Firefox Nightly on macOS: decrease in power usage by a factor of about 3x | Hacker News Reader

Firefox Nightly on macOS: decrease in power usage by a factor of about 3x

twitter.com

230 pointssohkamyung50 comments

I tried it after the comments in a previous HN post (https://news.ycombinator.com/item?id=20850135).

My results were very positive when idle (I can finally start FF and have it in the background without fans blasting).

However, just scrolling up and down in the default mozilla website increases power usage brutally (macOs activity monitor showed the energy impact as 200).

I'm still glad to see that fixes are finally coming, even though it's not usable for me yet.

If we get to a point of having it fixed and being able to use webrenderer on macbooks to have the gpu drawing, it could be a 180 for firefox's usability on macOs

I would really like an ETA for the rest of the changes.

Odd, I've just tried with Current FF and I can't get usage scrolling up and down that page to get above 6.3%

iMac (21.5-inch, Late 2013)

I wonder if this discrepancy could have something to do with switching to discrete graphics or something along those lines. iMacs can run their graphics card always and are designed to take that noise into account. If OP has a laptop, he might be hearing the graphics card revving up after being idle.

Does this also solve the 100% cpu usage on retina screens and the complete locking of macOS every now and then?

How long does nightly stuff usually take to get into the final release?

Have any details? I have a retina screen and don't have any CPU usage issues.

There are many bugreports known to Mozilla already. Here's mine: https://bugzilla.mozilla.org/show_bug.cgi?id=1407536

AFAIK, it's a known issue with retina screens and scaled resolutions. If you use the default resolution, you should not see any issue.

The default resolution has been a scaled resolution for years now.

I think GP means non-integer scaling

So do I. My MacBook Pro came out of the box supersampling at 3360x2100, or 1680x1050@2x. The screen is only 2880x1800.

The Firefox release schedule is pretty fast. This stuff should be in the beta release due out today, and the stable Firefox 70 release due out October 22.

Thanks!

> How long does nightly stuff usually take to get into the final release?

You can try it out now. Open a terminal:

$ brew cask install firefox-developer-edition

$ cd ~/Library/Application\ Support/Firefox/Profiles

$ ls -l

(Determine which directory is your current profile directory. It should be recently modified and end with .default-release. $ cp -R this directory to .dev-edition-default and start Firefox Developer Edition.)

Then in about:config put gfx.compositor.glcontext.opaque to true.

I have to say that with a lot of tabs Tree Tab Style is still taking up a lot of cycles. Up to the point it (nearly) hangs. At such point, I switch to Sideberry which does not, use OneTab, and start from there. Perhaps setting browser.tabs.20FpsThrobber to true solves this. Trying that out now.

Firefox Developer Edition is not the same as nightly from what I understand.

Setting gfx.compositor.glcontext.opaque to true makes firefox unusable as alerts aren't readable (might be theme dependent).

The only setting that you need to get FF to work on macos with a retina screen is the 'low resolution' mode trick. But then all text, etc isn't as smooth.

I am now running some tests with the new nightly, so far it seems like it fixes the issues and I am very excited about this!

> Firefox Developer Edition is not the same as nightly from what I understand.

It isn't, but this fix has already been around for a while. I've been using it for about a week. Back then, I was advised to run Firefox Developer Edition. I'd rather run that, than Nightlies.

according to @whimboo[0], these specific changes will land later this week on the developer edition.

[0] https://twitter.com/whimboo/status/1168476885766082561

Note the wording: "all of these changes". The first part of the change which I've been using for a week now, is already in Developer Edition. It has the largest effect. The most recent change has a small (but noticeable) effect.

FWIW, Nightly just managed to crash my MBP. Which, for the record, is normally rock stable.

Caveat emptor: You are likely to be running with altered config settings, as indicated by your advice above to others, and it’s occasionally found that about:config changes are what are causing crashes. Consider unwinding your changes to provide a more accurate outcome. (I’m not part of Firefox, just pointing out an especially common expert user self-inflicted issue.)

Thank you for the heads up. It hasn't crashed ever since (the nature of the crash made it appear related to using Nightly). I'll be switching back to Developer Edition ASAP as either b3 or b4 has the same fix.

For anyone interested, here is the firefox release calendar: https://wiki.mozilla.org/Release_Management/Calendar

FWIW, I get the CPU clobbered on a non-retina MacBook air, whereas FF works flawlessly on a retina one.

same here

Never had this problem. At worst some pages manage to spin up my fans.

Just installed it. And had gmail running on it and Chrome at the same time. Firefox Nightly showed to be using significant power while Chrome was not. I think this still has a long way to go before I switch to Firefox both in terms of performance and dev tools.

You're not the first to complain about the performance of Google property in non-Google browsers.

I'm not sure why it should be, or what it implies, but I've seen more than one person pointing out the difference.

Not being a Google user, I'm quite happy with the current Firefox performance, and delighted that Mozilla is finding even more room to improve.

“Web service designed to crush batteries when used outside of chrome uses a lot of power still.” Film at eleven.

Anyone knows how this compares to chrome?

https://bugzilla.mozilla.org/show_bug.cgi?id=1429522#c45 Linked in the Twitter thread.

It's in the link. The decreases in power usage put it on par with Chrome.

So, still terrible then?

The reason I use Safari is because Chrome is unbelievably power-hungry. I didn't know Firefox was even worse.

Same, I switched to Safari a couple of years ago. I was initially hesitant, because I used the developer tools in Chrome quite a bit. I was pleasantly surprised to see that the Safari developer tools is very mature, and in some ways better than Chrome.

Yes, compared to Safari. The Safari engineering is very impressive! I think there is quite a bit more to come on the Firefox side though, so hopefully they'll close the gap.

How does the situation compare to other platforms? Android, Linux, Windows?

That shows how shitty all previous versions where.

What the hell does "decrease by factor 3x" mean? Did power usage previously decrease by 1% and now by 3%? Is power usage now -200%? Seriously, I do not get it.

I think what they are saying is: "Current stable Firefox release uses N battery power. The nightly build will use a third as much (N/3) by comparison"

Honestly there nothing ambiguous about it so I don't get why you were confused.

Take the current power, divide by 3, that is the new usage.

So decreasing by 3 is the same as decreasing by 66%?

Exactly

Probably some US colloquialism? Doesn't make sense to me either, except in the sense "went down 300%" which obviously wasn't the purpose here.

nah, its quite common here even in Southeast Asia

I dislike this common usage as well because of the weird properties it has too (if "decrease by two thirds" and "decrease by three" is the same, the "decrease by" function has a rather weird shape, also, 66% faster = 200% faster?), but at this point, seeing someone use the term the way I think is "correct" is the exception that I notice rather than the rule. This fight is lost. When speaking myself I tend to just use an unambiguous construct (e.g., "the code takes only 1/3rd as long to run") and let everybody else do their own thing. It's rare for it to be a real problem.

Honestly, doing any sort of math with percentages in particular is just silly and should be avoided. Schools try to teach an official interpretation of "56 + 34%", but it's clear it doesn't stick and the adult world uses it fairly inconsistently in general. (In specific cases it is used unambiguously, for example, the accounting world, but in general, it's very sloppy.)

Decreasing or increasing by a factor is mathematically cleaner than decreasing or increasing by a percent. To decrease by a percent the formula is (1 - 66/100)*x, to decrease by a factor of three the formula is x/3. It might not seem like a big difference, but the latter can make things a lot cleaner in long finance formulas.

no, decreasing by 3 is `a - 3`

decreasing by factor of 3 is `a / 3` or `a * 0.(3)` or `a * 33.(3)%` or `a - a * 66.6%`

edit: math is hard!

Do you have similar issues with the phrase "increased by 2x"?

Why the need to convert to percent?

When people say 3 fold, they mean 3 times. I think fold should mean 2^3. When you fold a piece of paper, it doubles each time.

So on the news when they talk about a 10 fold increase, I know they mean 10 times and not 1024 times.

Does anyone else here feel the same, or is fold just another confusing colloquialism?

Well, when you fold something you don't necessarily have to fold it in half.

To one-third the original.

If a is 3b, then b must be 1/3a. I’m reading it as cutting 2/3 of the power requirement.

Reply on news.ycombinator.com