Chrome Might Not Eat All Your RAM After Adopting This Windows Feature (extremetech.com) 40
A new feature in Windows 10 might allow Google to streamline Chrome, and we know it works because Microsoft is already using it. From a report: According to Microsoft, its recent update implemented a new memory management feature in Edge known as SegmentHeap. In the latest version of Windows, developers can opt into SegmentHeap to lower the RAM usage of a program. Microsoft says it already added support to the new Edge browser, and it has seen a 27 percent drop in the browser's memory footprint.
This will be easier (Score:5, Funny)
Chrome Might Not Eat All Your RAM ...
I've been spritzing it with water when it does this, like my friend does to her cat when it chews on the sofa.
Re:This will be easier (Score:5, Funny)
Holy crap, Slashdot, you were supposed to destry the Sith, not join them!
Client-side bidding causes bloat (Score:2)
It's hard for a browser publisher to optimize anything properly when so many websites include client-side bidding adtech that shares the user's viewing history with over a hundred advertisement networks trying to squeeze out more fractions of a cent from a few square inches of the browser window. The easiest way to slim down any browser's RAM use is to turn off execution of JavaScript in HTML documents, or at least to turn off execution of those scripts known to track a user's viewing habits across differen
Re: (Score:2)
Re: (Score:2)
Check your memory every 6000 flops (Score:4, Funny)
I know with my chrome usage my memory is worn and less responsive after 6 months usage. I've switched to synthetic memory which gives me better up time. Remember to pull your DIMMs and blow on the contacts every few months because heavy Chrome usage eats away at your hard working memory! (Porn too)
Challenge Accepted!! (Score:5, Funny)
- Google Chrome Team
Re:Challenge Accepted!! (Score:5, Funny)
So opening this web page will now take only 400 mb instead of 600m
Re: (Score:2)
So opening this web page will now take only 400 mb instead of 600m
600 meters? Your monitor is HUGE. :-)
Re: (Score:2)
Explanation (Score:5, Informative)
Neither the incompetent ./ editors nor the website reference in this story could be arsed to actually provide any useful information. So here is a link to the bug report that explains what is going on.
https://bugs.chromium.org/p/ch... [chromium.org]
Re: (Score:1)
Actually, from reading through that bug report, I'd say lay the incompetence at the Chrome developers' feet. In particular, the underlying assumption they seem to have that all available resources are fair game; no other application matters. And, second to that, they're terrible at managing their heaps.
Re:Explanation (Score:5, Informative)
Chrome Dev - 10/15/2019: Performed extensive memory analysis of Chrome on Windows and found that more than half of the memory being used by Chrome was actually freed but Windows wasn't smart enough to release it for reallocation. Realized this could be fixed via Segment Heap and asked Microsoft how to opt in to that feature.
Microsoft - 2/10/2020 - "FYI, on Windows arm64 the Segment heap is the default."
Chrome Dev - 4/29/2020 - Notices opt-in for big boy apps (non arm64) in Windows May 2020 release. Notes that Edge in Windows May 2020 release has already implemented this.
Microsoft - 5/27/2020: Provides documentation to opting in to Segment Heap for Windows May 2020 release
Microsoft - 6/18/2020: Microsoft publicly announces that Segment Heap may be useful for Chrome and that they've already integrated it in Edge.
This is too funny. Google knew that Segment Heap would be useful for Chrome and have been trying to get access to the API since last October. Microsoft doesn't release access to the API until they've already integrated Segment Heap into Edge. Then, they finally give Google documentation to the API and release a press statement 3 weeks later acting as if they discovered something that could help out Chrome.
Re:Explanation (Score:5, Informative)
Microsoft doesn't release access to the API until it releases the very first version of the win32 API that implements the Segment Heap API. Shocking.
Microsoft tests a beta browser in conjunction with a beta OS for a coordinated release. Doubly shocking.
BTW, Segment Heap has been available to "big boy apps (non arm64) for quite awhile. They were UWP, not win32, but it's your definition, not mine.
Re: (Score:2)
Microsoft doesn't release access to the API until it releases the very first version of the win32 API that implements the Segment Heap API. Shocking.
I agree that there's no evidence MS was dragging their feet inappropriately. OTOH, this finally explains all of the whining I see about Chrome eating huge amounts of RAM, which I've never observed... because I've never used it on Windows. Its memory consumption seems pretty reasonable on Linux and OS X. But if Windows wastes 60% of the heap, I can see the problem. I wonder why some other browsers don't experience the same... different allocation patterns, I guess. Something about Chrome's approach interacts
UWP browsers have to wrap Edge (Score:2)
BTW, Segment Heap has been available to "big boy apps (non arm64) for quite awhile. They were UWP, not win32
Correct me if I'm wrong, but it sounds like you're trying to suggest that it's practical for the publisher of a Win32 application to port it to UWP just to be able to use Segment Heap. If so, this is not the case for web browsers. Last I checked, Microsoft expected publishers of UWP applications to distribute these applications through Microsoft Store, and Microsoft Store banned web browsers that don't wrap Edge. Quoting Microsoft Store Policies [microsoft.com]:
Re: (Score:2)
No, I'm expressly saying that this statement, "Notices opt-in for big boy apps (non arm64) in Win
Re: (Score:2)
sounds like garbage collection is now part of the main heap, probably for "performance" and would work fine, if only those pesky applications didn't just assume that the entire installed RAM was dedicated to themselves.
Re: (Score:2)
Re: (Score:3)
According to another source (don't recall who) Microsoft changed their HEAP memory allocation in the kernel with release 2004 feature update that was released last month.
Developers have to do the new syscall() as Windows is plagued with win32 apps that depend on bugs from ancient version to run and to change the HEAP automatically might break something expecting an ancient behavior from Windows XP etc. Ask any IE 6 developer about this?
Yes Google has to turn it on and it is not incompetence.
FYI release 2004
Re: (Score:2)
Enterprise even better, can defer feature updates 365 days, Though it is harder to get a lic for...
Re: (Score:2)
I think I can delay up to 365 days as well with Pro. I choose 180 as I do not want to be that far behind and my security and bug fix updates are 30 days.
Enterprise has very very slow and buggy Windows Update mechanisms, lacks WSL1 and WSL 2 for Linux, lacks driver support for some GPUs like my 2080Super, has outdated Powershell support, and has outdated Hyper-V virtualization where I can't run Microsoft's labkit SCCM lab (requires 1607 last I looked).
Enterprise isn't for everyone and is best for hospitals o
Re: (Score:3)
It is the heap allocator. Real quick. Memory can come in stack or heap (generally speaking). Without getting into nasty details, stack is limited (~1MB on Win10) but managed for you. Program ends? Anything in stack is cleaned up for you. Heap is memory that the OS allocates and manages. The OS manages the heap using an "allocator". The allocator has to keep up with a lot stuff to manage that memory effectively. Windows 10 has two allocators built into it. The old NT allocator is the one that came
ram doubler ideas? (Score:3)
ram doubler ideas?
RAM compression in Windows 10 and Linux 3.14 (Score:2)
Connectix RAM Doubler worked by devoting part of a Mac's RAM to a compressed page file, occasionally spilling pages to disk that could not be compressed adequately. Once the patent on RAM Doubler expired in the mid-2010s, Microsoft was able to add similar technology to Windows 10 [howtogeek.com], as were Linux developers to Linux 3.14 (under the name zram [wikipedia.org]).
Need more evidence... (Score:2)
Ya...I'm not quite on the "knows it works" boat yet...the evidence seems extremely flimsy.
Re: why not just fix the existing memory leaks (Score:2)
Virtual memory is not a useful metric. Find out how much resident memory is being used.
Fuck off with the clickbait headlines msmash (Score:2, Insightful)
You are literally ruining the site.
Re: (Score:1)
Agree. !00% agree. If this keeps up, I'm done.
Re: (Score:1)
Don't use Chrome (Score:1)