-
-
Notifications
You must be signed in to change notification settings - Fork 17
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[BUG] Frequent display freeze or blankouts #198
Comments
I think I'm seeing the same on OS X 10.9.5 and build 120.0.6099.199.1 |
Let's focus in on the situation where a tab freezes, and triggering mission control unfreezes it. I would consider the case of tabs going completely white to be a different issue (and one which is already tracked as #94). I have never seen anything like this on 10.9, including on my laptop which has only 8 GB of memory. @RJVB is it possible to provide more specific reproduction steps? I realize that might be hard! How much memory do your machines have, and is Settings → Performance → Memory Saver turned on or off? For reference, I have memory saver turned on for my laptop. It's off for my desktop, but at 32 gb of memory I don't expect to ever run into high memory pressure. |
Let's focus in on the situation where a tab freezes, and triggering mission control unfreezes it
TBH, I cannot recall having seen that myself either. The white page thing is rather frequent though, esp. on instagram and it does seem to be related to being low on RAM.
@RJVB is it possible to provide more specific reproduction steps? I realize that might be hard!
For the white page, I'd say get on instagram, follow a bunch of people/companies and start perusing the front page and then the "stories". For me that's a good way to trigger crashing extensions, too.
BTW, I've come to get the impression that CL can start up in 2 states, one where extensions start crashing almost from the go, and one where they don't. Since I do not currently use CL as my main browser I just restart it if I get the first state... I don't yet have enough data to say if the white page thing is linked to the extension instability but at this point I'd say it's not, though.
How much memory do your machines have, and is Settings → Performance → Memory Saver turned on or off?
Maxed out at 16Gb. I have the memsaver off in Chromium because I run "The Marvelous Suspender" (which *is* configured to use it for the tabs it suspends). Evidently I'm not complaining about white suspended tabs ;)
I also run Memory Cleaner (from the app store) but not in automatic mode. Triggering it when CL is running and acts up.
|
White page is being tracked in #94. I get it too (and I don't think it's memory pressure related). Copying the URL into a new tab (and closing the old one) sometimes helps, as does quitting and re-opening Chromium. I originally typed a bunch of additional thoughts here that I'm going to move to a discussion thread: #213 I cannot reproduce is the other issue described in the OP, where a page freezes but recovers after triggering mission control. Since no one else seems to experience this, I'm not entirely convinced this is a Chromium Legacy problem. |
Also: Please try going into Settings → System and disabling "Use hardware acceleration when available". Chromium Legacy forcibly disables GPU compositing, but some GPU rendering still takes place by default. This will turn it all off. My testing over the past 24 hours indicates that flicking this fully off substantially reduces "page randomly goes to white" issues! |
Also: Please try going into Settings → System and disabling "Use hardware acceleration when available".
I could try that for testing, but it's not a setting one would want to use as the solution for this problem, esp. if you indeed mean that it doesn't prevent all "white screening".
One context for which I use CL is video playback; it does a lot better with videos that were probably uploaded as crazy-hi-res (4k or even more). A perfect example of that: the IG stories posted by "Luna the Pantera" (hey, every legacy OS X user loves big cats, no?! ;) ).
|
I actually don't think I've seen a white screen since turning this on, although I'm not positive. Personally I download most videos to watch them in QuickTime, but you should also be able to use e.g. mpv to stream videos incrementally. I also wonder if there's a combination of command line flags which could disable all hardware acceleration except video decode... |
Well I take that back because I just saw one. |
On Tuesday March 12 2024 06:17:34 Jonathan wrote:
Personally [I download most videos to watch them in QuickTime]
For longer videos I use the "Video DownloadHelper" extension (https://www.downloadhelper.net/), fetch the stream URL and watch that in a self-built copy of QMPlay2. I also use that application (everywhere!) to search for and watch YT videos.
But that doesn't work for IG stories, reels and the like, and it'd be way too cumbersome if you're following even a few accounts that post frequent stories.
The question here of course is to what extent Chromium uses hardware acceleration for other things too. GPUs cannot just do hw video decoding and fancy rendering (I guess that any rendering via OpenGL should still use some form of hwaccel). OpenCL also uses the GPU and I wouldn't be surprised if modern Chromium leveraged some of that. Probably to little effect on our old CPUs and their legacy OSes but I'm convinced that every little bit of unloading of the CPU is going to help on such systems that are already getting pushed by modern software.
BTW, do you also get a technicolour show when launching CL (with the "restore session" setting active)? That reeks of video memory not getting initialised (and emptied) as it should I wonder to what extent that can have side-effects leading to some of the glitches we see further down the road.
|
I don't see that, no. I typically keep restore session turned off, but I tried with it turned on just now, and I didn't see anything.
It's worth testing with hardware acceleration both on and off. I can't detect a performance difference unless I'm running something like a 3D WebGL game (in which case, yes, the game is unplayable without hardware acceleration). Back before GPU compositing was force-disabled in Chromium Legacy, leaving it enabled led to obviously worse performance across the board (in addition to making the browser completely unstable). I assume this is because it wasn't really working properly, but I don't know that the remaining accelerated features are working properly either. |
Transition between progress circle and button should be animated. - fading element should go to 75% size and alpha 0 - appearing element should go the other way Each phase is 150 ms. Link to animation https://drive.google.com/file/d/1HsaTC_JnMEAODnQNNlEF9_fxe0ptb2N0/view?usp=drive_link (cherry picked from commit 69c865a) Bug: 329436073 Change-Id: I787d4a7d287035a2621ff93915dbc8be6a3ed788 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5372475 Commit-Queue: Olivier Robin <olivierrobin@chromium.org> Reviewed-by: Quentin Pubert <qpubert@google.com> Reviewed-by: Ewann Pellé <ewannpv@chromium.org> Cr-Original-Commit-Position: refs/heads/main@{#1275694} Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5388049 Auto-Submit: Olivier Robin <olivierrobin@chromium.org> Cr-Commit-Position: refs/branch-heads/6367@{#198} Cr-Branched-From: d158c6d-refs/heads/main@{#1274542}
Describe the bug
Not consistently, but apparently mostly when memory pressure is high, some tabs either get their display frozen, meaning not refreshed and not reacting to user input, or get completely white.
They get out of this weird state either by themselves after a while, or when triggering "Mission control" and coming back.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
Tabs should display their content, updated, and react to user input such as clicks or keystrokes.
Instead, sometimes one of these two situations will occur:
After a while, it solves itself out by itself. It also comes back to life when going to Mission Control and back.
Desktop (please complete the following information):
The text was updated successfully, but these errors were encountered: