Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
Chrome Safari

Developer Claims Chrome Uses 10x More RAM Than Safari (macrumors.com) 133

MacRumors writes: Under normal and lightweight web browsing, Google Chrome uses 10x more RAM than Safari on macOS Big Sur, according to a test conducted by Flotato creator Morten Just (via iMore).

In a blog post, Morten Just outlines that he put both browsers to the test in two scenarios on the latest version of macOS. The first test was conducted on a virtual machine, and the second on a 2019 16-inch MacBook Pro with 32GB of RAM. In the first round of testing, Just simulated a typical browsing pattern of opening Twitter, scrolling around, and then opening a new tab with Gmail and composing an email.

Under that test, Just found that Chrome reached 1GB of RAM usage, while Safari used only 80MB of RAM.

The two-tab test was only the start, however. With 54 tabs open, Just found that Google Chrome used 24x more RAM per tab compared to Safari. Both browsers, according to Just, were free of any extensions, and this specific test was conducted on his actual MacBook Pro, not a virtual machine. Per his findings, Chrome used 290MB of RAM per open tab, while Safari only used 12MB of RAM per open tab.

This discussion has been archived. No new comments can be posted.

Developer Claims Chrome Uses 10x More RAM Than Safari

Comments Filter:
  • Chrome probably also uses 10x more network bandwidth in the phone home to mommy direction.
    • Im not denying that Google isnt tracking people, but to say traffic is increased by a factor of 10 is silly.
      • Yeah, more like 100.

        • What possible information could Google be stealing (lets use that word) after each page load ?
          • Since opening a simple web page these days apparently requires downloading "content" from around fifty different websites, google then needs to know what those fifty different websites were. Fifty downloads plus fifty notifications back to the mother ship = 100.

          • "stealing (lets use that word)"

            Now before we go and do that, please remember that there are HUGE numbers of people on Slashdot who will contend to their dying breath that copying is not stealing.

            • And others who will claim it's perfectly all right to steal another's distribution.
            • > that there are HUGE numbers of people on Slashdot who will contend to their dying breath that copying is not stealing

              Not sure if stupid or if trolling. /s

              How do you "steal" a number?

              You CAN'T. You don't own a number -- you have (public) access to one like everybody else. You can't stop people from using them or from sharing them. They don't need your permission because you didn't invent them and you can do fuck all to change that.

              What you completely failing to understand is that you are trying to appl

              • I don't troll, generally speaking. I post either in sincere terms, or in humour. But in this case, I was definitely poking some fun at the underbelly of the forum. i spotted an absurdity I could wink at.

                And wham... out you came. Were I troll I would consider this a "good get". :)

              • I am curious. How did you infer all that intent on my part from my one line post? I'm curious at which point in that sentence I "lambasted" anybody. I offered no opinion on the topic at all.

                As far as research goes, I've been observing and participating in these discussions all the way back to dial-in bbs forums on 300 baud connections, Were I to express an opinion, it would be an informed one.

                Yours,
                The complete and Utter Fool.

              • I never claimed it was stealing, i said lets just call it stealing.
            • by tchdab1 ( 164848 )

              Corporate copying, which must be assumed will be used for profit, from vast numbers of uninformed users without consent IS stealing.
              Personal copying for non-profit non-destructive use is not.

          • Every single thing you do. How long you hover your mouse over something. How many times you scroll, how long you wait... every single thing you do, each moment you're on the page.

            That's what Viacom asked me to build into their music property sites. Maybe they're less invasive than Google.

          • by ceoyoyo ( 59147 )

            After each page load? Which page you loaded, of course.

            In addition to that, mouse movements.

      • On the "in the phone home to mommy" connection? Why is that silly?

        (ie. discounting the page content, measuring only the tracking traffic)

    • Anybody has 'ntop' installed to see how the bandwidth stats compare?
  • Invalid test (Score:5, Informative)

    by Guspaz ( 556486 ) on Sunday February 21, 2021 @04:38AM (#61085540)

    Reading the update and the comments, the guy didn't measure Safari's memory usage correctly, as he didn't include Safari's renderer memory usage. Chrome probably still uses way more, but it makes Safari look abnormally good by comparison.

    • Yep. Safari might be tied into the OS in a way that the resources don't show up on his RAM usage monitor.

      I'm not saying Chrome doesn't use incomprehensible of RAM (it does!) but I doubt that Safari RAM usage is that low.

    • It always tricky to measure memory usage, and some of the Chrome memory usage is probably counted multiple times because separate processes uses the same shared memory.

      Still there is no denying Chrome is bloated, but getting exact numbers is hard.

      • Also Chrome doesn't start clearing memory buffers until it has used 50% of system memory :D

        • by ceoyoyo ( 59147 )

          Seems reasonable. After all, why would you want to run anything other than Chrome?

          • I'm not defending Chrome's well known resource hogging, but if nothing else needs to use the memory then freeing it up doesn't do anything.
            • At least on Linux, the kernel uses much of the unused memory for file caches and I/O buffers. When applications allocate more memory, the amount of caching and buffering goes down. That slows the whole system down, and machines that are low on memory (even if not yet swapping) feel significantly more sluggish, especially when switching between programs.

      • by tepples ( 727027 )

        The way I typically measure memory usage for process groups that share a lot of memory is running free before and after completely closing the application. This procedure yields 90 MB for, say, Skype's main window and 450 MB for its taskbar icon.

    • Reading the update and the comments, the guy didn't measure Safari's memory usage correctly, as he didn't include Safari's renderer memory usage. Chrome probably still uses way more, but it makes Safari look abnormally good by comparison.

      Please forgive my general ignorance on this topic but I wanted to mention a detail about Safari that might put a wrinkle into the memory-usage discussion. If you pin a tab in Safari it becomes an instance. Whenever you open new windows the pinned tabs are there and they're not clones of those windows, they're instances of them. This means you could probably have 100 windows open with pinned tabs with only the memory usage of just the source tabs.

      Assuming I'm not misunderstanding how this works (I almost

    • Re:Invalid test (Score:5, Insightful)

      by AmiMoJo ( 196126 ) on Sunday February 21, 2021 @06:43AM (#61085740) Homepage Journal

      Indeed, all modern browsers render pages as large tile bitmaps that can then scroll nice and smoothly, and on a decent resolution display that alone is going to push past 80MB. And that's before allocating any memory to decode images and the like.

      In fact many web pages are more than 80MB alone. That number is completely unbelievable.

      The other point to make is that using RAM isn't bad, it's good. If your system has 32GB of RAM and most of it is unused then you might as well cache a few gigs for your browser. Free memory is wasted memory, it's sitting there using power and doing nothing when it could be cache. As long as Chrome reduces memory usage when there is pressure on free RAM (it does) there is no problem.

      • In fact many web pages are more than 80MB alone.

        80MB of HTML?

        • by AmiMoJo ( 196126 )

          HTML, Javascript, CSS, images (that must be decoded for rendering), cookies etc. The add overhead for all the network connections needed to fetch them, the Javascript JIT compiler, DOM, layout engine...

          Turns out what he was measuring was just the core per-tab stuff, the basic sandbox and resources to render that part of the UI. Not the actual tab process that fetches and renders the page, and then handles all the interaction. So Safari uses 80MB of an empty tab.

          • Unless you load videos, most pages won't fetch 80 MB of resources. So most pages alone won't be 80 MB in size.
            • Yeah, 80mb is exaggerated. For example this comment page on /. uses only 1mb of resources from only 80 different URLs. It's typical right? Most websites don't have any images or videos at all, right? Just text. Chrome fully renders it in 1.3 seconds. "Render" means it draws it on a canvas. How much memory does that canvas require? Do you want it to not buffer the canvas so that it has to do a complete re-render whenever you scroll?

              Of course I have an adblocker and noscript. Let's try in a fresh

          • Yes, it's incredible how bloated the net has become lately, and that's not even taking into account all the datamining some JS code does after the page is loaded (see how Facebook monitors all your mouse movements all the time). If one doesn't have an adblocker and a javascript blocker, browsing becomes an extremely painful experience if you don't have the latest 5G or optic fibre connection.

            • This comment page takes 18.5 seconds to render if I turn off uBlock. That's not because of bandwidth (I have 150megabit, and the render time is about the same when most of it is coming from cache). It's because javascript loads javascript loads javascript loads javascript loads javascript loads ads. I'm simplifying. The daisy-chain is several hundred URLs long. It's mostly because a 2.4Ghz i5-2430 is ancient and nowhere near Slashdot's minimum specs.
        • HTML comprises a miniscule fraction of the bandwidth used by a browser.
      • by Entrope ( 68843 )

        The current Apple Silicon Macs never have 32GB of RAM. They default to 8GB, with an option to upgrade to 16GB. If you "cache a few gigs for your browser", that's a much more considerable fraction of the total

        • by AmiMoJo ( 196126 )

          But if you don't allow the browser to use a few gigs when there is nothing else needing it then your browser is always slow. You are just wasting RAM you paid for.

          • by Entrope ( 68843 )

            Maybe you have a job that only requires a web browser. (If so, you are probably either easily automated or easily offshored.). My job requires doing lots of non-web activities: updating documents, managing requirements, writing code -- which for some reason now takes the combination of a web browser and a web server, and sometimes gobbles tens of gigabytes of RAM for code completion purposes -- and running analyses, some of which take tens of gigabytes on their own.

            I bought RAM for my purposes, not so some

            • by AmiMoJo ( 196126 )

              My job is similar. On my old laptop sometimes I close Chrome to save RAM, but it's no better with Firefox. That thing only has 4gb.

              On desktop with 32gb I never have an issue.

      • I have 64GB of ram and I have seen Chrome not reduce memory usage when under extreme pressure. I was debugging something in Visual Studio that needed about 50GB of memory just to hold the matrixes in memory and Chrome did not reduce usage. Instead, the system started swapping to the SSD. I had to kill Chrome to get the system to stop swapping.

        The current version of Edge behaves far better with respect to memory usage and CPU. Especially with the automatically sleeping tabs.

      • by nagora ( 177841 )

        Free memory is wasted memory, it's sitting there using power and doing nothing when it could be cache. As long as Chrome reduces memory usage when there is pressure on free RAM (it does) there is no problem.

        Yes, so long as it does. I highly doubt that it does.

        Free memory is available memory, which the OS can use for caching. Having it locked up in a specific application's allocation is inefficient.

    • by Ecuador ( 740021 ) on Sunday February 21, 2021 @06:55AM (#61085760) Homepage

      He is quite clueless. He saw Safari's memory usage not increasing at all when opening new tabs and still did not question his methodology. All because it fit his narrative about how Safari is lighter, because he sells a software that makes websites into apps, so it is whatever memory usage Safari has - minus a bit as from his own admission he just loads a safari based webview with the website as an app, so all the extra safari functionality is not included. Oh, right, what he highlights as the smart part of his app is that it loads the mobile version of the site. Genius! /facepalm

  • Chrome reserves a lot of memory for pre-rendering and other stuff, if it can. In a lower memory environment, Chrome takes less.
    • by Cley Faye ( 1123605 ) on Sunday February 21, 2021 @09:20AM (#61085990) Homepage
      That's the important part. Chrome uses what's available. On my "beefy" system, it takes a lot, but most of it is cache. It doesn't prevent other applications from taking up memory. On a work computer with very little RAM (4GB), it took way less and leave the machine perfectly usable for other tasks beside browsing.
      This race to "I use less memory on a benchmark" serves no purpose aside from calling one piece of software "bloated" when in the end it does nothing wrong.
    • The problem is that Chrome assumes it is the only application on the whole platform and tries to use ALL the memory. Which is really bad for people doing more than one thing or leaving a browser open in background with a lot of tabs to read later.

  • Seriously? (Score:4, Interesting)

    by fahrbot-bot ( 874524 ) on Sunday February 21, 2021 @05:01AM (#61085598)

    With 54 tabs open, ...

    And with 640 tabs open (which should be enough for anyone) ...

    Maybe I'm an outlier, but I've *never* had more than 10 tabs open at once -- and I shutdown the browser (Firefox) when I'm not actively browsing.

    • When I do day trading, I may have 20 tabs open for different stock graphs, so there are legitimate use cases for many tabs.
      • When I do day trading

        day trading? you're not the common user.

        • "day trading? you're not the common user." - No, especially since I actually make money day trading, which seems to be very unusual.
          • Oh? Prove it. And I don’t want to hear about your best week, or your best month, or last years results. Or even the past 3 years, where the market has risen for literally anyone who hasn’t been actively trying to shoot themselves in the head. How’s the last two decades been? I’ve met a single day trader who beat the market long-term, and it was a VERY old guy with immense self discipline who bought-and-held much more the usual retail investor. I’d be interested to met a second.
    • Reflecting on my own usage at work, as soon as I have to code something, do research, or look for documentation, I quickly end up with 20 to 30 tabs open at the same time.

    • by xonen ( 774419 )

      It may be laziness, but at the end of the day a frequently find myself with 2 or 3 browser windows each containing a dozen tabs, sometimes even a few dozen.
      Root cause: i mid-click links to open in a new tab, stuff that's relevant, then skim or read that information, and never bother to close it. I close all, and sometimes even restore entire session just because there was something that's still relevant the next day.
      There's no real need to do so, but also no real drawback. It just minimizes the amount of cl

    • by randjh ( 7163909 )
      Right now I've got 2 browser windows, each with 7 tabs, trying to restore a Kobo my wife buggered. Another window for /. and other techie news sites.. Another couple windows dealing with repair of my cellphone that I was in the middle of before starting to repair that Kobo. All told, I've got 10 Firefox windows open, averaging over 6 windows each. And I'm 80 years old. Am I just smarter than all you young whipppersnappers? :)
    • I use a lot of media sites with "front pages". If the new york times has five stories that pique my interest. They get opened up in tabs. Sometimes each article links to one or two more related articles-- so those get opened up in tabs.

      If I'm searching in the library for a book, the catalog may return 50 different items-- those that deserve closer inspection get opened up in tabs.

      Same thing with auction sites, shopping sites, etc.

      When you have plenty of memory and the browser is well behaved, it's a perfec

    • pretty much at all times. In fact, right now, this one I'm writing in has nearly 80 tabs. I have another one with 30+, and a third one with about 60. They're roughly segmented by use.

      My machine has 32GB, and after a few days I'm forced to restart FF's processes, not because it's out of memory but because it gets slower and slower to use, and with recent updates, noticeably worse at re-rendering already open tabs.

      This seems to be from extensive ad blocking and I've seen things like a reddit tab spiking to

    • Maybe I'm an outlier, but I've *never* had more than 10 tabs open at once -- and I shutdown the browser (Firefox) when I'm not actively browsing.

      Not sure what shutting down or opening up a browser has to do with the number of tabs you may need open at any given time. It sounds like you read Slashdot by clicking on links and clicking back like a chum rather than opening every story in a tab the way god intended. :-)

      • It sounds like you read Slashdot by clicking on links and clicking back like a chum rather than opening every story in a tab the way god intended. :-)

        Nope, use separate tabs, not a chum. :-)

  • by DrXym ( 126579 ) on Sunday February 21, 2021 @05:10AM (#61085626)
    If your computer has lots of memory, a modern browser will take it to use as cache, buffers, surfaces, textures, history, compiled js, state, js heap etc. Holding it in memory means stuff like loading & rendering is faster. If memory gets low (aka "memory pressure") the browser will free up this stuff so it can be used by other processes. Apart from that browsers might be using shared / mapped memory which could make consumption appear larger than it actually is if you're reading the process info wrong.
    • by AmiMoJo ( 196126 )

      His blog also notes that Safari uses more energy that Chrome, draining your battery faster.

      It seems that he measured it wrong anyway, the command he used to check Safari's memory consumption did not include child processes. Appears to be a bug in the tool.

    • Well, yes, that is the general idea. "Unused RAM is wasted RAM." But I recently replaced my old iMac (24 GB) with a new Mac Mini M1 (8 GB). All the benchmarks implied that architectural improvements and os optimizations had made the 16 GB configuration into a wasteful luxury

      And given that I needed a new machine fast, I specced it with a larger SSD, but not the larger RAM config. That turned out to be a mistake, as my large memory habits (and a leaky implementaation of Safari) made my mac mini experience "re

      • Well, yes, that is the general idea. "Unused RAM is wasted RAM."

        Unused ram is ready to go, and thus requires neither the cooperation of a greedy process nor the imaginary memory an OS might simulate.

        This stuff makes Chrome 1% faster and, at times, other stuff 50% slower.

  • by thegarbz ( 1787294 ) on Sunday February 21, 2021 @05:27AM (#61085648)

    Low RAM usage isn't some magical end goal. You want low RAM usage, you can do that quite easily by stripping out the web standards browsers support, security functionality like sandboxing, or features people come to expect like password management.

    And as for the "developer", sorry but there isn't a functional web browser on the market that is able to render in a standards compliant way using only 80MB of RAM. When your numbers look so outrageous instead of publishing your amazing mind blowing result that no doubt Apple themselves would have been shouting from the rooftops, maybe sit down and look at where you went wrong in your measurement.

    • I agree, one wants to always be using RAM where it helps a program's performance. The real problem is when every single program thinks it will be the only program on the system, so each one takes all the RAM it can use. On my work computer, Teams seems to have this attitude, crowding out my primary programs where I would prefer that the memory be used (I don't really care that much about the performance of a chat program beyond a certain point).

      I strongly suspect that there is a lot of wasted memory in most

  • All browsers are resource hogs. Stop with the silly browser wars.
  • You have to admit: They did a really good job at sneaking in their new version of compuserve + operating system with chrome and v8.
    Well done. Chromebooks are gaining, fast. Because these days thanks to Google a browser is all the regular person needs in a computer.

  • but the content being loaded in the browser, not all websites are created equal, some are light and some are heavy
  • Google Chrome has a long history of memory abuse and behind the scenes abuse. I have never liked it and will never install it. Research it if you don't believe me.
  • by martynhare ( 7125343 ) on Sunday February 21, 2021 @09:32AM (#61086020)
    ...because (s)he doesn't even know what processes to measure RAM usage against, (s)he didn't include all the processes involved, which is why the RAM use looked so low. In reality, the Safari favours casual users and Chrome favours power users in a default configuration, as demonstrated by the Brave team in their comparisons (see https://brave.com/brave-one-do... [brave.com] for more info)

    However, even Brave's comparison doesn't reflect reality. Chrome has three possible split process models and the most paranoid is used by default to help mitigate Spectre/Meltdown vulnerabilities. Running with --process-per-site --disable-site-isolation-trials causes far less RAM to be used overall (see https://www.chromium.org/devel... [chromium.org] for more information). WebKit2 as used by Safari is far less paranoid by default and does not do isolation per site instance. While I'm not going to go out of my way to run benchmarks, it's plain to see from Brave's results that Safari will lose to Chrome the moment users tweak the process model to be less paranoid.

    Also, if you don't care for security, there's always the option to go for a single process model and let multithreading do all the work, like the good old days. Operating systems like Windows, which are light on threads but heavy on processes would benefit from this the most.
  • I have 16GB on my laptop. Let's have it sit idle while I am doing nothing but browsing the web at the time.

    Any time I scroll, let the CPU burn trying to render new content. When I click back, read the HTML again from disk, re-run all that wasteful JavaScript; and not save its previous state in RAM.

    Windows XP used optimize for 256MB RAM at that time. That meant, if I task switched between two programs, it would swap the old one entirely to disk, regardless of how much RAM the machine had. Shall we go back to

  • Rumors that matter.

  • and used a method called p-hacking [wikipedia.org] that uses computers to find meaningful patterns in large amounts of data.

    FTFY

  • by going on a diet. Right?

To do nothing is to be nothing.

Working...