Why are current browsers so slow?



  • They are fast in benchmarks but when I try to use them they are slower than a few years old Opera. Despite being castrated from all the functions. And after adding add-ons they are even slower (and still have less functionality than the old Opera),



  • Let's test that shall we?

    For some reason, Peacekeeper is unavailable right now, but I have a bookmark for the old Sunspider benchmark so I'll use that.

    Here is Vivaldi's results on this machine, and here is Opera 12.16 on the same machine. Hint - copy the URL and paste it into the form on the other link to compare results. Remember though, Opera 12 is not multi-threaded and therefore is at an automatic disadvantage on such tests.



  • Browsers aren't slow. Browser speed was never really the issue.

    Browsers are slow because they lack usability.

    For example, two seconds saved using a mouse gesture or a rocker gesture is worth much more in terms of browsing speed than a few milliseconds in loading time.



  • @sgunhouse:

    Let's test that shall we?

    For some reason, Peacekeeper

    I believe he is talking about the general sanppyness not about the rendering engine.

    And yes nowadays browsers are are slow nowadays antivirus, operating systems and so on.

    With just few exceptions like win 7 which is faster than vista.



  • One tricky thing about comparing real-world web sites is that many of them adjust their output depending on the browser they detect. Fancy bloaty feeping creatures are often skipped on older browsers that aren't believed to support them. It can be interesting to experiment with one of those user agent spoofing extensions, sometimes you'll discover that horrendously inefficient sites have perfectly reasonable interfaces hidden away.



  • @sgunhouse:

    Here is Vivaldi's results on this machine, and here is Opera 12.16 on the same machine. Hint - copy the URL and paste it into the form on the other link to compare results. Remember though, Opera 12 is not multi-threaded and therefore is at an automatic disadvantage on such tests.

    So just to make sure I'm understanding your results, Opera 12.16 tested faster overall (significantly slower on nbody access and cordic math, significantly faster on 14 of the measures, and not statistically different on the remaining measures). Am I reading that correctly?



  • No, you are reading it backwards. Opera 12.16 was 1.78x slower, but did beat Vivaldi on a few tests. But note, Sunspider is a javascript benchmark - it factors out loading and rendering times.



  • @sgunhouse:

    No, you are reading it backwards. Opera 12.16 was 1.78x slower, but did beat Vivaldi on a few tests.

    Hmm…. That's what I figured your point would have been (and why I asked). But either I'm still not understanding the results (shorter times = faster) or I'm thinking your URLs are backwards:

    Vivaldi's results = 1552.4ms total and Opera 12.16 = 891.0ms total, making Opera 12.16 1.74x as fast.

    (It would be nice if the SunSpider page did some browser sniffing and included the User Agent or something to identify the browser.)

    BTW, I imagine you saw this Vivaldi HTML5 benchmark thread. Any comments?



  • New test then. Latest 64-bit snapshot on Win 7 (previous test was in Linux on a different machine), and for comparison latest 64-bit Opera Developer on same system, You'll notice this computer is faster than the previous tests. And here's Opera 12.17 (64 bit) on the same machine.

    Strangely, Opera 12 and Vivaldi were almost the same this time with Opera Developer bringing up the rear.



  • Such modern benchmarks nowadays are almost pointless, as they are telling nothing about the user experience.

    They don't tell about the startup time, they don't tell about the time needed to open a new tab, they don't tell about the perceived responsiveness, and they tell even less about the experience on a lowend/old/ low memory machine.

    Altough not always accurate there is something that tells a lot about the above, and is the download size of the installer.

    The smaller is the installer the faster is the program/operating system/browser/whatever.

    Such rudimentary "rule" has surely some exceptions, but they are still exceptions.

    Vivaldi will never be as fast as Opera12, and will never be as fast as Chromium 5, period



  • @sgunhouse:

    New test then. Latest 64-bit snapshot on Win 7 (previous test was in Linux on a different machine), and for comparison latest 64-bit Opera Developer on same system, You'll notice this computer is faster than the previous tests. And here's Opera 12.17 (64 bit) on the same machine.

    Your 64-bit Opera Developer link isn't working for me (no data displayed), but that is interesting that Vivaldi and Opera 12.17 are roughly equal.

    For comparison, here is 32-bit Vivaldi 1.0.151.7 on my 32-bit Win7 machine, Opera 12.14 in Sandboxie on the same machine, and Opera 12.17 on the same machine.

    Roughly the same for all three.


  • Moderator

    Giving a little more answer to the OP: Because they were not made to be fast.

    Opera and Presto were built with low resource usage in mind since the beginning. WebKit and later Blink are just copies of KHTML, they took the work of Konqueror team and just started to toss stuff they wanted in one of their services that no other browser had so they could limit the use for their browsers. Chrome was built to help them track us so they can show even more ads, that's Google focus for years, they don't care about their products any more because they know that everybody will use it anyway.

    There's one thing that for me makes the browser be faster and it's the way it renders the page. Let's compare the main engines:

    • KHTML & Sons: The page is only rendered after everything has been 100% downloaded. Only when the HTML, CSS & JS are loaded you'll see the page.
    • Trident: The page will start being rendered when the HTML is complete, then it will be as resources are loaded. CSS looks to have priority.
    • Gecko: The page will render right away as the resources are loaded. But Gecko doesn't prioritise CSS, keeping the page without style for some seconds.
    • Presto: The page will render right away as resources are loaded, but CSS have high priority. Style is applied right away.

    There's also the cache problem. Opera had a very strong cache, if the resource is in the cache it's loaded from it, Chrome on the other hand has a horrible cache, barely using it.



  • @An_dz:

    There's one thing that for me makes the browser be faster and it's the way it renders the page.

    I still think that the rendering speed is not the main concern of OP.

    We are not anymore in the times when Opera left everything else in the dust, especially IE4/5/6, to render a page.

    Nowadays, more or less, everything is fast enough. But most modern browsers (Vivaldi included) are two, three, four times bigger than Opera.

    And that is noticeable in the launch time and in most UI functions, epecially on PCs that aren't the latest novelty, equipped with slow mechanical HDD, relatively slow CPUs and relatively slow RAM.

    Vivaldi is not a monster but its installation file is bigger than the whole Win95 setup folder, and that means something… ;)



  • It means Vivaldi handles video better than 95 did? :P

    No, really - what was your color depth in 95? How large were your desktop icons? I remember *.ico files being 16x16 with 16 colors. Just the higher resolution and color depth makes all the little images take up 10 times the disk space or more.



  • @sgunhouse:

    It means Vivaldi handles video better than 95 did? :P

    Not yet :P



  • Today we have mighty pc, speedy internet and slow browsers, nonsens.
    Hm, maybe it's problem only for us old O12 users.



  • @The_Solutor:

    Vivaldi is not a monster but its installation file is bigger than the whole Win95 setup folder, and that means something… ;)

    Chromium was meant as an operating system, not as a browser, running apps from the Chrome web store. A modern "browser" can even emulate a complete x86 PC running Linux:

    http://www.bellard.org/jslinux/ (runs fine in Opera 12 too BTW and boots 3 times faster than in Vivaldi)

    This would have been a nice April's fool article in a PC magazine back in 1995. Now it's reality.



  • @sgunhouse:

    No, you are reading it backwards. Opera 12.16 was 1.78x slower, but did beat Vivaldi on a few tests. But note, Sunspider is a javascript benchmark - it factors out loading and rendering times.

    I think you're the one reading it backwards. Numbers in that are time measured in milliseconds, and almost every single category that was measured has Vivaldi taking longer to do the same tasks. When it's all added up, Vivaldi took 1552 milliseconds to accomplish the same things that took Opera 890 milliseconds, making Vivaldi 43% slower than Opera 12.


Log in to reply
 

Looks like your connection to Vivaldi Forum was lost, please wait while we try to reconnect.