Safari keeps crashing/freezing after install of El Capitan

Thanks, I enjoyed the sarcasm. But seriously, along those lines, that is actually Apple's "solution" to the "Other taking up space on iOS" problem discussed in other topics: just get an iOS device with more memory. They don't officially say this, of course, but customers are led to understand that it's their only option, because a fix is not coming.

So what I would like to ask (because I couldn't find the answer searching in this thread): has anyone experienced the freeze on one of the new 12" Retina MacBooks? The latest of these use the Intel HD Graphics 515 GPU, whereas my mid 2014 13" Retina MacBook Pro has an Intel Iris GPU, which many people who contribute to this thread are using. If the new MacBook (which I'm considering getting) doesn't have the problem, that will be my solution: get a new computer.

Hey guys,

I just got my iMac back after another week at Apple and it wasn't a hardware issue after all... so they say...

They also said they disassembled my fusion drive and put it back together for some reason.

Whatever they did, I hope they fixed my issue.

It must be reported by now at Apple that this is still an issue for a lot of people.

Nick

Can confirm sporadic freezes in Safari when closing tabs with HTML5 video, namely YouTube. Remains unresponsive for 60-180 seconds, then lags and stutters until restarted. Clean installs, dirty installs, betas, releases – doesn't matter, it's there.

OSX 10.11.5, rMBP15 Mid 2014, Iris Pro, Wipr content blocker, no Flash.

Has been this way since the Yosemite days. Really annoying.

Thanks, you've perfectly described my experience (minor detail being that my mid '14 rMBP is the 13" model). Also no Flash, here, and also since Yosemite (but might've been before that, I don't remember whether I ever had Mavericks on this machine).

I still say it's the GPU. Either a bug or an interaction with some low-level graphics code in the OS.

If I end up getting a new Retina MacBook 12", I'll definitely report back here either way, whether I continue to see the problem or not.

As I've said before, I'm 150% sure Apple knows what this problem is.

As I've said before, I'm 150% sure Apple knows what this problem is.

Yes. There's a large amount of comments over the last year or so in this thread and others that discuss efforts of reporting the issue to Apple. But the absolute proof of their awareness of the issue is that a customer who filed this as a bug with a detailed statement in the developer program had the report removed by Apple due to it being labeled as a duplicate of another already filed. This is documented in another thread on this support forum. However, it unfortunately never extends to their public-facing support team who treat the issue as a new and unknown one.

In terms of Apple knowing the exact cause: It seems so likely but still strikes me as odd that it hasn't been addressed after n number of OSX point major and point releases as well as Safari updates. This statement may appear to ignore that it could be a hardware fault but I would have thought there were subtle ways of maneuvering around this particular one by software. If it is indeed a hardware issue, Chrome, etc. are negotiating around any HTML5 video freezes. Their interactions with the system frameworks may not be as low level as Safari (a big assumption on my part) as Safari is far more integrated into the OS, but surely there's ways of bypassing freeze points without noticeable performance impedance - ala Chrome.

Hey,

I got my iMac 5K(late 2015) back from repairs again which was again another week.

after 2 days off working it froze again.

Hardware checks out fine, Apple telling me its probably hardware related at this point.

My reseller says both because they didn't actually had to fix anything...

I'm so done with this ****...

I removed 2 RAM modules (16GB) and it froze within 5 minutes, I'm running on the other 2 modules now and see how that goes...

Hardware or software... this is **** annoying and I really don't like the way Apple treated me in my situation.

Haven't been able to use my iMac since April...

Nick

There are plenty of good legal reasons (having to do with liability and exposure to class action) for Apple to avoid making any public statements about this. That kind of policy is the rule, rather than the exception, among vendors of all kinds these days (it certainly was at some of the companies where I worked). The best that you can hope for is insinuation and "process of elimination" where they help you to identify all the things that *don't* fix the problem and leave it to you to figure out what does (iOS device with bigger memory; Macbook with different GPU). Of course with all the discussion and analysis that fairly competent people have done, it would be pretty hard for Apple to avoid subpoena and discovery on this issue, should anyone decide to litigate, but they don't need to throw gasoline on the fire. Personally, I've never felt that class action resulted in great benefit to the injured party; it's mainly the lawyers who make out like bandits. So I'm in no hurry to instigate anything like that.

If/when they come up with a workaround in firmware/software, you can be sure that it will be delivered in an update. The fact that it hasn't after nearly two years since the first systems started having the problem (maybe more) tells me that they can't fix it that way for some reason, and that they almost surely aren't going to. Just one of those unfortunate things (for Apple, too, not just customers).

That kind of policy is the rule, rather than the exception...

Yes, sorry, I had my own assumptions about why they're not being public about this and similar issues under the same thoughts to what you've just expressed and what you've mentioned before. I may have suggested otherwise when I lamented that it's unfortunate their public-facing customer service doesn't acknowledge it as an existing issue but it was just a big inarticulate sigh rather than anything more.

Along similar lines, this is why it's been so helpful over the past few months for several us to have diagnosed the causes and symptoms up to the smallest shared experience that we could. Unfortunately, for your sake, this hasn't included reporting on this particular freeze, to the same degree, from anyone with a 2015 machine.

But it still leaves me wondering why a software bypass of any minor hardware issue in Safari rendering hasn't been implemented both to beat away the braying crowd (us and others) and with the motivation of their brand incorporating a perfectionist-like ethos. This is under the assumption that is a hardware issue and under the same comparison I made above that other browsers render engines manage to avoid the issue when rendering the same web languages and protocols under the same OS. Though it's a small minority with this issue, it's a large small minority if that isn't too much of an oxymoron. Search for posts around the time 2014 machines first came out when the noise was a lot louder*.

*Therein lays some of the problem, yes, but I still think Apple is very keen attached to its perfectionist image as part of its brand and would likely be aware of what part combinations cause this issue and, therefore, have a rough estimate of how sizeable the customer base affected is - even if they're quieter on the issue now.

Having dealt with exactly this sort of problem (a possible GPU bug) over the course of my career, I can say that it's not hard to imagine what could go wrong that would make a firmware/software workaround difficult. Let's say, for example, that some sequence of operations causes the GPU's own firmware to get into an unstable state, where it's either frozen or "spinning" or otherwise not responsive to commands from the OS. Depending on how the system hardware was designed, it may require a full system reset to clear the GPU (some designs allow for the OS to perform a reset on a subsystem without a full system reset, but this may not be one of them).

So it may not be possible to clear the problem once it occurs. But the better fix in any case would be to identify the combination of operations that cause the fault to occur. It's possible that Apple engineers may not have fully identified the exact conditions that cause the fault. These complex systems are amazingly hard to diagnose, partly because you can't get inside to see exactly what's going on (it would be an unusual engineering environment that has full time-coherent emulation of *all* of the major subsystems). The subsystems (CPU, GPU, storage I/O, etc.) are all pipelined and they have multiple threads executing concurrently. In operation, they can appear to behave nondeterministically (i.e., unpredictably, never exactly the same twice).

Just as likely, in my opinion, is that it's a combination that they don't want to prevent, because it might, for example, require serialization that would slow down system operation to an unacceptable degree (perhaps it would be impossible to display HD video while also managing the rest of the graphic content - more on that in a bit).

I think that this is a tough problem, and one that they've probably gotten to the bottom of but don't have a fix for.

On the HD video speculation above: I'd be interested to hear if anyone has noticed whether the freeze occurs after viewing video at lower video resolutions. These may not actually reduce the burden on the GPU (depending on how HD video display is programmed), but it's just a thought.

This is probably just my lack of experience in software development but I feel like your (thorough, thank you for that) explanation still talks in a theoretical form about being able to bypass the probable hardware issue but doesn't address what I mentioned before about Chrome interpreting the same HTML5 and video codecs on the same hardware and OS without triggering the problem. To me this feels like this shows evidence of a solution and is no longer theory. Apple would be more familiar with their own SDKs, etc. than anyone so surely any Chrome rendering implementation that makes calls to OSX and Apple-selected hardware could be understood by Apple engineers working on their own core software (while acknowledging that they use separate render engines)? I'd be interested in hearing your thoughts.

Just to expand on the above with Chrome... I've mentioned previously in this thread that Chrome did freeze my laptop in the same way when I first got it. I can't be 100% it was when switching/closing tabs as I hadn't become familiar enough with the patterns yet. But it was definitely HTML5 YouTube videos. I switched to Safari, believing it didn't have the issue, and stuck with it for months until moving back to Chrome - which I discovered no longer froze, unlike Safari.

I can only speculate on why a different browser (or viewer, or any other difference) would not cause a freeze. Perhaps something that Apple considers essential in the way Safari handles video (something that they aren't willing to sacrifice) increases the likelihood of the problem - a specific sequence of operations, perhaps. Or maybe the problem does occur with Chrome, but less often, because Chrome does things slightly differently.

Obviously, we, in this forum, have very little hard facts to go on. It's really pure speculation. Since you (and others) said that you've seen freezes with Chrome and other browsers, and since people see freezes with and without Flash, we have not in any way isolated the specific conditions that cause the fault to occur. As I said, that can be a very tough job, and it's one that only Apple engineers are equipped to tackle competently (which I honestly do believe that they have, but without producing a fix).

From reading the many messages on this thread where Flash was involved, I personally believe they're unrelated freeze that share the coincidence of a video being involved. The few people who had it that went a bit deeper with diagnosing it didn't seem to share any any other trigger or symptom. It was a system freeze, not a GUI one. Closing tabs did not cause it. Often it was a CPU spike., or Safari extension issue, etc., etc. Often it was also a misunderstanding of over cluttered system with bad preferences, library settings, etc. I don't think there was any other freeze in this thread (and excuse me if I'm wrong) that was shared by a large group with exactly the same symptoms and triggers that occurred on a completely fresh system with no configuration changes, or software added.

(something that they aren't willing to sacrifice)

Perhaps that's it. If so, I'd love to know what it is but doubt I will ever find out.

I'm inclined to agree that there are several different "freeze" issues being discussed in this thread, but when I see a description as detailed and precise as what @alksv reported a few notes back, which, in particular, matches my hardware very closely, I conclude that *that* freeze is exactly what I've been seeing. And, as I reported in the past, while my system was "frozen" like this, I was able to connect to it with ssh from another computer and see that it was running fine. Well, obviously not fine, but process activity was very normal looking, with no high CPU, and only a few "stuck" processes (not unusual).

So right, there's not much precision in the discussion, but a lot of people are experiencing freezes related to viewing video in a browser.

there's not much precision in the discussion.

Sporadically, there has been. And those comments were foundations for others for anyone who was following the discussion for a long period with a vested interest. It got to the point where a significant number of us confirmed that similar hardware and all the usual checks (clean install/no pref changes, etc.) could trigger a freeze only when a tab in Safari was switched from or closed that contained the other shared elements that were agreed. Forgive my repetition, I know I'm like a parrot, but I can't help holding significance in that and am a little dubious of a shared cause with the specifics of some freezes mentioned elsewhere here - which does not include yours or alksv's. Perhaps there is a common root fault, perhaps not. Your freeze symptoms were the first I've come across that were so similar and yet had a tiny difference. I'd certainly see a possibility of a shared root cause easier in this case. Especially that the hardware is the same (well, on the broader level of year and screen size at least).

I remember the specifics of the freeze you were having as we were both discussing it and I was mentioning at the time that it was pretty much confirmed that anyone having the tab triggered freeze was seeing only the GUI stuck and all other processes continued, including SSH ability etc. I believe you were surprised I didn't have multiple computers laying about to SSH into the problem one 😉

LOL right, you remember quite well the entire history :-) So anyway, I agree that there have been freezes reported that may not share the same root cause.

One problem with so many people reporting symptoms and trying to analyze and narrow to a minimum set of conditions to produce the problem is that many of us aren't equipped to do the same analysis. For example, I can't, at this point, go back to Yosemite or Mavericks (well, I could, because I archived those systems). And others (yourself, for example) don't have a convenient way to try ssh from another system (more on that in a bit). Etc., etc. All of these variables make it very difficult to say whether there's a common thread in what people have been reporting.

On using ssh or some other method to connect to a frozen system, my memory fails me a bit, here, but I think that at one point I was able to connect to the "frozen" mac with Screen Sharing from another mac. Now if you don't have another computer, you can still connect to a Mac with an app on a smartphone (Android or iOS and probably even Windows Mobile). There are apps that can do ssh and others that can do Screen Sharing (specifically using the VNC protocol). If you set your Mac up to allow "remote login" in Sharing prefs, you'll be able to use an ssh app to connect. If you enable "screen sharing" or, preferably, "remote management", you'll be able to connect with a VNC app. Testing with any of these might prove interesting - I don't know what you'll find in your case.