Hacker News Reader

Comment by sxates | Hacker News Reader

What's missing here for me is some kind of performance evaluation. If I upgrade my 2011 MBA from Mavericks to Yosemite, should I expect any change in performance, for better or worse? Did the power management change in any significant way?

Apple's mobile OSs have a way of obsoleting older hardware. I'm curious to know if their desktop OSs are trending that way as well, or if they're making performance gains instead.

View on HN

I've had serious performance issues with Yosemite. For one, dragging around windows became extremely janky; this appears to be caused by the transparency (and the limitations of my first-generation Retina MBP), as the option to disable it fixes the jank. For another, opening a new tab in Safari started lagging about a full second before I could start typing in the address bar, which is a ridiculous amount of time compared to the previous version or competition. I was able to fix this by disabling Top Sites, but while I don't actually use Top Sites much, I never had to disable it before, and most users won't know to do that, making it a serious regression. (Chrome actually takes quite a while to load its equivalent - I don't know whether it's for the same reason or not - but it does it asynchronously, so I can start typing in the address bar almost immediately.)

On top of that, there are serious performance issues with the built-in Japanese IME, although I'm not completely sure how much of this is new, and I experienced a bug where WindowServer would randomly start hogging CPU, which may or may not be related to said IME.

Given all these problems, I'm surprised that the general consensus seems to be that the performance is good. But maybe I'm somehow a special case. (The Safari issue is the most egregious - is it that many OS X users don't use Safari in the first place?)

I've had some pretty bad performance with the demo, too. It seems to be much better (but not perfect) now with the official release. I'm not really sure why.

I'd love to try and get to the bottom of the Safari performance issues you're seeing. My email address is in my profile, please drop me a line if you wouldn't mind.

Okay.

Did you just update last night? I was getting janky/bad performance and a lot of glitches immediately after updating, but it smoothed out (presumably due to Spotlight indexing).

My post was based mainly on experience throughout the public betas, including the last beta which is supposed to be almost identical to the final release.

In using the beta, I haven't noticed any actual performance/battery difference between the two. Honestly, when I first installed it the first thing I noticed was the lack of sound when changing the volume.

Heh, me too - you can turn it back on in System Preferences though. It's a new sound effect.

Do you know exactly where? I couldn't find it.

I don’t have Yosemite, but in Mavericks, you can change it in System Preferences > Sound > Sound Effects, by checking the checkbox “Play feedback when volume is changed”.

It's still there, but now you hold shift+vol up/down to hear the feedback. This is in contrast to earlier OS X versions, where holding down shift will mute the audio feedback.

Without quantitative metrics, the responsiveness is just way better than Mavericks. With Mavericks I just felt like I was waiting for it all the time, even on a maxed out laptop with SSD.

With Yosemite, even my 3 year old iMac is doing way better.

Safari has always been slow for me until this release.

The trade off is that I really hate the way Yosemite looks but I can live with the tradeoff.

Note: all my machines have 12 GB or 16GB of ram

There might be a small battery time reduction because these new vibrancy effects will cause a bit more cpu consumption than Mavericks. It's probably negligible though.

Yosemite is completely unusable without Reduce Transparency on IMO.

Unusable in terms of performance?

That's a very small part of it. I just hate the new look in some places, like the color-changing Safari toolbar, and it's just useless and annoying to me. Maybe 'completely unusable' was a bit of an exaggeration…YMMV.

Mavericks added performance and battery life, but as far as I'm aware there isn't a focus for that on Yosemite - the work having been done in Mavericks.

Apple has a pretty poor track record in recent years on perf. Especially with iOS updates. If Yosemite didn't focus on perf then it seems likely to have gone down and older machines will suffer. Thats not necessarily wrong but it is important to understand before you dive in.

> Apple has a pretty poor track record in recent years on perf

Apple has a bigger problem where they're popular enough that every release has a ton of people who post subjective problem reports but rarely provide repeatable benchmarks or failures. They're definitely not perfect but if you see a report which doesn't have a specific test with the exact steps needed to duplicate it, it's wise to assume it's an urban legend.

I have experienced it first hand with both MacBooks and iPhones. That's enough to make me hesitate to click upgrade. It's certainly anecdotal but it's my anecdote that I would be a fool to not take under consideration. You should treat my story with more salt. Anecdotally a lot of people have experi need what I have and are also hesitant to update based on their own personal anecdotes. Take it as you will.

The other thing people need to remember is that our perceived experience is notoriously suggestible if you don't take efforts to correct for it. If you expect a new OS release to be slower, you're more likely to experience it as slower – and that first reboot will likely "confirm" it because every persistent cache has been invalidated, one-time upgrade tasks are running, all of the app updates which blocked on a major OS version requirement are installing in the background, etc. Very few people will take the time to measure before and after multiple times to know whether there's something objectively different.

It'd be educational if someone like Apple or Microsoft shipped an update which changed only the version number and then recorded feedback, particularly since there would be e.g. some percentage of people who had something like a recent hardware problem which they hadn't noticed and assumed was caused by the update.

I don't know that that's been the case on OSX though — since 10.0 the story has consistently been that each release is as fast or faster, albeit more memory hungry.

I'm sure I'm missing one but I can't remember any OSX release being associated with performance complaints.

Mavericks is very buggy (locking w/ multiple monitors; finder bugs). I'd wait for at least update 2.

Oh, and the .0 release will, if the pattern holds, have shit perf in some common situations. I presume because apple doesn't test for that.

Reply on news.ycombinator.com