Why are browsers so slow?
ilyabirman.net
ilyabirman.net
This is exactly how I'm using Firefox, right now -- 273 tabs open. It's sitting at ~5% of CPU and 450MB of RAM. And switching tabs is instantaneous. And it doesn't reload every page when I quit and restart my browser -- it only loads a tab when I click into that tab. And it has tab groups so that my current group has only about 8 tabs in it, and the other groups are sorted by topic. And on and on.
I find it frustrating when people post these articles as if everyone has this problem, and don't provide enough details as to their setup so that people can help them fix it. I promise you -- if everyone around the world right now had the problem the author was having, it would have been solved. No one would stand for it. Rather than assume everyone is suffering just like you, assume that other people either a) don't behave the way you do, or b) have found a way to fix the problem.
Which platform are you on? Which Firefox version? Stock or tweaked? Any special add-ons? I use Firefox on OS X and it crawls after 10-15 tabs. Consumes many gigs of memory. Chrome isn't any better. It just splits the memory consumption among multiple processes.
I don't remember browsers being this slow on Linux. Time to go back to the magic land.
The Developer Edition replaced Aurora, which was the alpha release, one version between the beta and Nightly.
Credit to them though, Aurora was surprisingly stable for most of its lifetime, despite not even being a beta.
1: On Ubuntu you can use their Firefox Beta PPA² as long as you don't mind that it replaces release-version Firefox.
2: https://launchpad.net/~mozillateam/+archive/ubuntu/firefox-n...
Nightly is amazingly good even if it is built from the HEAD of the trunk (branch in git terms).
Firefox is working on this. Doing it without breaking things like music streaming sites is ... delicate.
Earlier this year I realized a major offender was Google search and search results pages which would spin up a core more than once a minute.
This explained to me at least why it was still hogging cpu even with ad blocker.
https://billmccloskey.wordpress.com/2016/10/27/mozillas-quan...
and a preliminary version is in Chrome 56:
It's depressing that an image viewing site and a video playing site are in the same list of resource uses. YouTube plays 1080p videos. Imgur shows photos. It's a testament to the brazen disrespect of client resources that went into building the latter.
If you're viewing imgur without ad blocker your computer isn't working 10x harder than it should, but 100x.
1. All the existing image hosting sites suck
2. Someone gets sick enough of it that they launch a new one, with all the features people want: free, direct linking, no/unobtrusive ads, etc.
3. Because this new site is so much better, everyone flicks to it. Bandwidth costs skyrocket.
4. The operators need to cover their costs, so they start adding more ads and blocking direct links. The site starts to suck. Go to step 1.
Sometimes investor dollars get involved, but it doesn't change the basic formula. This is why we've seen Photobucket -> minus -> giphy-> imgur -> gfycat, and it will continue in his vein forever. I don't think there's any way to run a site like this profitably without pissing people off sufficiently that they go elsewhere.
I'm thinking about something like a P2P solution, IPFS-style. You could pay for upload space in two ways - either directly, the bog standard way, or with your own drive space. So for instance, you get 1GB free space for image uploads if you agree to set aside 1GB on your drive as a cache. Couple that with your own "seeding" CDNs storing all the images, and maybe this would be enough to distribute bandwidth costs across all the people who want their images hosted?
Not sure if the browsers are capable enough to pull something like this off without relying on some sort of plugin or an application.
Before imgur, there was photobucket, imageshack, whatever - they were beyond dumpy jankey messes that you truly didn't know what you were going to get when you clicked on them. There was a good chance a new window was going to open behind your current target, or that the images would be intentionally low res until you clicked into them (for more CPM's for the host).
Imgur came along and stole their bread and butter, but then they took on a bunch of VC money and had to figure out how they were going to profit. I'm not sure if they've figured it out, but their site has definitely suffered - although not nearly to the degree as the predecessors that they replaced.
There's a lot of gifs that are converted to video (because it is 10x smaller when converted to video)
Whether you're displaying a static image or an animated GIF or a H.246 stream, drawing the next frame doesn't require interpreting or JIT compiling any JavaScript, it doesn't require thrashing the garbage collector, and it doesn't require re-computing the page layout.
I particularly like how imgurl flat-out refuses to work if you disable JavaScript. It's an image-viewing site. Browsers have been able to download images since the beginning, and have been able to render them inline since almost the beginning. There's no need for JavaScript at all, and yet imgurl deliberately choose not to work. 'Brazen disrespect' doesn't begin to describe it.
I don't mind that they try to use JavaScript to minimise the data they serve. They could always show one or a few images without JavaScript, and provide a link to more (and hide it when JavaScript is enabled).
Note that this has nothing to do with ad blocking — they could display all the ads they want to a non-script-executer.
'Brazen disrespect' really doesn't begin to describe it. 'Rank unprofessionalism' starts to.
But then it would be harder to track you.
(Well done uBlock!)
Except for when Adblocker (used to) makes your browser work harder https://blog.mozilla.org/nnethercote/2014/05/14/adblock-plus...
(FWIW, before electrolysis Firefox was unusuable with many tabs for me, now it is perfectly fine, so either I made a random setup change at the same time, some other optimization landed or it actually helps)
Tab Tree (instead of TreeStyleTabs) and Suspend Tab (instead of UnloadTab.)
Edit: and yes, Linux often feels like magic land. I happen to like Windows 10 better than most earlier Windowses but a good Linux is better for me as a developer on every point except support for MS Office and (niche) tools.
Yes, the browsers (Chrome, Firefox) work noticeably worse than on Windows for me. Also they tend to freeze the computer completely at some point (after taking all the available RAM for themselves).
Browser performance seems to be pretty much as random as printers these days.
Stock Firefox 50 (32bit), untweaked with no relevant add-ons (well, except AdBlock Plus), Windows 10, just over 400 tabs open, under 10% CPU, under 2GB of RAM (on a 16GB machine). If I freak it out by doing a bunch of flipping through dormant tabs, I can spike CPU pretty good as it does layout on fifty tabs at once, but who cares, I don't do that. I do find that I need to restart my browser every few hundred tab open/close cycles, which takes under 30 seconds (just did the restart, that dropped RAM use to 1.3GB, but it'll get back up to 1.9 pretty soon). Restarts used to be a lot less frequent, and a lot faster, before work required me to keep two slack clients open. Running multiple YouTube windows at once seems to be bad for uptime.
Regarding OP's complaints, opening a new tab take imperceptible time, a new window is about half a second (only marginally slower than notepad), and switching tabs can be done several times per second.
I have put ZERO effort into make Firefox run faster, unless you count running AdBlock. I haven't even bothered to figure out which is the best adblocker.
Why do people use Chrome, again?
(edit to add a few details)
HN looks like: http://i.imgur.com/lfjkhkl.png
It is dishonest and bad for the general public. Users expect their adblocker to block adds; not block 'some' adds because of Mafia practices, or corruption (ie. advertising companies paying large, inappropriate sums of money to the company behind ABP in order to be whitelisted by default).
> I like having the option to support sites I enjoy without massively inconveniencing myself.
You can still do this on a case-by-case scenario with uBlock Origin.
> You prefer maximum adblocking, and that's fine, but there's nothing at all wrong with giving us both the choice.
This is false dichotomy. You have the same very choice with uBlock Origin. And uMatrix for that matter. I'm also not taking away your choice; nothing prevents you from installing and using ABP. What I will not do is stopping calling what ABP does sellout. Because as I argued above it is corruption.
On top of that, uBlock Origin is performance-wise better than ABP [1].
You still seem to be under the impression that ABP is deceptive. It's not. It explains the "acceptable ads" list and gives you the option to disable it first thing after install. It is not doing anything without the user's knowledge. They're providing a perfectly honest service that you aren't interested in, and you think they should feel bad for that.
> You can still do this on a case-by-case scenario with uBlock Origin.
I specifically said without inconveniencing myself. Building up custom adblock settings for various sites is a chore and I have better things to do with my time. I choose to use a product that takes care of that chore for me, for the most part. You choose to take the time to personalize things, and that's fine; I've done the same in other circumstances. But that's a personal choice, not a moral one.
> I specifically said without inconveniencing myself.
You said you wanted to support your fav site. That is not hard to do with uBlock Origin. You are actually supporting most sites.
You also make it seem like the ABP whitelist isn't inconveniencing. It is; see above.
So do I, which is why I use Patreon and subscriptions.
The side effects of viewing ads are that in addition to supporting the sites you enjoy, you're supporting malware distributors, and helping companies compete on ads rather than the quality of their products. Maybe you don't care about your own attention or data security, but your actions don't just affect you.
No, I'm not, because I'm using ABP's vetted ad provider list. This risk of getting malware from one of those is greater than zero, but a whole lot less than the risk I accept by using the internet at all. I might get struck by lightning too, but I don't waste a lot of time worrying about it.
> and helping companies compete on ads rather than the quality of their products.
I'm sorry, you do not get to blame me for the continued existence of capitalism. That's just ridiculous.
> Maybe you don't care about your own attention or data security, but your actions don't just affect you.
Once again, you are trying to turn your personal software preference into a moral issue. My actions re:AdBlock affect you or anyone else exactly as much as my choice of desktop wallpaper.
If you choose to participate in harmful systems, you are in fact partially to blame for the harm done by those systems, and simply staying that I don't get to blame you for our doesn't change that.
Incidentally, criticizing ads isn't a criticism of capitalism: capitalism could exist just fine without ads, and in fact I think removing ads would make capitalism drastically more likely to yield the positive results capitalism purportedly yields.
Indeed. Ask yourself the following question: say a user is not interested in advertisements. Is it therefore not a waste of the advertiser's time and/or money to force the ads upon this user?
I think its a combination of badly written code, webgl and maybe even adblocker breaking some otherwise mostly working code.
(Lots of extensions, including Tree Style Tabs, Session Manager, Tab Mix Plus, uBlock Origin, RefControl, Cookie Controller, etc.)
now watch your memory and CPU usage :)
Same setup, but my Firefox leaks memory like a sieve. It will crash 2x a day easily.
That said, I also have LastPass installed, I am using uBlock instead of ABP (though uBlock uses less memory anyway), and I also have RES installed.
I have seen websites leak incredible amounts of memory. To some extent I don't even know how much it matters what browser is used, if the JS is bad enough.
Anyway. I have Firefox with uBlock, on Archlinux. With JS disabled. 400MB, about a hundred tabs. I do recommend using it that way. I also have Chromium, for things that require JS - youtube, facebook and the like.
window.leakyArr = [];
function leakStuff() {
leakyArr.push("This is leaked" + Math.random());
}
setInterval(leakStuff, 100);
Of course if you unload the page this will all be collected. But if you have a page you leave open for a long time and it does stuff like this, it's possible for the page to use hundreds of megabytes of RAM. In fact, twitter does just that if you leave it open for a day or three, for reasons more or less like the above: they're showing or caching all the stuff that came in since you opened the page. That's more and more stuff as time goes on.That's your problem, right there.
I've been told that the main version of LastPass currently available from addons.mozilla.org has known problems that cause the browser to lock up and hang indefinitely.
Instead, if you click through to list all releases and select the latest 4.x version, those problems should all be fixed:
https://addons.mozilla.org/en-US/firefox/addon/lastpass-pass...
I use Chrome for websites that use huge amounts of JavaScript links, which are pain to use with noscript. Chrome does not like a lot of tabs open and chews up memory and performance if forced to run large numbers of tabs.
Any proofs? Stock firefox can't go further than 100 tabs, and you can't do literally anything else on your computer at that time.
about:tabs (see https://addons.mozilla.org/en-US/firefox/addon/tab-stats/ ) is showing me "151 tabs" and "57 tabs have been loaded" for my current session. The other 90-some I haven't looked at this session, so they're left unloaded. I can post a screenshot, but I could fake that just as easily as I could fake my claims above, so it's not any more "proof".
I'm certainly doing other things right now (e.g. compiling).
It _really_ depends on what you have in your 100 tabs. If you have 100 copies of gmail, you're more likely to have a bad time than if you have 100 sane pages. ;)
Ha ha, what proof would satisfy you, where my word does not, O doubting Thomas? If I was going to lie to you on HN, I'd also be able to fake a screenshot. Maybe you'd like administrator access to my machine, so you can satisfy your smug certainty of what is not possible?
I haven't had my FF (stock plus a few plugins) under 300 tabs in months. It doesn't stop me from playing AAA titles, max settings at high resolution, at 30-60 fps. Sometimes while running a VM (because Windows is a gaming OS, not a work OS). I only mention that to disabuse you of the claim that...
> you can't do literally anything else on your computer at that time
I assure you, I can.
Perhaps it's because of what else is running on your system and your system's hardware and OS. I regularly use Firefox with a few hundred tabs open across a couple of windows. On Windows, I find that if I have IE or Edge or Chrome running, each with a few tabs, most of my RAM just goes to them. I also hibernate my system and restart it only once every few weeks.
I use several extensions with Firefox too:
- uBlock Origin (block ads)
- Privacy Badger (tracking blocker)
- Perspectives (distributed certificate checks and better error handling)
- HTTPS Everywhere (default to https on sites and avoid plain http)
- Self-Destructing Cookies (self-destruct cookies for a tab after closing it)
- Tab Mix Plus (better tab management)
- Session Manager (better session management between restarts)
- Tree Style Tab (hierarchical tab list management)
- TooManyTabs (almost self-explanatory)
- Lazarus (save form data to restore or reuse)
- Link Alert (so I know if a link is a file or an internal site link or external site link or a popup and can then decide if I want to click it)
- FoxyProxy (switch proxies easily)
- Foxclocks (handy clocks in different times zones)
- DownThemAll (easier downloads of multiple items, which I need occasionally)
- and many more
Although I experience slow starts and long shutdowns, in general, and in comparison, the browser is not RAM hungry compared to opening just a couple of tabs on Internet Explorer/Edge or about ten tabs or so in Chrome (with fewer extensions than above in Chrome). On OS X/macOS though, Firefox seems more RAM hungry, is a bit more sluggish and consumes more energy than Safari.
Shutting down the browser and opening it again will also help since Firefox by default does not load tabs until they get the focus (this "load tabs on demand" behavior has been around for quite sometime). It will load some stuff from the cache though. One more thing you can try is create a new profile (take care to get bookmarks from the old one) and see if that avoids any problems with extensions or other cruft in the older profile.
After being burned by incompatible bookmark formats years ago (I used to keep everything in bookmarks that were tagged within firefox) I ended up with a habit that whenever I have something that I want to read (or refer to) later I leave it open in a tab.
Only browser that could handle this habit was indeed Firefox, running on Windows 7. How it usually went down was that whenever I hit 1000 tabs I reserved some time to read what was worth reading and close tabs that weren't interesting anymore, until I was down to about 200-500. I didn't use addons much, I remember only having vertical tab list, one of the (ad)blockers and an addon that displayed current tab count below the tab list.
That setup had 8GB of RAM, and I didn't have problems playing games without closing the browser. IIRC it usually reserved around 2-4GB. When I installed Windows 10 on the same machine the browser had 1370 tabs open. I know I still have the sessionstore.js backup somewhere, but I did start from 0 again with the new install.
I'm quite sure that some Mozilla people have noticed my crash reports every now and then :) (though mostly the browser was running fine without problems)
I've since started to use things like Google Keep and Stash to keep the tab count down a bit. Browser that I'm currently on has 229 tabs open, mostly temporary, work related things.
EDIT: Oh, I do have 22 extensions installed.
Chrome came out when Firefox was in a noticeable slump. Remember that in ?2008 being better than IE was the yardstick browsers were measured by and Firefox had passed that many years earlier, so I think they had lost their way a little bit. I remember Firefox 4 was noticeably bad on OSX and it took a long time to fix.
PS I use a browser salad daily (Safari and two channels of Firefox so I can have two instances open at once, and Chrome for Facebook / flash / testing).
$ firefox --no-remote --new-instance --profile /path/to/alt/profileEspecially Google maps and YouTube. I always have to switch over to Chrome to use these sites because they always bring FF to its knees. I wonder if this is deliberate on the part of Google.
It might be deliberate, at least in sense that MS pages never worked properly on Netscape / Firefox / Konqueror /... in the days of IE dominance.
Oh you are like my mother. I do appreciate Firefox respect for standards, but their UI hasn't innovated since 90s. Two months ago it took me 20 minutes on my moms Windows laptop to find current version of Firefox! It lacks unified search bar, tab expose, even some basic hotkeys are backwards.
I can't believe it uses so little resources though. It's not uncommon to see a single tab use 400 MB of memory on Chrome/Safari.
/LARGEADRESSAWARE would give it access to an extra gig but needs to be compiled that way.
About that, I've noticed some weird stuff as well while debugging why our web app was crashing the tab on chrome after extended use without a page reload.
After a hour normal workload, chrome on mac used almost a GB ram, while on windows was at 120mb. Same page, similar workflow etc.
I still have no idea what's going on but there's definitely something about how memory is handled across platforms.
I get a fair bit more resource usage than you're reporting, though...
#!/bin/bash
killall firefox
for file in ~/.mozilla/firefox/*.default/*.sqlite; do
echo sqlite3 "$file" VACUUM
sqlite3 "$file" VACUUM
done
I do it in my once-a-month cleanup script.What happened?
find ~/.mozilla/firefox -name \*.sqlite -print -exec sqlite3 {} VACUUM \;find $HOME/Library/Application Support/Firefox -name \*.sqlite -print -exec sqlite3 {} VACUUM \;
Forked from Firefox 27 and super fast.
Pale Moon 27 is based on Firefox 38 for back-end and Firefox 24 (ish, with some 26-isms) for the front-end. That said, I'd still recommend it. :-)
I'm not sure how the GP has over 200 tabs open with less RAM used than 98 empty tabs though.
Backup/Restore your bookmarks etc. afterwards as it cleans up everything. Resetting everything afterwards took 1h for me but was totally worth it in retrospective.
I wish something even better would come along but I don't really know what that would look like.
But I agree about the UI in general. My habit seems to require some monthly to quarterly clean up to keep it from getting to be too much.
1. I dont have to think. I Cmd+Click links as I find them interesting, switch tabs as my attention changes, etc. and the tabs just stay there, waiting.
2. AwesomeBar searches over open tabs by default, and selecting a suggestion switches to that open tab rather than opening a new one.
3. When I have time, I can sort my tabs into labelled groups ("Work", "Research", "Shopping", "Politics", etc.)
4. Tabs are sync'd to and from my Firefox for Android automatically, so if I'm on the move and I want to pull up that thing on the tip of my tongue, it's right there on my phone as well.
I'm sure it's not perfect for everyone, but it works for me /shrug
Firefox: https://addons.mozilla.org/en-US/firefox/addon/tabist/
and chrome: https://chrome.google.com/webstore/detail/tabist/hdjegjggiog...
It's also open source: https://github.com/fiveNinePlusR/tabist
If you end up trying it, let me know if you find it useful/hate it or have something that should be added :)
If it's easy to add a by-date-last-opened to the order-by at the top, that MIGHT be useful? If it's easy to add the text of the window title to the table, that MIGHT be useful? Those are just top-of-my-head spitball ideas.
One thing that is not 100% clear is that you can press cmd-shift-e to bring up the tabist page. or ctrl-shift-e if you are on windows/linux.
> by-date-last-opened to the order-by at the top
That's planned but not easily attained information from the browser so it'd have to be tracked by the extension from the startup of it.
> If it's easy to add the text of the window title to the table
That's how the extension works but there was a bug[1] that I fixed in firefox that needs to be propagated through the builds in order for it to show up in release. I believe it will land in firefox 51 which comes out early next year... yeah 2017-01-24 is when it will be released[2]. If you want it now you can get on the beta build and it will be fixed.
1. https://bugzilla.mozilla.org/show_bug.cgi?id=1289213
2. https://wiki.mozilla.org/RapidRelease/Calendar
cheers,
hope this helps.
It'd be nice to include the favicon (if available) next to each link so that a visual scan is easy to perform. Otherwise a "Find on page" (Ctrl+F) would have to be used to find something in a list of a few hundred tabs. Sort options like the ones mentioned in another comment here about the date added, and additionally grouping links by domain, would also be useful.
Sorting by the date added/visited is a planed feature but a little tougher to get from firefox so it will take a little bit of work. The extension actually has a group by domain sort(but not strictly group by domain as it's grouped by window first)
Or one can specify what to search with this micro language:
Add ^ to search for matches in your browsing history.
Add * to search for matches in your bookmarks.
Add + to search for matches in pages you've tagged.
Add % to search for matches in your currently open tabs.
Add ~ to search for matches in pages you've typed.
Add # to search for matches in page titles.
Add @ to search for matches in web addresses (URLs).
Add $ to search for matches in suggestions.Another nugget I just found out about: alt-enter opens the what you typed in in a new tab.
If you're visiting a site by entering the address (as opposed to using search or a bang command on DuckDuckGo or using a bookmark/history entry), then you'd also appreciate the following shortcuts:
Ctrl+Enter (Cmd+Enter on Mac) for "www." prefix and ".com" suffix on the domain. Eg: type "google" <Ctrl+Enter> (or Cmd+Enter) to go to www.google.com
Shift+Enter for "www." prefix and ".net" suffix on the domain. Eg: type "jsfiddle" <Shift+Enter> to go to www.jsfiddle.net
Ctrl+Shift+Enter (Cmd+Shift+Enter on Mac) for "www." prefix and ".org" suffix on the domain. Eg: type "wikipedia" <Ctrl+Shift+Enter> (or Cmd+Shift+Enter) to go to www.wikipedia.org
In my knowledge, the www.<domain>.com autofill with Ctrl+Enter was pioneered by Internet Explorer, but Firefox is the only browser that took it to the next level as a built-in feature.
it should be integrated with history (and the back button should not just show a dumb list on right click)
It's a small bookmarklet that when clicked takes the current tabs URL and adds it as an escaped param onto the WebBookmarks URL and launches WebBookmarks. Then in the WebBookmarks form the Title and URL are already filled in, I assign a category, and click save. Job done.
I do this and do not use local machine bookmarks at all, as I use so many different machines (linux, windows, tablet, phone etc) during the course of a week, that keeping all those locally stored bookmarks in sync would be a nightmare.
It works for me, although I've not touched the code in forever so it's looking a bit dated - simply because it's rock solid.
It surprises me that more people don't do this.
For example, a search engine tab might lead to a stackoverflow tab which might lead to a few documentation tabs. As long as you remember what the documentarion is for, you can tell what the SO thread is about, and what is was you searched.
For example, say I'm looking for a sofa to buy. I might open the ones I like in a bunch of tabs to mull over. However, if I'm at work, I'd like to just close the window. Since the project manager has associated the window with a project, I can just close the window. Later, I select "Sofa hunting" from the Projects menu and off I go with the same set of tabs.
I'd go one step further: Think of the project as a "pile of bookmarks" where the visible set of tabs is a subset of all boomarked tabs. For example, say I find a nice sofa. I add it to the pile and close the tab. The project now includes that URL, but it won't open a tab for it unless I go into the project browser and find it there and open it. So this disassociates the tab from the bookmark, but retains the "working set" that is my project session. Similar to how an IDE or editor might preserve which files I have open, but still maintain my entire project.
There are some extensions out there that do similar things, but don't get the ergonomics right. There's a couple of "session managers", but they are dumb: You "load" a session, and then "save" it. Changing the window doesn't automatically update the session, and closing the window destroys your session. Safari also lets you bookmark a bunch of tabs as a folder and reopen them again, but there's no link between the folder and the open window.
I once started on an extension like this for Safari, but the current browser extension APIs aren't great. For example, I'd want to sync the projects with something like iCloud, so that they're available on all devices; but I don't think there's a way to do this without involving a server app. I might try again, though.
Image PID Hard Faults/sec Commit (KB) Working Set (KB) Shareable (KB) Private (KB)
firefox.exe 10604 0 1,529,308 1,491,692 62,348 1,429,344
I'm on Windows 7, 39 days and 5 hours up time.This Firefox instance is running since: 8:05:29 AM 05/12/2016
Firefox 50 (with a pending update).
Also, I see the process as a 32-bits image but I was sure it was 64-bits, will need to check on this.
And I have only: Facebook Ads Block, Ghostery and uBlock Origin.
So being Windows 7 and probably an 32bits Firefox install may cause all this difference?
I currently have 37 tabs open and Firefox is using ~800MB of RAM. Basically, either you're doing something really magical with your Firefox, or are just spewing bullshit.
Someone's going to say its the fault of extensions (what good is Firefox without extensions, anyway) so let's see:
- Decentraleyes (locally emulates CDNs)
- HTTPS Everywhere (why is this not the default behavior?)
- OverbiteFF (support for the gopher protocol)
- Perapera (Mandarin word splitting and Mandarin↔English dictionary)
- Pinboard (I can't imagine this uses any significant resources whatsoever)
- Random Agent Spoofer (change user agent at random)
- Reddit Enhancement Suite
- Tab Groups (I miss when tab panorama was default, but I can see why it's an extension)
- uBlock Origin
- uMatrix (possibly my single favorite Firefox extension; control JS and more, like NoScript on steroids and easier to use)
- VimFx
That's… actually quite a lot. I wonder if extensions alone are my problem, but then again Firefox without extensions isn't much better than Edge (and Edge seems faster on average).
I don't think I've tweaked any settings in about:config.
Is that enough details about my setup for people to help fix it?
Anecdotally complaints about how slow Firefox is seem fairly common. I think your last paragraph is a bit off the mark.
--
1. Both Windows 10 Pro; 16GB/32GB of RAM respectively; both 4GHz i7-4790k; anyway, point is they're pretty good computers.
2. Kidding, but fuck Mozilla
has to intercept every request
> - HTTPS Everywhere (why is this not the default behavior?)
has to intercept every request
> - OverbiteFF (support for the gopher protocol)
might have to intercept every request, and why do you need this
> - Random Agent Spoofer (change user agent at random)
has to intercept every request
> - uBlock Origin
has to intercept every request
> - uMatrix (possibly my single favorite Firefox extension; control JS and more, like NoScript on steroids and easier to use)
has to intercept every request
Stop installing so many stupid extensions. Random UA switching is especially pointless; “Decentraleyes” is a close runner-up. Firefox has performance problems, but I think you’re definitely seeing extension bloat if you have that kind of device.
Those BS plugins do very little for actual privacy (and even less if one is actually a target) and very much for CPU/memory.
FF 1.3GB memory with 10 tabs and 5% of CPU, with FF in the background.
Switching tabs is absolutely not instant, opening new tabs is reasonable, bet definitely not instant.
I'd love your numbers ;)
BarTab Heavy - Allows unloading of unused tabs. Lets tabs act like temporary bookmarks. Tree Style Tab - Just great tab management. uBlock Origin + uMatrix - Lots less cruft loaded and running to slow things down. Session Manager - Lets me save windows with groups of tabs like bookmarks, works amazingly for looking things up and being able to restore a lot of references at once.
When I'm in my working tab group, that's like 8 tabs or so. Most of the tabs in the storage groups are months old (more like bookmarks I didn't have to hit Cmd+D to save)
https://chrome.google.com/webstore/detail/the-great-suspende...
I'm on a 2015 MBP with 16 GB of ram and like 20 extensions installed.
Modesetting is faster too.
Some sites are just ridiculously faster and more-responsive, especially general news sites where all I really want is some text I can immediately read and scroll through.
I have ~300 open tabs in Chrome on Android.
Although they're probably not all loaded, I don't notice any major performance problem.
Literally the only reason I'm on Firefox is tab groups, but my performace is basically identical with tab groups disabled.
Used to run Win8.1 on this machine, Win 10 runs it better.
Used to run Ubuntu on this machine, Arch runs it better.
It would make my day if there was a silver bullet that made FF fast, or at least not crash all the time. The same machine can handle literally 10 times more Chrome tabs without any trace of struggle. But I prefer FF for a few things, so sort of stuck in between.
You could try checking your RAM, I used to have an issue with FF crashing quite frequently, but found that it was part of a larger issue of having corrupt RAM on board.
But it is still better then Chrome performance vise. Chrome is just generally laggy and unresponsive often (no idea why).
If browsers are slow, it's mostly because of some horrible JavaScript running in the background, i.e. poorly optimized websites.
Marvellous. By the time I get to maybe 70-90 open tabs Firefox will consume 4-6 GB RAM and constantly hog a CPU core.
Some pages with JS cause the usual instant entire-browser lock-up.
And I admit I miss the days when Opera 10 was still an up to date browser. I don't think I ever had a (subjectively) faster browser.
no, you have 273 labels displayed in your browser UI, not open tabs. Opera 12.x can hold >100 real loaded and rendered tabs open at 1-2GB ram with ZERO influence on browser responsiveness.
Nowadays every single tab in Chrome/Opera 40.something/Vivaldi progressively slows down UI. ~50 tabs means about 1 second UI delay when going fullscreen on YT. Not to mention the most trivial website takes ~100MB of ram, while Opera 12.x was somehow able to do the same thing in ~10MB per website.
Old browsers didn't do this, so creating tabs was so fast! But they were also much easier to compromise. Tradeoffs.
Source: Electron maintainer
If paging in is the problem then you need more memory (or fewer processes consuming your memory) or a fast SSD so that paging isn't slow. If you don't have a fast SSD then you should try to always have a few GB of RAM available, otherwise your system can't cache frequently needed resources and everything will be slow.
Creating a new tab in Chrome is very fast for me, even on my six-year-old laptop. I can type Ctrl+T and instantly be typing a new URL/search-term.
There are hundreds of things that can make tab creation slow - anti-virus, low-memory, malware, extensions, CPU contention, etc. It is likely that different people on this thread are having their browser experience slowed by different things. Only by monitoring your specific system, or performing experiments (try disabling all extensions) can you find out why browsing is slow for you.
In some cases the slowdowns could be bugs in the browsers which happen to only affect some people, but in most cases I think it is external factors.
chrome://tracing can be used to investigate slowdowns in Chrome. ETW tracing (see UIforETW) can be used to investigate slowdowns on any Windows machine.
I wish there was some system of grouping tabs, e.g. if I visit a newpaper site and Ctrl+Click 3 article links it would keep those tabs "under" the main site tab, and the tab would only display "nytimes.com (4)". As long as I'm viewing one of them, the tabs of that particular domain would expand and other sites would collapse.
Is there a browser or chrome plugin that does this?
Like Tab Groups? FireFox has had them since forever, but since "nobody" used them, it was removed. You now need an extension to regain that functionality:
https://addons.mozilla.org/en-US/firefox/addon/tab-groups-pa...
You also have tree-style tabs, which organizes tabs based on their parent tabs (those they were opened from):
https://addons.mozilla.org/en-US/firefox/addon/tree-style-ta...
You have several options. Unless of course you use Chrome, in which case Google has said "fuck you" to actual power-users and choice ;)
Yes, I'm using chrome and edge, only because FF sucked at the time I switched and even though I have heard good things about FF recently I tend to only switch every 5 years or so. So I guess I'm not really a power web user, as I seem to rather not switch browser than be able to customize it.
... then you could try Vivaldi, which has nice features when it comes to tabs (stacking, tiling, vertical tab list). No tree-style, though.
I have the same feeling, but judging by all the others keeping hundreds of tabs open I feel like I am missing out on a big reason.
Could anyone explain why they keep all these tabs open? None of the sites I frequent do I care about page load times, plus often I would want to refresh for latest news anyways. Most documentation sites are very fast to load. I guess I could see SPA apps that auto update as being something to keep open..
Do you just type into the address bar and switch tabs based on url or tab title thus not caring if you can see the titles or not?
But then the pressing question is, why can't most of the tabs not just be closed? Because I haven't read them yet.
So the real question is, why do I open them in the first place and I'm shy of an answer.
When I bookmark something I add various tags to it (b/c Ill never remember the title) and then can search by tags.
Anywho :)
I've got an Atom-based 2014 laptop at home. That one might take a second to open a new tab if I'm doing anything else on it at the time.
Modern browsers handle a higher load than older ones did; a more complex web means a greater amount of more complex data, and that means more complex data management. The overhead is higher (there's a lower floor), but it allows the browser to scale better (higher ceiling).
I really feel the slowness on that one. Tab switching takes seconds, other interactions take seconds (especially on sites with javascript), Facebook.com is unreadable (script takes too long).
It is just an anecdote, but at least on Linux the browser is what makes older computers obsolete.
PS: Interestingly, some minimalist WebKit-based web browsers seem to work fine on there (like uzbl), of course that comes with a huge downside of no sync, practically no addons, and so forth.
The author was talking about a 2014 iMac, 5 years newer than your netbook and likely 3-4x the price. They're running a modern OS with a browser (Safari) claimed to be 1.4x - 7x faster (by Apple's numbers on their Safari page) than the Firefox browser that I use. So that's what I'm wondering about. By the author's description, their premium 2014 computer isn't running a browser as well as my super-budget refurb 2014 computer.
- ignores usability basics: horrible UI with > 10 tabs, reloads all tabs, many other issues - all of these are things people have complained about forever, they refuse to fix
- memory hog: Chrome uses more memory than almost any other browser
- Google forces Chrome upon you since Hangouts etc will not work with other browsers
- for reasons passing understanding, the Chrome UI has been copied by every browser
Firefox is better at every single thing and has a far better extension model, which sadly they will deprecate again to copy Chrome. I find it very sad that its popularity and market share has fallen as the world has collectively dumbed down the browser UI.
> "Firefox is better at every single thing" Ouch. I mean, your post is littered with unsubstantiated claims, but that has to be the biggest and least substantiated of them. For starters, I think Chrome is a good deal further down the line in terms of ES6/ES7 and WebGl. Chrome is certainly ahead in the implementation of Web Platform features like Web Components and a large number of CSS properties.
I see a lot of websites which advise non-Chrome users to switch to Chrome to view the website in question -- which is horrible, but one could hardly say Firefox is therefore better at WebGl, for example.
I've also used Firefox/Firefox Web Developer Edition/Firefox Nightly many times of the years and I have generally preferred Chrome's dev tools. The ability to simulate slow network connections and to drag and reorder DOM nodes is nice in Chrome, and in general I find the dev tools to be more reliable when I'm messing with CSS properties.
The reality is that I have to use both browsers in my line of work anyway, but just wanted to say that. Obviously your retort will be that Firebug is the best, but that's kind of my point: I do not think you can say of Firefox that it is better at 'every single thing' if it must first be augmented by add-ons. Firebug will be discontinued eventually anyway when multi-process lands in stable Firefox.
So, I wouldn't say Firefox is better at every single thing. Maybe Firefox with a dozen add-ons is better than Chrome at some things, but that's the point. Regular users, and even busy developers/tinkerers, do not want to waste time configuring and setting up their browser. They just want something that will work. Chrome works better out of the box for the regular user.
-----------
Regarding Chrome's UI, I think it caught on for a reason: just like websites have become flat and streamlined, the browser has also become more minimal in its UI, and Chrome very much led the charge there. Of course, Firefox is to be applauded for allowing the user to pick and choose which bits of the UI are visible and which are in the menu -- that's a great feature, I wish I could configure/style Chrome.
But before Chrome, this option did not exist (unless there was an add-on, it's been so long I forget), and the FF user interface had a dozen buttons, two text fields (address & search), a bottom status bar and so on. It was too much clutter. Chrome streamlined the browser UI and put things in a more logical order: tabs, then search/address, then web content. It was and is well-designed from a UI/looks perspective for the regular user.
Clearly, you are a power user and possibly a developer, so you will perhaps be more prone to having hundreds of tabs open and more inclined to spend time making everything about your browser just so. I can appreciate that. I've poured hours into browsers, tiling window managers, etc. I just don't think I'd advocate dwm/xmonad/awesome-wm to a regular user, and neither would I say a window manager represents a superior product to a desktop environment. A product isn't superior if you have to assemble/mod it yourself. This is a cornerstone of Apple's success, and VLC Media Player's success, and Google Search's success, and so on: to provide something that just works, rather than something which needs to be modded.
-----------
No one can deny that Google has far greater reach and clout when pushing its browser down people's throats, but many developers prefer Chrome now, and not just 'rockstars', but older developers too. I do not believe they are 'casuals', I think they genuinely prefer the way Chrome works, and I think its market share is not testament only to Google's marketing power, but also to the fact that it is a successful and effective web browser.
Google Search maintains dominance today in search because of its marketing and its huge lead in terms of data, but it only reached its current level because it ultimately worked better than its competitor products. I think the same is true of Chrome -- when it started out, it represented a fast and sleek new browser product, and while it has gotten overweight as time has progressed, it still beats the competition out of the box.
----------
PS: I am not a Chrome salesman. I have many complaints about the product, (especially on mobile, where I think it is spectacularly weak, but that's another story). I just found your claims a little too rash. Even 'polemic' isn't adequate to describe it.
If that is your impression, I got news for you.
Regular users don't even know tabs exist, and every single tab opened goes unclosed.
My wife had some 200 on her iPad. Every time she opens a tab and can't click "back" she says the internet is broken again.
(And ofcourse she can't find her previous tab in that mess of unmanaged tabs)
Regular users have a lot of tabs too, but for completely different reasons.
This is debatable.
Web components is implemented in Chrome because it was a spec pushed by Google, so they had an implementation being built while things were still being decided. Firefox has an implementation too, just not turned on by default (presumably folks are waiting for the standard to settle?). Similarly, Firefox used ES6 internally for years, so any lag that may have existed (I don't recall any, but ICBW) would probably again be due to standardization. Chrome might be ahead in the implementation of features. You can't call them Web Platform features if they've not been standardized fully.
> and a large number of CSS properties.
Chrome being ahead in a "large number of CSS properties" is objectively false. I keep track.
Firefox has 324[1] properties implemented (I am not counting prefixes as separate properties) when I last counted (a few months ago), Chrome has 304[2]. Firefox has 88[3] properties not in Chrome and Chrome has 68[4] properties not in Firefox. Furthermore, if you look closely most of the Firefox-but-not-Chrome properties are solid green in the Firefox column, which means that they have been implemented in an unprefixed form (which in turn means that the property is probably specced and stable), whereas nearly half of the Chrome-but-not-Firefox properties are light green, which means that they're only implemented in the prefixed form, so the property is either custom or unstable.
I'll admit that these numbers are generated from Firefox Dev vs regular Chrome and Firefox dev has newer features enabled. I made that page to get an idea of Stylo/Servo progress and to help inform what stuff folks should work on next, not for comparing Chrome and Firefox.
Running in Firefox stable to get fairer numbers gets me 301 properties enabled, 69 Firefox-but-not-chrome, and 72 Chrome-but-not-Firefox. This does mean that Chrome has a slight higher number of properties implemented when comparing stable versions. 3 properties is certainly not "a large number of CSS properties", and again, many of the Chrome-only properties are prefixed and will either never be implemented or are unstable and have good reasons for not existing in Firefox.
The numbers aren't perfect. In particular handling implemented-but-disabled properties, properties that are still getting standardized, and properties that will never be standardized is tricky and can't be done in an automated way. But the numbers should be close enough, enough to say that at the very least Chrome and Firefox are at par when it comes to implementing CSS properties.
[1]: https://manishearth.github.io/css-properties-list/?stylo=sho...
[2]: https://manishearth.github.io/css-properties-list/?stylo=sho...
[3]: https://manishearth.github.io/css-properties-list/?stylo=sho...
[4]: https://manishearth.github.io/css-properties-list/?stylo=sho...
What's great about an extension model where every single extension is root in your browser?
Chrome also has the benefit of letting me use my Google account for managing my bookmarks, as well as automatically being logged into my Google services.
There is simply very little compelling reason for me to use Firefox typically as it offers me a subpar experience, and every time I do have to use Firefox's dev tools, I am saddened by their choices of hotkeys/UI/etc.
Why not all three? The browser implementation, user's choices, and the implementations of each individual add-on suck.
The browser didn't limit the resource use of greedy add-ons (or inform the user of conflicts between them), the add-ons were coded either to do too much, or to perform their function with suboptimal code, and the user is the one that chose that set of software to use together.
I have a custom blank page loaded from SSD that has only a small amount of html5/pure js and it takes maybe 1000ms on an intel i5/3xxx. If I open a new tab without any other windows/tabs open it's very fast, maybe 100ms. A new private tab (doesn't load add-ons or my custom new tab) opens quickly but not as quickly as a fresh restart w/ no windows/tabs.
And then it was fixed. No reason to assume it still applies, unless you have new data.
Yes, having a great number of intrusive extensions can cause issues, but there is no need to jump to conclusions as these issues can be present without the user doing anything "wrong".
firefox does not do this.
If it didn't load all the tabs, then you'd see an outdated version of the page when you clicked on a tab, or it'd load when you clicked on it, forcing you to wait right at the moment when you want it immediately.
Reloading all tabs sounds like the best option for people who use tabs to temporarily use several websites at once. That's what I do - never having more than half a dozen and closing them all when I'm finished on the computer. For people who use them in place of bookmarks, it would be better not to load them.
For some it is obviously not, but for others it is a good default.
What's bad is that it cannot be changed. That is not debatable.
This seems so obviously correct to me that I wonder what I'm missing.
You are missing that
> should be marshalled to disk at quit
is not easily doable.
I'd think it's the sort of thing which ought to be child's play.
Sure, you could find a way to serialize it, but that'd be a lot harder — you can't hust go back to halfway the middle of receiving something through a socket.
I really hope that JavaScript in a browser can't launch processes!
Even if you have 300 tabs open, your browser won't try loading everything at once, so you can do what you want.
Otherwise, I press Ctrl+Tab long enough to go through all the tabs and wait for it to load them all.
I don't think that the disk space necessary to cache all this is scary, since after all the whole thing has to fit in memory anyway, and disks are order of magnitude larger than RAM.
* SSD or slow spinning disk (makes a massive difference if your computer is swapping out tabs to cache).
* Have the tabs been loaded or not (a new browser session will open placeholders for the tabs without loading the site, but an older session will likely have dozens -if not more- of those sites loaded)
* Frontend code. Are we talking low footprint sites like HN or dozens of JS and CSS heavy sites like Facebook, Office 365, etc)
* Similarly to the above, media content. Streaming videos? Flash? etc.
I bet you a 100 active tabs of MS Excel Online would perform crap in any set up while 100 inactive HN tabs would stand a better chance of not crippling your whole system.
I can't really blame people for doing that given our hectic modern lives.
If it is a dynamic website, most of them have sessions that expire.
Let me reiterate the first part of that sentence you've quoted: "I doubt many people set out to work that way, it's more a symptom of the "I'll leave that there and come back to it when I get more time" mindset."
> If it is a dynamic website, most of them have sessions that expire.
That depends entirely on how long the session cookies are set for, whether the user browses to other pages on that site since (thus potentially renewing their session cookie) and whether one even needs to have an active session to use that specific dynamic website. Suffice to say there's more exceptions your point than there are examples of it. eg message boards, youtube (and other similar video streaming services), imgur (and similar), etc. And even in instances where the session only manages access to that page, it's not a great chore to log in again.
To be honest I don't really understand why you're trying to lecture HN readers about sensible browsing habits. We're all tech-savvy enough to know how to use a browser. If we chose to adopt a workflow that's less CPU / system memory performant then we do so fully understanding the problems that might arise. The problem is many of us - myself included - grew up in an era when websites were a document rather than an pseudo-desktop application so often our ranting stems from a frustration at how bloated the web has become.
I don't understand what is so upsetting about mentioning a feature designed for that use case.
Then, not all HN readers are tech-savvy, being enthusiastic about technology is not the same as being tech-savvy.
Then, if you don't like having your point refuted, don't engage into a conversation by refuting others' point of view in the first place.
Indeed. Which was the point I literally just made after you argued about sessions timing out.
> Then, not all HN readers are tech-savvy, being enthusiastic about technology is not the same as being tech-savvy.
It sounds very much like you believe you're educating people here. I'd be surprised if anyone reading this thread was unaware of the existence of bookmarks, or confused by the revelation that more tabs mean their "computer thingies" have to do more "thinking". But I can't prove that any more than you can prove most HNers are as technologically inept as you repeatedly suggest.
> Then, if you don't like having your point refuted, don't engage into a conversation by refuting others' point of view in the first place.
You're not refuting my point though because my point was never that tabs are better than bookmarks. I just commented on why people often end up working that way despite knowing better. It was an insight not a recommendation.
I made a reasonable suggestion without an intention to offend people. Hacker News is about sharing information. I am sharing information.
You want to start your Monty Python argument clinic in the Hacker News comment section? I am not interested. Take your unreasonable accusations elsewhere.
1stly I said "tech-savvy enough to run a browser". The context here is critical (hence my emphasis) because I'm not talking about our level of technical competence but rather the basics of using a browser (ie using tabs and bookmarks). You may disagree and think some on here are unaware of bookmarks but I think that's extremely unlikely. However I think taking proper context of my point into account, you likely don't disagree me. Which is why your rebuttals always crop out the context I included.
2ndly I was never offended by your suggestion to use bookmarks. In fact I didn't even disagree with you. I wrote about why some people don't use bookmarks and how I don't blame them for their arguably bad practices. It seemed that you misinterpreted that as an argument in favour of tabs and thus constructed arguments around disproving that point. That wasn't necessary as there wasn't ever a disagreement there. Maybe I should have been more clear early on?
Lastly I've not made any accusations about you. Let alone unreasonable ones. I may have disagreed on a few specific points but I think the "accusations" term is a tad exaggerative. If you feel I have said things that could be taken personally then I apologise, that was never my intention. :)
Couple the above points with your reiterating my counterargument about login sessions and hopefully you can see why I was a little frustrated at the weird direction this discussion took. At every turn you seemed hell bent on disagreeing with me even when we were agreeing. If you don't mind me asking, maybe your initial comment getting downvoted (wasn't me by the way!) left you feeling you needed to prove your point? I've never agreed with negative rep but that's a whole other tangent.
Anyway, since we never actually disagreed on the crux of the argument I think we should just chalk this down to a misunderstanding and get on with our lives :)
page resource total url
(b) (b) (Kb)
----------------------------------------------
33857 143 34.0 news.ycombinator.com
62690 55975 118.7 arstechnica.com
11844 223862 235.7 seldomlogical.com
218837 765943 984.8 newyorktimes.com
l592771 545304 1138.1 guardian.co.uk
The one thing I notice watching the index page being broken down, is the sheer number of resources requested. Didn't bother with JS/browser interaction which you can check with browser dev tools.Agree with that. A driver of the page size (JS, images, video) is advertising. Advertising and advertising techniques are distorting the news and ability to read it.
implemented in py using
- Beautiful Soup 4
- Requests
- htmllib
Core idea is to grab an index url and determine size of page and resources by looking at urls likely to be pulled by page. JS not implemented from what I can read. Method is something like:
* pull index page with requests by url
* extract headers
* build list of resource urls
* look at content type to determine new request
* extract images, text, css from urls in page
* look at content length to determine size
* total sizesRegardless of performance, with that many tabs open I just find browsers unpleasant to use because you can't see the title of each tab and switching between them is a mess. OneTab is fantastic - I have it as a pinned tab on all my browser windows and I send it all the tabs I'm not actively using. For me it also doubles as a short-term bookmarking system so I don't clog up my pinboard.in bookmarks with articles I have not yet read etc.
https://chrome.google.com/webstore/detail/the-great-suspende...
https://chrome.google.com/webstore/detail/the-great-suspende...
It frees up lots of memory but still lets me keep lots of tabs around without having to be militant and killing them off as I stop using them.
It's marked as experimental, but I haven't had any trouble so far.
The fact that you can do this https://m.youtube.com/watch?v=BV32Cs_CMqo in a browser should tell you we've come a long way from them just being Document readers.
Firefox also loads suspended tabs from cache. It reloads them(on first switch to tab), but no data is downloaded from server.
Try Ctrl+C,Ctrl+T,Ctrl+v,Enter combo. In opera it worked flawlessly everytime. In Firefox, it sometimes works, sometime it opens new tab without URL, sometime it pastes half of URL (?!) and navigatest there.
In chrome you never know what sugesstion will the address bar give to you. In opera, you were 100% sure that your bookmarks will be there, so you could do Ctrl+T, type first few letters and press Enter without looking.
If you have 100 tabs open of which 10 are rather badly-behaved, maybe the browser cannot do enough to mitigate it. But again, although this is a common power-user setup, it's not standard.
If none of the above, it's likely due to an extension. I was starting to blame Safari for slow tab opening and page loading recently and then I uninstalled an extension and everything became lightning fast again. I have around 100 tabs open across 4 windows. All major browsers are very fast for normal browsing in the default configuration.
Site downloads almost 500KB of fonts.
I, like the OP, don't understand why that part is hard. Does it have to rerun and redownload all the javascript each time you switch to a tab? Are we misunderstanding how websites work? Will removing all JS solve the behavior the author mentions? Obviously doing that will make pages load faster. But that's not what they seem to be asking. Is the OP question not meaningful because of some aspect of page rendering?
Honest questions, I have basically no knowledge about this topic, but have wondered the same thing.
[0]: https://addons.mozilla.org/en-US/firefox/addon/automatic-rea...
Filesystems are a great "database" for large quantities of information, as long as it doesn't change, and you don't need to do any queries on it. I think there is quite a lot of range between "reasonably small site" and "news site" for which static pages work well. You could even have a successful news site with static pages, as long as you didn't want things like comments and trending articles, because while there are lots of articles being created, news articles don't usually change.
Changing layout and themes, however, could be no fun with static files.
I'd prefer to generate all the HTML files for a site every update, just like Jekyll or Hugo does.
He's not (just) talking about the slowness of websites, but of browsers themselves.
And it supports Chrome extensions, to boot.
I found that quite amusing..
Opera 1x, managed to be a email client, RSS reader and Browser while keeping the size below 12MB. Chrome/Blink or Firefox all takes 40-50MB Download.
I really wish they could Open Sources the old Opera.
P.S- I am not sure if these Tab Problem is nerd only. Basically people are trying to gob too much information. But i have yet to see a Sane UI to fix this.
I never really understand comments like these, what are you saying? You want smaller binaries? Less features?
Don't you have a spare 50MB on your 512GB SSD?
I currently have nearly 400 tabs open, across 7 browser windows (Firefox).
The main reason for this is context switching. I have tried several context switching type addons but not found one yet that is reliable, simple and fast. Right now if I am programming, I have 2 browser windows I use for that. For news and related articles of interest, 1 browser window. For games and related, 1 window, etc, etc.
Firefox and Tabmix plus makes this easy and relatively painless. Chrome would have about 12 million processes and use 4.7 petabytes of memory for same use case (I exaggerate a smidgen)
Back in the day when I was writing on the forums "guys, forget Firefox and use Opera, it's fast it's usable you have everything you might want in a browser with undeniable proof", the reply would eventually be "yes but Firefox has _this_ webdev extension and I can't do without it". So Firefox would get installed in their families' and friends' PCs, because that's what the web developer used. Also, the "safer faster better" marketing campaign, although that slogan would have been much more appropriate to Opera than Firefox. Fast forward a few years. Google Chrome comes out and everyone stops caring about those damn extensions; suddenly performance becomes a huge decision factor (nevermind Chrome was a resource hog compared to Opera, nevermind the horrible lack of features, nevermind the bad usability). I could go on, but you get the drift.
There's no incentive for large market share browsers to significantly improve because the (mediocrity+marketing=success) recipe is proven to work, over and over. What would it take for Chrome to lose market share? A new hot hip popular company with lots of money thrown in advertising, like (extremely unlikely example) Tesla, with a new cute interface. Because people like new and shiny trinkets.
For what they are, I consider modern web browsers to be exceptionally fast and reliable, and they tend to put a lot of other applications to absolute shame. They are imperfect, but still very, very good.
799 open tabs. 8.5GB. Restarted in around 10 seconds and it's currently using around 2GB. "Delay loading of background tabs" is checked, which I believe is the default now.
Not had any serious responsiveness problems. The browser extension I use to give me a list of tabs on my sidebar is a little slow when opening a new window, that's about it.
Is "blank" actually blank? If you have new tabs set to show Top Sites (the default setting), there's a notable lag that doesn't occur if you set it to Empty Page. Of course, changing it is a reduction in functionality, and it's silly that loading a simple grid of screenshots is so slow. Chrome, for its part, is even slower (on my machine) to actually load the new tab page, but it loads it asynchronously with the address bar responding instantaneously, whereas Safari lags out the whole UI.
I still use Safari myself, with new tabs set to Empty Page, because I don't really care about the screenshot-grid functionality (in either browser), and Safari has better performance in other respects, as well as lower power consumption. But it irks me that the default macOS experience makes you wait to open a new tab; it's a terrible UX for something that shouldn't be hard to optimize.
"I've got 450 tabs open, in any browser, ever, and I experience 0 lag - in fact, it tends to finish rendering before my eyes even focus! And I've only got 6 gigs on some old 386!"
vs
"I can't even move my mouse in FF/Chrome/Safari without massive memory spikes and random reboots!"
At least some of the reports have to be expectations - I've sat next to someone and they thought I was "moving so fast" in my browser, and I was - at the same time - cursing how slow everything was running.
Someone reporting "no lag" might actually be experiencing what someone else reports as lag. Is there any way to adjust for our vague descriptions? Videos, perhaps, of what someone considers "fast" or "no lag" on a browser with 100s of tabs open?
Before the "Not me! I use w3m on a p3!" replies pile in, accept this as a generalisation warning..
I think most of us in this thread though generally use reasonably recent gear (our machines are the tools of our trade, personally I upgrade each year) and that we use our browsers extensively enough that even small issues around perf become quite noticeable...
I can't remember the last time I couldn't scroll smoothly or a tab didn't open instantly...
Maybe the difference here is those in the church of noscript and those without? After the initial pain white listing the various things you need, I've found the performance of Firefox on Linux to be brilliant. (currently around 100 tabs, some of which are sat on prime video etc, arch Linux, 16G, 50.1.0).
A lot of the extensions listed in this thread are doing quite a lot of work also, and sometimes on every request which probably doesn't help..
Disk caching could also be a problem for those of us without shiny new NVMe's in their laptops (pull the trigger on that one, it's worth it!).
The only downside of all the blocking stuff for me though is when you book a flight or something and the noscript/ublock combo prevents me from getting past the callback from your payment provider... I need some kind of giant banner that says turn off noscript and ublock when using britishairways installed somewhere :}
Or I'll just send them yet another snarky feedback message the next time I fuck it up (seriously, it calls t.co and doubleclick in serial!)
We have SSD, CPU, Memory and Network 100 to 10000x faster then what had 20 years ago, and yet Apps didn't get 100x faster. It is still not instantaneous.
In someways i think it is because we have created far too much abstraction /layers.
I've discontinued use of optical media circa 2009 and replaced my OS drives with SSDs in 2010.
Eliminating spin-up times and random seek times for multiple small files has probably been the biggest performance boost in my user-side computing experience.
I'm no expert in the development of browsers; but I'd hazard a guess that the problem isn't as straightforward as it seems… and performance will suffer as a result, especially since all this has to occur at runtime.
But When it comes to clicking on links and opening a new page, Browsers aren't slow. The software that runs on them is slow: all those applications loading too many libraries, widgets, videos, video ads, Angular and lots of code that's not needed.
Too many modules in this "Bot Machine".
Certainly that's my problem - I don't even try. Enough people with that attitude and browsers never get optimised...
Pressing 'back' used to be quick, on the modern web its rather sluggish, like it re-loads the old page from scratch.
I can forgive sites that are implemented as single page with their own 'fake' URLs (though that is still annoying). But even static content.
Given the emphasis on browsers handling lots of tabs I'm surprised 'back' isn't implemented by, effectively, keeping the old page around for a bit and pausing its Javascript, then just picking up where you left off.
. compute resources, the cpu cycles allowed in the next epoch are halved
. RAM it's memory accesses per second are halved in the next epoch
. network traffic it's kB/s are halved in the next epoch
This would allow all pages to still work but the resource intensive ones would be flagged quickly by users and developers.
Being generous, I think the average user might actually push this btn once a month?
Then again, as is Chrome.
build and use sites that template html, without js / huge styling, like HN or pinboard, and I promise you everything will be lightning fast
Nowadays we have monopolies forcing onto us mediocre things developed by mediocre teams, like Chromium/Blink. And they are big and so deep in the government, that no one seem to be able to do anything about it.
No one is forced into anything. It's just that the browser went from simply rendering HTML and some associated eye-candy JS to being an OS.
So JS, which used to be a relatively safe, interpreted language, turned into Java, a complicated JIT VM where every millisecond counts. In the process, security holes are introduced, and competing with the "Big Guys" is just too hard.
It's like why no one can compete against Intel for power and speed. It's just too expensive.
By the way, if you are referring to speed rather than features, you need to take into account that when you try Opera 12.17 right now, you are trying a browser that has been frozen in time for years while the JavaScript on websites evolved. For this reason, it's now very slow (if not incompatible) in JavaScript-heavy websites like Facebook, Gmail, etc... but I can attest as a long-time user that it was indeed the fastest browser in its time.
That said, there are improvements for that on the nightlies.
Is there a particular site that causes this to happen when you have it loaded? If you're willing to spend some time troubleshooting or profiling this, please let me know and I'll walk you through some ways to do that...
If you're using a nightly, you should be able to use https://developer.mozilla.org/en-US/docs/Mozilla/Performance... to profile it, because nightlies have useful symbols. Might be worth capturing a profile at the point when the spinner stops and the page is finally shown.
If this starts happening again and you get any more data on it, please email me: bzbarsky at mit dot edu.