[andreasley]

Note that these tests were conducted on a MacBook – presumably running OS X. Apple invested a lot of time optimizing Safari and OS X for low power usage, so it's not exactly surprising that Safari wins in most cases. It would be very interesting the see a comparison with Windows and Linux (if possible on the same hardware).

[stephengillie]

Lots of other commenters have remarked about the lacking rigor and narrow scope of the tests. It would be interesting (and time consuming) to re-run this test with many more parameters:

  * Fresh install of OSX vs fresh install of Win8.1/Win10 on capable Macbook. (I'm unsure about Linux or other distros running on Macbooks, add that to the test if it's feasible.)
  * Javascript enabled and disabled.
  * Flash enabled and disabled. As well as Silverlight and other media players.
  * Various adblockers, tracking blockers, and content filters (like Adblock vs Adblock Plus vs uBlock vs uBlock origin, etc. And uMatrix and Ghostery et al.) This could become its own research project.
  * Logged into tracking sites (e.g. Facebook) vs logged out.

Maybe this testing could be automated with Puppet/Chef?

Edit: Windows 10 would be a definite option instead of Win8.1. I was unaware Win10 would operate on a Macbook, but in retrospect I suppose there's no reason it wouldn't.

---

Note that high-power-consuming sites also consume more electricity on desktop PCs; this is less of a concern because they are on a constantly-replenished power system. The concern may be closer to that for CFL vs incandescent lights, where we can make a few changes to the system and make it vastly more efficient, multiplied by a large userbase.

[bhauer]

> Fresh install of OSX vs fresh install of Win7 on capable Macbook

Why Windows 7 and not Windows 10? The article is capturing fascinating data, and I'd like to see how Edge stacks up.

[josephmx]

Windows 8.1 probably makes more sense, I'm not sure how stale a 1 day old OS would be

[zobzu]

Specially on a macbook woth notably nbad windows drivers...

[joshuapants]

I upvoted you because Apple's drivers for Windows are notoriously bad, but the major impact is on user input (trackpad, for example) rather than on performance (in fact, Windows tends to perform better than OS X on a number of tasks on a given hardware configuration) so I think it would still make sense to run the test on Windows on identical hardware.

[zobzu]

its not about speed its about power. if hardware isnt even instructed to go into sleep mode for example, there will be large differences.. which tends to happen on macbooks running windows.

[maaaats]

A post on HN a few weeks ago showed that Windows 10 preview was smoother than OS X on a MacBook: https://news.ycombinator.com/item?id=9555628

[mwcampbell]

Regarding Windows 10 on a MacBook via Boot Camp, I can confirm that it works; my boss has been running insider builds on his MacBook for a little while. Just make sure that if you get to Windows 10 by upgrading from a previous version, you repair the Apple driver installation afterward.

[eridius]

Desktop PCs don't have battery concerns, certainly, but electricity still costs money, and if you can reduce the power usage of your computer by a significant amount, that can translate directly to saving money.

[semi-extrinsic]

The cost of electricity for running a desktop computer is on the order of 2 cents per hour (assuming it draws around 200 W while browsing). If you reduce that by 20%, you save a whopping $8 a year if you spend 5 hours browsing every day.

No, this is mainly motivated by laptops, and I'd say it's even more related to thermal management than battery life. Macbooks are notorious for being very hot under high load, so naturally Apple wants a common task like browsing to be low load/heat.

[stephengillie]

Yes, that is what I meant to say.

[melling]

Hasn't Don Melton discussed in a number of podcasts that Apple places a priority on energy efficiency? Battery life is a key selling point for Apple devices. Chrome's priorities have simply been different.

[psophis]

Yes, I believe this is the episode of Debug that you are thinking of: https://overcast.fm/+I_IAnJ5Y

(Couldn't find a link on imore)

[ArtDev]

Except battery life is really bad on Apple devices. Compared to Samsung at least (my Galaxy Note4 that has a 3-day battery life).

[Steko]

This is baloney, Apple has consistently delivered great battery life on laptops, phones, tablets and beats the Note 4 in several categories:

http://www.anandtech.com/show/8613/the-samsung-galaxy-note-4...

[ksml]

My Note 4 also has a 3 day battery life. I just have to turn it off for 2 days and voila! /s

[mkozlows]

Yeah, it'd be a lot more interesting to see results for Linux and Windows, too.

The test that'd be the most interesting to me is actually Chrome on ChromeOS vs. Chrome and Firefox on Linux on the same hardware. My subjective impression from using ChromeOS is that it's much faster and more efficient than Chrome on other platforms, but I'm curious if that's really true.

[TheCowboy]

I wouldn't be surprised either, but the methodology for the testing doesn't seem to be fully explained, so it might lack rigor. Is it a fresh install of the OS? Are all of the caches and histories clean? What plugins are enabled?

Performance can vary based on how you configure your browser and what extensions you have enabled. I get less battery using Chrome, but it's configured in a way that I think is a superior experience, and most of the time my notebook is plugged in.

There might be unsurprising trade-offs between speed and battery life. (And if you get things done faster, does that offset the drain?) For me, usually time, not battery life, is the most scarce resource. Obviously I'd prefer to have more of both.

[TD-Linux]

Yes, I'd also like to know the status of Flash in particular. However, they were pretty clear about the methodology of their test - their Macbook was factory-restored, they measured power using current measurement at the DC input and integrating over time, and all of the tests were automated usage. It's certainly the best power analysis that I've seen posted to HN in a while.

I'd love to see more of this in the future - a lot of tech gets thrown around on HN for its power efficiency, (GPU acceleration, media decoders, even different CSS animation types) and it would be nice to have actual numbers.

[daenney]

A little down the article:

> We used a factory-restored MacBook Pro Retina 13” to test each website on one internet browser at a time. No programs other than the browser were open.

I'm pretty sure a factory resetted Mac is a fresh install and that it won't have any caches or history. And since they don't mention plugins I'm going to make a wild assumption here and say they didn't install any after having loaded up the browsers on the machine.

[TheCowboy]

Ah, I skipped down to "Testing Philosophy" out of habit when looking at benchmarks. It still doesn't say if the computer is rebooted fresh after each test, but I'd like give the benefit of the doubt on more obvious things like that.

[toyg]

Agree. My experience matches the test, but that's not surprising: I'm sure Office runs better on Windows, too...

[MBCook]

There is nothing magic about Safari. People have been complaining for years, the only reason for Chrome to use that much more CPU is that they clearly don't care. Whether they're not telling OS X when they're idle and can sleep or they're using inefficient older APIs or what, Google needs to fix it.

[psophis]

There is quite a bit of magic that Safari is doing. One example is that it complies javascript down to byte code in LLVM.

[MBCook]

That's not magic, Crome could do the same if Google chose.

[dgrab24]

Wouldn't surprise me if os x is hard coded to display its own apps more favorably. Brings to mind how apple hard coded the wireless signal bars to display better AT&T signal on the original iPhone to boost sales.

[wyager]

It's not as if the signal bars on a phone have any standardised meaning anyway. That's just clever UX psychology.