Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Chrome Chromium Google Microsoft

Google Will Disable Microsoft's RAM-Saving Feature for Chrome in Windows 10 (zdnet.com) 140

Posted by EditorDavid from the game-of-memory dept.
"Google has decided to disable a feature in Windows 10 version 2004 that allowed Chrome and Microsoft Edge browsers to use a lot less RAM," reports ZDNet: Windows 10 gave Win32 apps including Chrome access to a 'segment heap' API to allow apps to reduce memory usage, but as Techdows spotted, Chromium engineers have decided for now to turn off the feature by default in Chrome 85 after discovering it has a negative impact on CPU usage. Chrome 85 should reach stable status in August...

The CPU issue was discovered by an Intel engineer who found that when Chrome used segment heap, it led to significant performance regression in benchmarks on a PC with an Intel Core i9-9900K processor...

Microsoft has defended the trade-off between memory and CPU but conceded it can be implemented better to reduce the impact on CPU performance. "It is common practice to trade one resource for another. More often it's increased memory usage for reduced CPU usage. In this case it's increased CPU usage for dramatically reduced memory usage, or more accurately commit," wrote a Microsoft employee... "In the short term this is a good trade-off of one resource for another as memory/commit usage is a significant pain point for browser users," argued the Microsoft employee....

However, Chromium developers want to see more evidence about the possible impact of Chrome using segment heap... "The CPU cost (10% slowdown on Speedometer 2.0, 13% increase in CPU/power consumption) is too great for us to keep."
This discussion has been archived. No new comments can be posted.

Google Will Disable Microsoft's RAM-Saving Feature for Chrome in Windows 10 More

Google Will Disable Microsoft's RAM-Saving Feature for Chrome in Windows 10

Comments Filter:

  • 16 GB of RAM to open the browser on Windows 10 (Score:3, Insightful)

    by lessSockMorePuppet ( 6778792 ) on Monday July 20, 2020 @03:54AM (#60309681) Homepage

    And they're worried about CPU time.

    I'll take a 10% slowdown on CPU over using all my RAM and instantly degrading to an EMACS situation (8 megabytes and constantly swapping).

    Hint: drive is slower than RAM. If you blow through all the RAM, you're left with drive or network speed. CPU sits around twiddling its clock while waiting for meaningful work.

    • Re:16 GB of RAM to open the browser on Windows 10 (Score:4, Informative)

      by thegarbz ( 1787294 ) on Monday July 20, 2020 @04:02AM (#60309693)

      No it does not take 16GB of RAM to open a browser on Windows 10. It takes a couple of hundred MB to run the system, a few more for process isolation, and in general you should see 1-2GB of RAM at the most on a browser with many tabs open and plugins running.

      RAM is cheap, welcome to 2020. What isn't cheap however is a CPU that refuses to enter a battery saving C state, especially given browsers are likely to be the most active piece of software on any given laptop.

      Please use all my RAM, that's why I bought it.

      • Re:16 GB of RAM to open the browser on Windows 10 (Score:5, Interesting)

        by GigaplexNZ ( 1233886 ) on Monday July 20, 2020 @04:16AM (#60309713)
        2 days ago Chrome was chewing up 28GB of my RAM. 16GB of that was for a single tab - Gmail. Chrome really does have a memory problem. Though it does seem that it's more of a memory leak problem as the longer I leave it open the more RAM it uses.

        • Re: (Score:2)

          by bn-7bc ( 909819 )
          Chrome has a problem, or does the gmail js running in the tap eating 16GB have a problem, how much dose the same tab, open for the same amount if time doing exactly the same task (thus reading/replying yo the same mails (or at least similar enough as to be of the same size), consume when used in another browser? No, I'm not defending Chrome here that is googles job, and I don't seriously expect you, or anyone else to spend the time on doing that test, my point was just to point out that the fault might no
          • When both Chrome and Gmail are both from the same vendor, I really don't care which one is the root cause. Using Gmail from within Chrome is their primary use case.

            • I've never used either Chrome nor GMail. I find this whole shebang rather entertaining. I would say that using Gmail within Chrome is nobodies primary use case.

        • Re: (Score:2, Insightful)

          by thegarbz ( 1787294 )

          2 days ago Chrome was chewing up 28GB of my RAM. 16GB of that was for a single tab - Gmail. Chrome really does have a memory problem. Though it does seem that it's more of a memory leak problem as the longer I leave it open the more RAM it uses.

          Congratulations. You had what's known as a memory leak. This is widely known as a software bug or flaw and has nothing at all to do with the feature being provided by Microsoft.

          • You do realise I said it was a memory leak, right?

            • Indeed you did and I congratulated you for it and even quoted you. Though I'm still confused as to why you think this is as all relevant to the discussion at hand, which is the GP's claim that his browser consumes a lot of RAM and this Microsoft feature would magically fix this.

        • Re: (Score:2)

          by AmiMoJo ( 196126 )

          Something is badly broken on your system. Maybe a really dodgy add-on? Malware perhaps?

          I have Gmail open. Using 230MB of RAM for that tab. Entire system with several other apps open is only using about 5GB of RAM.

          • No add-ons and no malware. I just don't close the browser very often. ~200MB per tab sounds about right when it's freshly opened.

            • Re: (Score:2)

              by AmiMoJo ( 196126 )

              Mine has been open for about a month, in fact it's prompting me to restart it for an update now.

            • > No add-ons and no malware.

              There is *something* going on that apparently you don't know about. If you double-check for extensions and truly don't have any installed, then update Chrome and still have the same problem - well then it starts looking pretty likely that you don't know about the malware. There is something going on there. A Malwarebytes scan would probably be a good idea.

              • No extensions. I tend to update Chrome roughly monthly (schedule is defined by the network admin where I work). There's no malware either, or at least nothing found by either Malwarebytes or the corporate AV scanner.

        • I've used Chrome for years and never saw it use more than a few GB at most. The exception being those people with hundreds of tabs open.

        • Re: (Score:2)

          by Luthair ( 847766 )
          Maybe you should look at your extensions.

        • There is a unique opportunity here to take the Pepsi challenge. Try the new Chromium based Edge browser and see if it's different.

          I have ludicrous amounts of RAM on this box so I wouldn't notice either way. As of this second Edge Chromium has 1.1GB and that's with 2 windows, and ~12 tabs. It's been open since I rebooted on Patch Tuesday.

          (Full disclosure: I work for Microsoft. I'm not shilling, but I understand if you choose to believe I am.)

      • Re: (Score:2)

        by fazig ( 2909523 )
        The high performance RAM so the Infinity Fabric of my Ryzen runs at 1900MHz without abysmal timings wasn't that cheap, as I also wanted a bit of capacity because some of the applications I use can easily gobble up 20GB of RAM during normal operation.

        But then again I neither had RAM nor performance issues with Chrome. Perhaps because I keep my number of open tabs low.
        I don't really get why people need hundreds of tabs open at the same time. You can't look at them at once! So why not use bookmarks instead?

        • The high performance RAM

          Let me stop you right there. You bought high end stuff and traded off price vs size vs performance. RAM is cheap for the 99.99% of people do not get 1900MHz RAM. Heck many higher end gaming rigs are perfectly fine running 1500MHz RAM.

          I assume with RAM like that you're also rocking a 3950X or similar high end processor?

          • Re: (Score:2)

            by fazig ( 2909523 )
            3900X, the 3950X wasn't released at that time. But even if it was, I would probably still have gone for the 3900X.

            For the Ryzen 3000 series most of the larger hardware reviewer have recommend and still recommend to use 1600MHz (3200MT CL14) or 1800MHz (3600MT CL16) to get the performance sweet spot for various applications. Beyond that point capacity becomes more beneficial than higher bandwidth and lower latencies in general (of course before that as well for any application that needs more than 16GB).
            Th

            • Indeed I have 1600MHz RAM. I also spec'd 1600MHz RAM (high latency) for a budget build for my father (not sure why I said 1500, dividing by two is hard for me). But point is, performance sweet spots are necessary when building a performance system. If you're reading hardware review sites you are speccing well above average compared to the common PC.

              The vast majority of systems out there don't have 1600MHz RAM just as the vast majority of the systems out there aren't 3900X or even 3700X systems.

              Don't forget,

              • Re: (Score:2)

                by fazig ( 2909523 )
                Yes, prices can drop sharply if looser timings are acceptable. Then you're also less reliant on the memory being Samsung B-dies thus having a lot more options to choose from. Although most people probably don't even look up that information

                For example F4-3200C14D-32GVK (14-14-14-34, ought to be Samsung B-die from the timings alone and confirmed by a lot SPD readouts from CPU-Z and Taiphoon) kit still costs just a bit less than $200. It is probably binned well enough to overclock to 1900MH with timings li

      • RAM is cheap, welcome to 2020.

        What ISN'T cheap is a new laptop or motherboard with free slots to put the cheap RAM into. Or even the knowledge of how to do so.

        Yeah, yeah, we should have all bought more expensive computers... but doesn't that detract from the "RAM is cheap!" message?

        All this so that browser developers don't have to optimize anything.

        • All this so that browser developers don't have to optimize anything.

          False. Browsers are actually optimised pretty well. The question is do you have a suitable computer to run a modern browser meeting modern internet standards, effectively acting as its own mini OS, sandboxing individual tabs for your safety, running typically multiple plugins, across many tabs each with their own processes, own JIT compilers for scripted languages, each serving up dynamic content and offering features we expect in 2020.

          The internet changed. It's not 1995 anymore. The high RAM usage is the n

          • All this so that browser developers don't have to optimize anything.

            False. Browsers are actually optimised pretty well.

            Here's me thinking we were discussing an article that's talking about a tradeoff between RAM vs CPU usage that's relevant to millions of people but is being dictated by an Intel Engineer ... on an Intel Core i9-9900K processor.

            Decisions like this are "optimization", not loop unrolling. This decision is boneheaded because disk thrashing causes performance problems, too.

            Older hardware should not prevent technological progress.

            Basically you're saying that Grandma's Internet experience should be dictated by a few rich kid's gaming PCs.

            • Nope. I'm saying Grandma's Internet experience should be dictated by the most common laptops on the market. "Rich kid's gaming PCs" being the only ones with more than a couple of GB of RAM just shows you're completely detached from reality.

              And if you were a good grandson maybe you should install an outdated low-resource browser for your grandma to use her hotmail account rather than holding up technological progress for the rest of us.

              • OK, boomer.

                • I don't think you understand what Ok Boomer means, or what a Boomer even is considering it's precisely Boomers who stand against technological progression.

                  Boomers know this, Millennials know this too, but your UID is too low to be a Gen Z child so why do you speak like one?

                  • Getting back to the point: RAM may be cheap but upgrading exising PCs isn't so that's not the answer.

                    Neither is throwing away the computers of everybody over the age of 30.

          • One can argue that given the features of modern browsers, the per-tab memory usage is kinda reasonable.

            One can also argue that several GB for your browser is a problem.

            The issue I see I most cases, including my own, is that the UI encourages people to have 50 different sites open simultaneously. We leave a page open and running because we might want to use it next week. That creates a problem.

            The solution is a quick and easy way to open the site a week from now if we end up wanting it, without leaving 50 s

          • Here's a whopper of an idea: put in a fucking checkbox and let the user decide, instead of dictating to us what is better for our system when you cannot possibly account for every system combination out there.

            I have 32GB of RAM in my laptop, so I don't mind using a bit more memory to keep available CPU for other tasks, especially since I run three displays on it, so I always have other tasks running. However, if you don't have 32GB of RAM in your laptop, you may want to conserve memory at the expense of a

            • Don't be a retard! Do you know how difficult it is to make something configurable when you could just force everyone to use only one particular method that you intend to flip back and forth anyway? Thing of the poor Google employee's. If you provide a configuration option, then the option can simply be toggled to and fro and once in place and tested you are done.

              However, by making it non-configurable you now have to pay all these Google workers to go though and make changes to the code, then pay someone

            • As a developer I hate user configuration. It's one more thing to ask from them when they report a bug. It is another row in the test matrix. And another thing to document, yet almost nobody will bother reading that document before they monkey with a setting.

              On the other hand, waiting 15 years for browsers to reintroduce themes as "dark mode" is a bit of a joke. We could have configured that ourselves if themes worked right and didn't constantly get re-architected.

            • Here's a whopper of an idea: put in a fucking checkbox and let the user decide, instead of dictating to us what is better for our system when you cannot possibly account for every system combination out there.

              No, that's impossible.

              I was told by a vendor this week that a similar change was out-of-spec, and would require at least a month to design and implement, if not two or three. You see, the old code and the new code can't exist together, and it would be very hard to find a way to toggle behavior based on user input. There's just no way to let a user choose between auto-update and manual update, and the auto-update code is already written and it's replaced the manual update code so they'd have to re-write that

      • Re: (Score:2)

        by xonen ( 774419 )

        If you'd told me around 2005 that browsers in 2020 would use 2GB just to display some web pages i'd laughed my ash off and mumbled something like 'no, it can't be that bad'.

        On-topic: not everyone has much RAM. Not everyone has the resources to buy a brand new PC or laptop. Plenty people that are working on boxes with only 2-4GB of RAM, yet having a multi-core CPU that's per core not that much slower than a modern CPU.

        Google would be wise to take that into account and making it optional. Much RAM? Don't both

      • Re: (Score:2)

        by Hydrian ( 183536 )
        RAM is only cheap if you can upgrade it. Some new machine have the RAM soldered to the motherboard. This is becoming more and more common in consumer-level laptop. Since the majority of non-gamers/non-professionals, buy laptops... RAM is NOT cheap.

      • So you're cool with running out of RAM and paging to disk. Click, click, click, click... Enjoy that. I am trying to upgrade my
        8 GB of RAM to as much as 24 GB because I can't even run an old game and browse with Chrome for very long before very bad things start to happen. Game CTDs. Browser freezes. Have to reboot. This happens almost immediately if I let windows do memory management. If disable windows memory management things worth smoothly until they don't and it takes longer for bad things to hap

    • Re: (Score:2)

      by DThorne ( 21879 )
      RAM is there to be used. A lot. If you spent money on a computer you want to squeeze every possible usage out of the hardware. I'm constantly baffled by people obsessively tracking every time an app goes beyond what they consider to be "necessary" to get the job done - I want every app, everywhere, to utilize all of my hardware absolutely to the max *in a resource friendly fashion*. Back in the days of XP, yup, memory gobblers were a serious problem but since Win 7 Windows isn't really all that bad for

      • RAM is there to be used. A lot.

        One problem with this philosophy is that greedy programs do exist - that is, programs that will take and take without actual need. If a program needs it and uses it, fine, but to just say that it's there to be used therefore programs can be used I fear might just lead towards apathy that will lead to the problems that have been worked past / we wanted to avoid in the first place.
        Responsible memory usage IMO isn't a sin... IMO, wanting to utilize hardware is fine, but the problem with how you define it is

      • The same could be said for CPU time. Are you unhappy if your CPU is not pegged at 100% all the time?

    • If my CPU hits 100% I can still open task manager and kill the asshole process, if my RAM hits 100% it's like trying to wrestle a chewtoy from a very angry rottweiler.

    • Comment removed based on user account deletion

    • Well I am looking at my Task manager and I see only less than 600 megs dedicated to Chrome. While 600 megs is a lot, It it is 4% of that 16GB that you are using. Perhaps you have some spyware on your PC that you need to clean up.

      CPU processing makes a lot of heat, heat turns on the fan, and degrades your battery life much faster than what RAM uses.

      RAM is rather Cheap, and so is Solid State Storage. While CPU is expensive, and generates a lot of heat, and uses a lot of power.

    • On Kubuntu 18.04, my Firefox memory usage after starting, going to Slashdot, and reading a few articles is 474,408 KB, totaled from all Firefox processes. Chrome on the same system after a fresh start and doing nothing but existing is 482,452 KB, totaled from all processes. So they both take nearly half a gig.

    • CPU Use directly correlates with battery usage.

      I would rather my browsing use less power and more memory.

    • I wouldn't jump to a conclusion even though I like Edge. The question I would ask is "Which uses more battery?" Does 10% longer CPU burst use more power than 100MB more memory usage?

      My desktops already have plenty of RAM so I don't care. Even my laptops generally have plenty of RAM so I would be curious about which impacts battery life the most which is a metric I care about in a browser.

  • Comment removed (Score:3)

    by account_deleted ( 4530225 ) on Monday July 20, 2020 @03:55AM (#60309683)
    Comment removed based on user account deletion

    • If you have crappy CPUs, it would be slow, but still usable.
      If you do not have enough RAM, it would not run at all.
      So, the priority should be RAM usage first, CPU usage second and then disk I/O.

      We should not be designing for potatoes in a way that negatively impacts performance of current equipment. If you can give a user a switch to active this do, and leave it disabled by default. If you can't make it a switch then maintain a "Lite" edition of your software.

      Someone else being cheap should not negatively impact my system especially since battery life is a new king of the performance requirements in the modern world.

      • Re: (Score:2)

        by AmiMoJo ( 196126 )

        They use the same code base for the Android version of Chrome. I have an older laptop with 4GB RAM where it runs fine too. It scales impressively well really.

        • They use the same code base for the Android version of Chrome.

          And Chalk and Cheese both contain carbon atoms. That doesn't make them even remotely similar.

      • Re:For low-end PCs... (Score:5, Insightful)

        by wierd_w ( 1375923 ) on Monday July 20, 2020 @05:08AM (#60309801)

        The issue is that it is not just chrome's software people holding this view.

        this is classic tragedy of the commons writ large.

        "RAM is cheap! It is PLENTIFUL! Why SHOULDNT my program consume every single byte of it that it can, if it makes my (otherwise totally unoptimal and shitty, especially since it is trying to do things it really should not, because other kinds of software do the job better and more efficiently) code run faster!!

        Leads to:

        OS Developers: RAM is cheap!
        Browser developers: RAM is cheap!
        Application developers: RAM is cheap!
        Web based developers: RAM is cheap!

        By the time you get through all the layers of "I refuse to accept that this is not an infinitely available resource, and want to staunchly hold to the maxim that my software can have *ALL* the ram in the system whenever it fucking wants, and that is the way I will write my software, fuck you people and your assertions otherwise!", you end up with even high end systems running like potatoes.

        Kindly sir-- Divorce yourself of such notions.

        • Yes and no. RAM being plentiful achieves two things:
          - Faster loading when you need something.
          - The ability to add more features.

          Both of which I find desirable outcomes as an end user. Dear developers: Use my RAM, please. Every bit of data that doesn't need to be read from my SSD at the time it is needed is of benefit to me!

          • I think you might be forgetting some important things here.

            A CPU (even a multi core CPU) can only shuttle data in and out of that RAM so fast. When you have multiple stacks of software, all demanding huge troves of RAM (because they cant be bothered to better decide what data to keep and what data to discard in their caches, because "A single cache miss causes the benchmark to look shit!!"), and all being able to do tasks in true parallel fashion (because multiple logical CPUs that are only tied to each ot

            • I'm not forgetting anything. The point is the user wants functionality. That can be delivered in 2 ways, pre-load required data into RAM or load it as needed from a disk. The demand itself is set by the functionality required.

              Having the CPU access it in RAM is faster leading to less weight time and even more so reduced power consumption allowing the CPU to get back to its idle state faster.

              You keep claiming that this is all due to some programmer cutting corners and using outdated libraries or something lik

          • Comment removed based on user account deletion

        • That horse left the barn long ago. Software is already slow and bloated. How much faster is your desktop today than 20 years ago? Orders of magnitude. Launch a large application like Photoshop or Solidworks. It still loads as slowly today as it did two decades ago.

    • Yes. Especially for software thats supposed to run everywhere.

      This isnt a 3D modeling program where its reasonable to require gigabytes of memory at your customers expense. This is a web browser that should be able to run on a potato.

    • Re: (Score:2)

      by Luthair ( 847766 )
      If your battery is empty your PC won't work at all. Further, there is such a thing as virtual memory.

      • You realize that if you care about performance at all, virtual memory should be avoided if at all possible, right?

        I'd rather have the CPU work harder and stay off the disk than save a bit of CPU and have the whole damn system bog down on IO wait, or even worse, have the Windows analog for oomkiller step in and do it's work.

        Application software should not act like gas in a vacuum expanding to fill the available space. It should learn to share.

      • Further, there is such a thing as virtual memory.

        Which uses your hard drive if I recall correctly - accessing disk is a lot slower than accessing RAM though.

  • WAIT A MINUTE!!! (Score:3)

    by tgeek ( 941867 ) on Monday July 20, 2020 @04:10AM (#60309705)
    NOW Google is concerned about Chrome performance???? When did this start???

    • Re:WAIT A MINUTE!!! (Score:5, Funny)

      by Luckyo ( 1726890 ) on Monday July 20, 2020 @04:19AM (#60309717)

      Probably when someone at Google noticed that their HEDT machine had a couple of gigabytes of free RAM while running Chrome. Can you imagine the PTSD that person must have gotten from that? "Wait, I can actually run things other than Chrome while Chrome is running? What is this sorcery?!"

    • Re: (Score:3)

      by AmiMoJo ( 196126 )

      Chrome has always been about performance. They made this video about how fast Chrome is 10 years ago: https://youtu.be/nCgQDjiotG0 [youtu.be]

      Back then every other browser was pretty slow and it made web apps second class. Now web Outlook is actually more responsive than desktop Outlook. Chromebooks are viable for many people because web apps are all they need and perform as well as local apps. Google nearly made operating systems irrelevant.

      • I remember what Opera 12.x landed. They finally patched out all the delay in the UI, which meant things like tabbing, typing in the bar, or using any in browser features where instant.
        Chrome still has some rather severe UI delay. Hitting new tab freezes the browser until it renders, where the delay is high enough that it can eat several inputs if you assume it accepts inputs. Bookmarks and history has the same issue, if you use those features. It won't help to increase Javascript performance if the UI lags,

        • Re: (Score:2)

          by AmiMoJo ( 196126 )

          Chrome doesn't freeze when I open a new tab. Maybe it's an extension you have installed, I did find one that caused freezes (Privacy Pass).

        • Chrome still has some rather severe UI delay.

          I think it's time you reinstall Chrome. Honestly I'd need a process profiler to even attempt to measure "delay" between hitting new tab and it appearing, and even then I'm not sure if that delay is simply because I don't have a high refresh gaming monitor. Your setup has a problem.

          It happens. I experienced the same with Firefox, the difference is that after 2 versions came and went with the same problem I recognised that it was probably my machine and reinstalled the damn browser clearing the user data in t

          • Chrome still has some rather severe UI delay.

            I think it's time you reinstall Chrome. Honestly I'd need a process profiler to even attempt to measure "delay" between hitting new tab and it appearing, and even then I'm not sure if that delay is simply because I don't have a high refresh gaming monitor. Your setup has a problem.

            I'm often baffled by complaints about Chrome's performance or memory consumption. One common thread I notice, though, is that the complainers are always running Windows. This may be an example of the base rate fallacy -- because I nearly all computer users run Windows (sadly, even on /.) -- but I wonder if maybe the reason I never see these problems is because I don't use Windows.

            Is there anyone who sees serious Chrome performance or memory consumption issues on Linux or OS X? (Note that I'm asking only a

            • One common thread I notice, though, is that the complainers are always running Windows.

              I run Windows, I also run Linux. Schedule problems, performance problems, software bugs, and configuration degradation (especially as a result of constantly changing things like syncing user profiles or uninstalling and installing plugins) are somethings that are truly universal. The benefit of Linux is they usually put the offending files in /tmp /lib /opt or in your home directory, whereas with Windows if you're lucky they are somewhere buried in your profile director and if you're unlucky buried 40 fold

    • NOW Google is concerned about Chrome performance???? When did this start???

      In the first beta of Chrome, back when Google demonstrated a browser that could execute Javascript and render a webpage faster then competition. Don't confuse the adoption of a a large featureset and resulting high RAM usage with performance. Chrome has always been the fastest browser out of all the browsers that had the same abilities and feature set.

  • Options exist. Don't expect me to hold your hand, people.
  • that "Google has decided to disable a feature in Windows 10 version 2004". Since when does Google control the code of Windows?

    • that "Google has decided to disable a feature in Windows 10 version 2004". Since when does Google control the code of Windows?

      If I rewrote that to read

      ""Google has decided to disable a feature that Windows 10 version 2004 takes advantage of to ..."

      would it make more sense to you?

      Because that's really the situation here.

  • They rarely measure memory usage. As long as this doesn't change, it shouldn't surprise anyone that browser vendors are not willing to trade even slightly higher CPU usage for a significantly reduced memory footprint.

  • ... to store bitfields (arrays of boolean values).

    Preferably boxed and wrapped in variant data types at every layer.

    So ...

    • Re: (Score:3)

      by v1 ( 525388 )

      When you consider the size of even 1MB of ran it becomes impossible to see a uint64 bitfield as anything more than a drop in the oean, unless you're trying to store a movie one bit at a time.

      But while that sounds absurd, that's basically the basis for "trading memory for speed". The problem is that like with the movie example, programmers are making horrible tradeoffs while searching for faster ways to do things. A 1% increase in performance isn't worth doubling your memory footprint. And if that 1% you

      • Sometimes I miss the days when memory was a serious limitation, and code had to be lean and efficient.

        If you want to go back to those days, you can load $OLD_OS in a VM. Then when the browser running on the host OS lets you have a few MB, you can get some work done.

        • Re: (Score:2)

          by v1 ( 525388 )

          yep... just think of the hundreds of thousands of lines of code that have to run every time you press a key on your keyboard, before MS Word can put the letter on the screen.

          Then look back at something like AppleWorks, running on a 1mhz 6502 processor, as it runs rings around the speed of Word, all while using less than 128 kb of RAM.

          Efficient programming is becoming a lost art. So many lazy programmers.

          (and hilarity ensues when they encounter something like an Arduino, wait what? 16k of ram? but but but

          • People here always seem to take potshots at Word's performance, but have you actually run it recently? Or are you just repeating Slashdot lore?

            Word is a native application written in C/C++, so unlike many "modern" apps, it's actually quite fast and svelte on modern systems. On my dev machine at home (ten years old and running Word 2010), Word starts in about half a second, and takes 15MB of RAM for an empty document. At work I used more modern machines and up-to-date versions of Word, and it seemed just

    • These are the coders who use uint64 linked lists to store bitfields (arrays of boolean values).

      Are you serious? People actually go that far for bitfields?

    • Preferably boxed and wrapped in variant data types at every layer.

      No, you have to box it because it's const.

  • IBM mastered VM storage and paging 50 years ago. One can even configure different paging and heap strategies for at the task level. I dont understand why a heap API is needed at all - the compiler should have heap size options and storage pool numbers,. Crikes, I wonder if MS engineers even understand about VM and page stealing and LFU taught in 2nd year. Now all these Intel CPU speculative flaw workarounds are probably more understandable. The downside to operating system code that simply just works, is th

  • 10% slowdown in browser benchmarks? (Score:5, Insightful)

    by Ecuador ( 740021 ) on Monday July 20, 2020 @07:29AM (#60309997) Homepage

    Seriously, they are disabling by default a feature that can help with the #1 problem with their browser - that it uses so much memory, people can run out with just a browser given enough tabs, regardless the GBs of memory they have, to avoid a 10% slowdown in browser benchmarks???

    When you run out of RAM, the slowdown is in the order of 10^6% - it just stops working!

    If you use websites that behave like a browser benchmark, then you are visiting the wrong websites - they are either mining on your CPU or are so badly designed advert-ladden that you should avoid them like the plague anyway. If visiting a website starts my CPU fans, I don't visit it again. Normal browsing should not be affected by a 10% CPU slowdown.

    Have an option to turn it off for benchmarking so you don't lag other browsers if you want to, but default off is insulting.

  • That's been the case in software design approximately as long as there's been software.

  • Chrome already unloads from RAM as-needed when the resources are demanded by other apps through the OS. Now aggressive RAM management uses the CPU and extra time reloading/decoding/rendering assets for the browser instead of using a cache in RAM will further use the CPU.

    All this does is burn through your battery faster.

    • Chrome already unloads from RAM as-needed when the resources are demanded by other apps through the OS.

      Stupid question, but I wonder what would it be like if Chrome et all used a prediction based means of determining if it requires, and thus will request, more RAM through the operating system, rather than just taking up RAM as it sees fit (irrespective of if it will use it or not)? I mean, of course I see an increased CPU load for sure, but I mean in terms of whether or not it would assist (EVEN IF slightly) in better RAM management, or make things worse, etc.

  • We have seen how security mitigations for Intel processors caused a reduction in performance. So, could one or more security mitigations be responsible for the performance issues with Intel processors? Might a Ryzen based machine perform better with this feature than the 9900k? I don't use Chrome since I am not a fan of the UI and I find that Firefox has fewer problems(no unexpected add-ons getting installed without permission the way they are in Chrome).

  • Capable of running more trackers and serve more ads.

    How much processing power is necessary to render a few static documents like a news article or web forum? Very little.

    It's clear that is not what is going on under the scenes when I look at some random webpage and see 10% of my CPU is used continuously. Optimizing Chrome's performance is treating the symptom of our broken internet rather than going after the bad behavior of the modern web.

  • Can the code be used to trigger the memory optimizations if less than say, 12GB of RAM, or if prior executions of chrome detected swapping was becoming an issue, but if you have enough memory, don't worry about it? It seems that this should be a tool to add to the tool-chest to ensure optimal performance, not an either/or answer.

Slashdot Top Deals

Top Ten Things Overheard At The ANSI C Draft Committee Meetings: (5) All right, who's the wiseguy who stuck this trigraph stuff in here?

Close