So has Firefox become the new memory resource hog over Google Chrome ?
I frequently will see it run between 15GB ~ 20GB of usage with all the tabs open after a day or two.
I’ve noticed that Task Manager will show my memory resource usage getting pretty darn close to 99 % or so so I started looking at the processes and noticed this usage.
Closing Firefox regained the memory usage and the sudden stutters in other web browsers was gone due to the memory usage getting darn close to 100 %
I’ve installed a free app named Wise Memory Manager which will run in the systray and monitor the usage and you can setup a threshold to have it attempt to clear the RAM for you.
The gaming rig has 64GB installed so I didn’t think there was be an issue with leaving the web browsers just running in the back ground but that may no longer be the case.
Yes, I see it happen occasionally, with memory leaks either in some webpages or occasionally parts of the browser. I’ve also had my GPU process balloon up to 50GB, found out you can kill it and firefox will resurrect it without crashing, so I do that regularly now too. I also use this Auto Tab Discard – Get this Extension for 🦊 Firefox (en-US) and I turned off the auto part and now just use it for selective tab discarding.
I watch mostly YouTube on my HTPC (Win10 Pro, Elitedesk 800 G1 SFF, i5-4570, 8GB RAM, RX/550 low profile). This leads to me having 15 or 20 tabs open at a time, and normally the computer goes into a sleep state shortly afer I do. Might get a reboot once a month.
I have never seen it tick past 5.5GB RAM utilization in this state.
Maybe this is a type of “Linux ate my RAM” situation? Some kind of opportunistic use of unused memory, effectively.
I don’t think so, I run Firefox on Windows 10 and I do track Firefox’s memory usage both by its own process manager, and an external task manager, and note the commit space usage because if I don’t Firefox will cause me to OOM (and my system maxes out at ~200GB of commit space with 64GB of physical RAM). I do try and keep a lot of open tabs though (currently limited by my commit space). Another thing I learned was about:memory has a minimize memory usage button. It often does not decrease the amount of physical RAM used but for me it consistently removes 7+GB from my commit usage.
I refuse to use a chromium based browser, but I really do wish the Mozilla org ( which I’m pretty sure does have a lot of cash) seemed to care about the browser, I want to fork it and fix or change a lot of things but I haven’t been able to find the time. I would try to upstream all of my code but I have seen them be unwilling to adopt patches for tracked bugs, so I would rather just fork and worry about upstreaming later.
Is all the ram, from holding firexox data, or the data from pages in tabs.
External tools will not be able to address what is being held inside, if firefox is trying to hold too much.
So, perhaps a tool that does not load the contents of tab, rendering them as semi-ephemeral bookmarks, instead of running tabs is indeed the best approach, as already mentioned?
If the auto-tab discarder does not work, give other ones a go?
I use the external tools to track total commit size usage (because it is very annoying to deal with system OOM) and seemingly unreported usage, the 7+GB of freed commit space that comes from hitting minimize memory usage does not seem to reflect in about:processes.
The discard add on works (and it uses a firefox native discard call), but there is currently no workaround to address the main process using unnecessary memory (mine is currently sitting above 10GB) , and the GPU process will occasionally memory leak and has to be restarted (like I mentioned above had this hit 50GB) .
I do have some ideas that I will eventually either turn into a browser add on, or a browser feature to help deal with sites that leak memory without having to discard them. I also have looked at the memory reports that you can generate because I was and still am planning on trying to fix some underlying issues.
The original thread creator had an issue with a specific webpage, that was effectively leaking memory by attempting to create infinite vimeo subframes, which was identified (and resolved?), but I was more talking about the state of memory usage in Firefox in general because the opening post does ask about that. Discarding a tab will often leave data because firefox will often have tabs share a process (which is why the about:processes page will only allow you to discard entire processes, not individual tabs).
I don’t actually care about reporting niche bugs I find anymore given how slow I see things happen on bugzilla (AFAIK the GPU memory leak bug was reported by someone else years ago). My usage of Firefox is probably also very unorthodox, and thus I do agree that it probably isn’t worth anyone else’s time to fix given the state of bugzilla and the countless other more important bugs.
I’m sorry but I have lost faith in them having the bandwith to address minor problems (and according to their own triage basically all of the occasionally memory leak/OOM issues are minor).
I do appreciate the work that the Firefox devs do and I hope I do eventually make time to contribute to the project, but I don’t think traction alone gets you anywhere (HDR support was added to MacOS has been asked about for a while for Windows and Linux, I understand Linux has had blockers with that with Wayland, but windows seemingly does not have external blockers)
Also a bit ranty below:
Also a lot of the issues I run into, I can either immediately find a bugzilla ticket that is very old (such as the GPU issue) , or if it is a very new issue (like firefox android becoming very hard to use for many users since it would OOM your whole phone for typing 3+ characters on the address bar [icognito was not effected ], this took months for a resolution, not including the time it went unreported, and I am confident this lost them users) I will find eventually, but some are very very niche and would be triaged accordingly (for example swapping tabs while the minimize memory usage process is running can very very rarely [ I don’t know how to reliably reproduce it, but it does seem to be very timing dependent, and may have other dependencies] cause the entire browser to go POOF, without any crash report). They followed Chrome on not supporting jpeg xl and even stopped taking community patches for it ⚙ D119700 Bug 1709814 - Add support for color profiles for JPEG XL. r=tnikkel⚙ D122158 Bug 1709818 - Add support for animated JPEG XL. r=saschanaz⚙ D122159 Bug 1709815 - Add support for progressive decoding for JPEG XL. r=saschanaz . There are more than a few features (besides jpeg XL) that they refused or have not gotten around to that I want to add in a potential fork.
I found something interesting with the Firefox tab memory issue, the site that is causing this problem uses embedded video links to vimeo so I placed vimeo.net/vimeo.com in my hosts file and pointed it to localhost and now that same tab is not using nearly the same amount of memory, just about half of what it used to use.
Also noted that the overall memory usage has stabilized too.
I’m glad you fixed your issue, although your solution does work and doesn’t require any changes to the browser, I do think a more fine grained solution that does not fully block vimeo could be done and would either involve either tampermonkey or ublock origin or even just a custom extenstion.
@235SAS thanks for the feedback, but I don’t actually use Vimeo for anything so it makes little difference to me that it is blocked at the hosts file level.