www.forensicswiki.org/wiki/Mozilla_Firefox_3_History_... for unofficial documentation). Several hundred thousand rows here...

This is one of the things which really slows Firefox down for me. I often found in the past I had to blow away chunks of Sqlite in my profile to get back to a decent speed.

Dillo seems to be the fastest on mine. I imagine w3m would be faster, but I use it through Emacs which makes it a bit slower.

Unfortunately, as long as Firefox security infrastructure does not match Chrome, I can not use it. Note that errors like use-after-free and buffer overflows are marked as Critical (run arbitrary code on computer) for Firefox, but marked as High (execute code in the context of, or otherwise impersonate other origins.) for Chrome because of sandboxing.

Firefox: https://www.mozilla.org/en-US/security/known-vulnerabilities...

Chrome: https://chromereleases.googleblog.com/search/label/Stable%20...

Hopefully with their sandboxing project done (Also with Rust) will make Firefox much better in the security department, but until then, no.

Firefox sandboxing does not use Rust. Servo sandboxing (work in progress) does.

It's not accurate to say that Firefox has no sandbox. Sandboxing work has been rolling out for many months now, and later versions have increasingly restrictive sandboxes. You can see the current status per version here: https://wiki.mozilla.org/Security/Sandbox#Current_Status

Hardening the sandbox is a constant work in progress, but it's not as simple as "compromise of browser engine automatically means arbitrary code execution as running user" any longer.

Sorry I didn't mean Sandboxing used Rust, It was a separate thing (That Rust is less susceptible to such class of errors, so the more code written in Rust the better) Also didn't mean it has no sandboxing (I knew about efforts), my point was its security is not a match for Chrome yet, as seen from the amount and type of critical errors.

When it is there I would definitely use it along with Chrome.

In case it wasn't clear, it is now the case that content process compromise is no longer critical unless combined with sandbox escape.

Looking at historical vulnerability reports can be misleading in this regard, as the sandboxing features are rather recent (pre-stable-version in some cases).

If I wasn't clear as well, As far as I can see from the link you gave It still does not match the level of security Chrome offers, it is getting there, but not yet.

As Mr. Patrick Walton over there said, Mozilla is still building up the exact, minimal file-system permissions. So, yes, in that way it is still not yet on quite the same level as Chrome.

In terms of security architecture, there is to my knowledge only one bigger difference left and you can change that, if you want.

The difference is that Chrome will spawn a new process for every new tab (unless the webpage in it is from the same domain as another tab). Firefox instead will always round-robin the tabs across a fixed number of processes to achieve lower RAM usage and as result of that also somewhat better performance.

But you can tell Firefox to round-robin across up to 1000 processes or what have you, so that it then does spawn a new process for every new tab (and therefore sandboxes each tab individually).

To do so, go into about:config and set "dom.ipc.processCount" to a high number, like 1000.

I guess as long as it is not default behavior it will stay as a weakness then? I am not sure if it is worth to exchange it for memory or performance gain.

Also is using thread per tab approach more susceptible to memory leaks, resource sharing issues or OS scheduling shenanigans than process per tab approach? You close the tab, the process is dead and resources are mostly guaranteed to returned. My knowledge of modern browsers is limited, maybe this was considered an acceptable compromise though.

Perhaps that now they are using Rust instead of C++, leaks and memory corruption should be much less of an issue.

Though the Firefox developers aren't going to skimp on sandboxing (or other exploit mitigation techniques) just because of Rust. Defense-in-depth is the name of the game; Rust just provides a layer of language-level defense that C++ previously didn't offer.

Chrome will round-robin across processes too; it has a cap on the process count. The cap depends on things like amount of RAM. See https://www.chromium.org/developers/design-documents/process... fourth bullet.

I believe they made that change in Chrome 55:

https://v8project.blogspot.ro/2016/10/fall-cleaning-optimizi...

I'd like to see the same comparison on a Mac. The Windows version of Firefox has been close to Chrome performance for a while. The Mac version has been significantly slower.

I've been using Firefox Nightly (57) for a while on a 2013 MBP running Sierra. Firefox definitely feels significantly faster than it used to, and the new UI is great.

But this comes at the cost of battery life. I moved back to Chrome because it was constantly chewing CPU.

+1 for the significantly improved performance on Firefox on OSX (I'm on a 2015 MBP), not just in the current 57 nightly, but since Firefox 55 came out.

Absolutely stunning improvement in performance!

Great work by team FFx! :D

Yep, switched to FF from Chromium since 54/55 when multi-threaded tabs became default. Don't need to restart once in a while when Chromium has eaten up all the RAM. The extension support is just as good, and some of the labs features are very interesting.

But Chrome is also guilty of absolutely guzzling battery. I always try other browsers on macOS but in the end keep switching back to Safari because it easily gets me ~2-3h more battery life. This means that on 10h average battery life you'd pay a ~20-33% battery life premium just to run a different browser. At that point, finding replacement Safari extensions is well worth the bother.

What I don't understand: why is it that other browsers don't put battery life before all else? We live in a portable world after all..

>We live in a portable world after all..

Do we? I'd say most people with laptops use them either at home or at work, both places were they can plug in.

The majority is not "Starbucks laptop users" or road warriors.

I mostly use mine on the train during the commute. I'm one of many. At home and work I use desktops.

Mozilla will also want to target their engine at tablets and phones at some point. This is presumably why Samsung committed funding to Servo (IIRC).

I switch to Safari when on battery. But the cost is pretty high - it's the only browser I have on my mac that I find actively unreliable, and the feeble range of available extensions makes it seem crippled compared to Firefox or Chrome

A few releases ago Chrome (57?) finally fixed the battery problem and its been fine for me since.

Personally I can't use Safari, as I want my bookmarks/history synced between devices and don't use an iPhone.

Just ran the tests on my Macbook Pro (mid 2012) running High Sierra:

  Safari TP 11.1 Release 40  - 62.71
  Chrome Canary 63.0.3221.0  - 55.45
  Firefox Dev Edition 57.0b3 - 44.17

Not sure whether this would be borne out on later Macs but this seems to confirm my anecdotal experience on my Mac in the last few days, even since 57 came out.

While we are doing scores, here is Fedora 26, Threadripper 1900X with 32 GB RAM:

    Firefox 55.0.3              - 71.4
    Firefox Dev Edition 57.0b3  - 113
    Chrome 61.0.3163.100        - 136

Firefox still does not use hw acceleration for composition.

These are Browserbench 1 though, I have no idea how to run v2.

+1 for mentioning huge CPU usage by Nightly on Mac.

I've been sending feedbacks for Nightly for a while now, but they don't seem to read them. The two biggest issues are:

1- High CPU usage (which leads to high temperature and shorter battery life) 2- No "Look Up / Define" in right-click menu.

Fix them and I'll switch from Chrome in a blink.

> in a blink Funny even though you probably didn't intent it to be.

I had to switch back to Chrome since Firefox has been almost unusable on mainstream site like Netflix. Can't wait to go back to Firefox as soon as 57 is out!

I switched to Chrome years ago and every time I try Firefox it feels sluggish, so I don't switch back.

Last time was a week ago :\

This has been my experience also.

However, I'll try Quantum and see if it improves the situation. I really don't want to be using Chrome.

57 really is a huge improvement, I can no longer feel any difference in responsiveness between firefox and chrome.

In some ways I think firefox has actually surpassed chrome (e.g. on my xps 13 touchpad scrolling is much more responsive/smoother in firefox)

Re: touchpad scrolling, make sure you disable "smooth" (animated) scrolling and enable precise scrolling; on Linux this requires setting MOZ_USE_XINPUT2=1 in the environment.

Nice, thanks!

Everything feels 10 times faster without smooth scrolling, while I have to say that the nightly already felt blazing fast _with_ smooth scrolling, haha.

This should be off by default. Feels and looks faster. What is the downside?

Tracking the thread through https://bugzilla.mozilla.org/show_bug.cgi?id=1207700 and https://bugzilla.mozilla.org/show_bug.cgi?id=1170342 (and ultimately https://bugs.freedesktop.org/show_bug.cgi?id=93539 ) it appears that when XI2 is used, focus-out events were not sent in rare cases under old versions of Xorg. If you're running modern software there should be no downside.

Thanks. Could the Mozilla installer detect the version of Xorg and set ?

The key is, which version a week ago? Unless it was nightly you wouldn't have seen any of the stuff in the OP.

I see, okay.

I just had the feeling that Firefox has news to be on-par with Chrome for years now, but I never really "felt" it, hehe.

The current news relates to merging parts of Servo (web engine written in Rust) to Firefox, which started relatively recently.

I know that. And I have high hopes :)

I just wanted to say, that it isn't the first news about Firefox getting faster since Chrome came out.

I tried the new versions every now and then and Chrome always felt faster to me. Maybe it really is, maybe it just uses some perceptional tricks, I don'T know.

Firefox really has been slowly getting faster for a while; especially with Firefox 54, I no longer felt as if I was taking a significant performance hit as the price for not using Chrome. Firefox 55 got even a bit faster, but still was not quite as fast as Chrome. However, the recent changes in Nightly (Firefox 57), which now landed in the developer edition, really was a huge leap forward in performance.

I encourage you to try out Nightly now (https://www.mozilla.org/en-US/firefox/channel/desktop/). I’m not promising that it’ll be faster than Chrome, but it just might be fast enough now.

Did you try current stable firefox or Developer edition or Nightly edition?

If I understand correctly, the photon changes haven't landed on current stable firefox yet. They have just landed in Developer edition. So you might want to give it (or nightly edition) a try.

Just try firefox nightly 58.0a1 ! Firefox nightly is my default browser nowadays. I love the modern UI.

I installed that nightly and it really feels fast and modern.

They even got rid of that search bar, haha.

Only fonts and the scroll bars seem a bit old school compared to Chrome.

Also, WhatsApp Web keeps logging me out.

> Also, WhatsApp Web keeps logging me out.

This is an issue on my Arch Linux laptop, but not at work using Ubuntu. Both run Firefox nightly. Been going nuts with this issue!

I have it on my Windows7 machine with Firefox Nightly, but not on Chrome.

Works on my Arch …

> They even got rid of that search bar, haha.

It's actually configurable by the user. I like having it on the side, so I added it back.

Mozilla is experimenting with which default is preferred by fresh install users right now, so users on this thread may have different initial defaults. I don't think that there are plans to change the default for existing users.

How? I was looking for this last week and couldn't find any settings that worked. Nothing relevant in about:config, and the toolbar customization shows the search bar there until I close customization.

Go to customize mode and hover your mouse pointer on the search bar. You'll notice that your mouse pointer turns into a hand, then just drag the search bar and drop it into the center of the screen (blank space).

(Tested on Firefox 55.0.2/Linux)

Cheers. :)

They actually very recently added a preference for this: https://dolske.files.wordpress.com/2017/09/searchprefs.png?w...

> fonts

Chrome has a bunch of bugs related to font-rendering on Windows, e.g. https://crbug.com/414307. I'm not sure why switching to DirectWrite didn't just fix these bugs, but subjectively appearance-wise on Windows, Firefox > Edge > Chrome.

okay.

To me it seemed like the input font here on HN could need a bit more hinting in FF.

Check your addons. They can really slow things down.

Was it nightly? I was in the same boat with the standard release, everytime I tried it it felt sluggish. Then I tried nightly and all I can say is wow, no.other browser is as snappy as FF nightly

Nightly has been amazing. I’m just waiting on LastPass before I switch full time.

Switch to Bitwarden, already has a WebExtension extension. Lastpass is basically dead ever since LogMeIn proved they don't care about security of the vault.

Hmm. Yeah I’ve been wondering about lastpass for a little while now.

Is there a comparison of these tools that you’d recommend?

There is no good comparison website I'm aware of, but Bitwarden ticks every box Lastpass does, and is already working on ticking boxes Lastpass doesn't.

Bitwarden doesn’t have support for credit cards/ addresses/ SSNs...

Yes, it does. Source: I just switched hours ago and it imported my CCs from LastPass. You can see the stuff they're working on via their Twitter account too.

Thanks for mentioning Bitwarden. I didn't know about them, but I've had the same concerns about Laspass/Logmein. Yubikey functionality is important to me, which many alternatives don't have, but I see Bitwarden does have that. Their GPL v3 license on all of their software is nice to see too.

Dead to me ever since they decided to raise the price with 100% for no reason (from consumer PoV) whatsoever. Classic bait & switch.

Is there an easy way to convert your current Lastpass database to Bitwarden?

Bitwarden can import Lastpass-made CSV exports (plus exports from literally everyone).

Yes: https://help.bitwarden.com/article/import-from-lastpass/

Thanks, both of you.

Seems 2FA (YubiKey) costs 10 USD/year instead of 24 USD/year which LastPass costs.

I've been looking for a recent feature comparison, the Wikipedia one is rather shallow and incomplete [1] [2].

My (incomplete) take on feature comparison:

Both are relatively cross-platform, widely supported by what I consider status quo platforms (Windows, Linux, Android, macOS, iOS, Chrome, Firefox).

Advantages of Bitwarden are open source and public development, approx half the price (and approx same as Lastpass before purchased by Logmein), and the ability to self-host.

Advantages of Lastpass are different login for accessing to account and decrypting database. This is important because without this, a MITM attack gets access to the master password.

I also found the announcement of Bitwarden, approx 1 year old now, on HN [3]. It includes some feedback about the software. Bitwarden uses AES256-CBC. How secure is AES256-CBC in this case, and what does Lastpass use? Another criticism is usage of Google Analytics. This can be disabled in the application, but I don't feel it serves a purpose for the user. It should never have been included in the first place. I also don't think either software is audited.

[1] https://en.wikipedia.org/wiki/List_of_password_managers

[2] https://en.wikipedia.org/wiki/Comparison_of_password_manager...

[3] https://news.ycombinator.com/item?id=12676979

I just switched... Thanks for the recommendation, it's excellent.

I'm a quite happy user of KeepPassXC as my password manager. I seriously dislike how LastPass stores the passwords on their servers. And KeepPassXC is open source and actively developed.

https://keepassxc.org/

Chrome is an insane ram hog. It is inexcusable to be using >400MB per tab, but I see that all the time.

I don't know how much my tabs use, but it isn't 400MB or I'd have ran out of RAM a hundred tabs ago...

I just checked on my work computer and you are right, only a few hovering around 300 or so. They may have made improvements or it may be a macOS thing

I am averaging around 150 MB per tab here.

Agreed. Microsoft Azure Portal is taking up 934 MB, Gmail is 502 MB, Pocket is 309 Mb. Google Keep takes 180 Mb or so.

What really amused me a couple of years ago is that even processes handling about:newtab were constantly leaking on linux.

And here I am just enjoying my Opera. Built-in ad blocker, built-in WhatsApp Web, great UI, ultra fast, supports all modern web technologies.

the WhatsApp web feature is something that i never thought i use as much as i do. Really nice to have in the sidebar.

Opera is just rebranded Chrome, with some badly-advertised "VPN" slapped on top of it, a shadow of it's former self (the Presto version).

Opera is using Chromium's engine ("Blink"), but the whole UI shell ("chrome") is custom, with many UI conventions from old Opera.

I use that built in whatsapp a lot more than I expected. It's great. I love opera, it's my go to browser for dev.

That said, I'm definitely going to give quantum a try

Enjoying the Chinese consortium that owns Opera too? I am not even joking here, given China's recent track record with WeChat I would think twice before using Opera to handle all your https (banking..etc.) traffic. Of course Chrome (US) suffers from somewhat similar issues. Firefox really has unique position in the market in that it's fully Open Source and has non-profit driven org behind it.

I am not a Chinese dissident, so Opera being owned by a Chinese company doesn't concern me. China has absolutely no interest in my activities.

I find your confidence in that to be rather touching. A testament to faith in humanity.

For historical note: that is Speedometer 2.0-r221659. Somehow missed the last two digits when I wrote this comment originally.

Also, with gfx.webrender.enabled, gfx.webrendest.enabled, gfx.webrender.blob-images, and gfx.webrender.layers-free all enabled, same version of Speedometer and Firefox on the workstation scores 63.47 (or 4% faster), but has render errors on the Speedometer's tach needle (SVG?) and renders fonts with very strong (very fatiguing to read) subpixel rendering.

I run 60 tabs in Firefox atm, takes 2.14GB (obviously I'm using Tree Style Tabs). Runs smoothly.

I'm at 204 with about 1 GB of usage. Firefox is great with many tabs http://www.techradar.com/news/firefoxs-blazing-speed-with-hu...

I've become a big fan of "The Great Suspender" extension since I am always opening new tabs in the background and may not get around to them until hours -- or sometimes days! -- later.

On Chrome I presume? Not available on FF, and afaik FF doesn't even support unloading tabs (actually unloading from memory rather than making it appear that way but still consuming the memory).

Ahhh, yes, on Chrome, sorry. I didn't realize it wasn't also available for FF.

Here's mine (fx55): http://i.cubeupload.com/AvH2Cc.png profile1: 6 windows, ~900 tabs; profile2: 2 windows, ~300 tabs. cold start takes about 2 minutes (on nvme ssd), next are slightly better. Startup time improvements were supposed to be in 55, right?

I've found that google maps performs worse in firefox and I get random %100 cpu spikes from websites in firefox that I didn't get in chrome. I'm on a mac using firefox 55.

I've found out that many google products (web pages) have bad performance outside of Chrome[0]. But I've used Netscape/Mozilla my whole life, and I will not change from it, specially now that its performance is measurably rising (again).

[0] which is of course something google devs should consider and fix the sites, but hey... why do that when you're the market leader?

Firefox 55 doesn't have any of the (massive) improvements from 57 and Project Quantum which are being discussed here. Give Beta, Developer Edition, or Nightly a spin.

I used Firefox Focus on my Pixel for a few weeks and was blown away by how fast it was. The mobile web went from borderline unusable to borderline pleasant with it, I imagine largely due to its tracking and limited ad blocking. Sadly, I migrated back to Chrome because anything that needs to work with Google login or opening an app simply doesn't work. This causes issues with a lot of things I use.

I feel like I'm being ActiveX'd into using Chrome now which is very non-Googly. Not sure how feasible it is to use a different mobile browser until Mozilla and Google work together to make sure they're all on the same API page on Android. Ironically enough, Windows is far more open and friendly to competing browsers than Android.

It's not as fast as Focus, but if you want an alternative to Chrome Firefox for Android should work better with Google Apps that are (probably) broken due to Focus's anti-tracking features. It is based on the same rendering engine as desktop and should benefit from many of the future speed improvements of Firefox 57.

On Android Chrome is arguably better if you want Google account integration or raw speed, but Firefox for Android has a more traditional UI, extensions, and is fast enough for most of my browsing.

I haven't had any issues using Firefox for Android... Firefox Focus is not a general purpose browser, so some issues are probably expected. Try Firefox for Android and install uBlock, or any of your other favorite desktop extensions!

I was blown away by Firefox Focus as well. I ended up making it the default browser, so its used when I read an article on HackerNews, Reddit, Twitter, etc. But when I need the "logged in" experience - autofill passwords, bookmarks, cookies, etc. - I launch Chrome. Best of both worlds!

I'm the same but my "logged in" mode browser is Brave. It's chrome + privacy

Yes! I do this as well after reading about it on some Reddit thread. Fantastic use case for opening random links from email and chat (and Focus is actually very smooth)!

Focus is, uh… WebKit/Chromium with no traditional browser session features (history/bookmarks/etc.) and a tracking blocker. Useful, but not "true Firefox" :)

I like Firefox Nightly on Android though, it's really good.

> Firefox Focus

Sorry, but unless you have a Gecko-based version (which - AFAIK - is not what you can get from Google Play or F-Droid), it's system WebView (i.e. Chromium) under the hood.

> I feel like I'm being ActiveX'd into using Chrome now which is very non-Googly

Um

Define "googly"

(If you're saying it's atypical of the behavior you'd expect of them: based on what?) Considering there's no trivial way to log into one of their services (gmail) without automatically doing so on others (yt etc) without fucking around with browser sessions - among many other non-customizability gotchas from them - saying google doesn't want to control how their offerings are used is going to be a very tough sell.

Reply on news.ycombinator.com