Microsoft Boasts of Tiny Energy Saving With IE

judgecorp writes "Microsoft has sponsored research that indicates that its Internet Explorer browser uses less power than the competition, Firefox and Google (there's no explanation of what causes the difference). However, the difference in power use is not really significant — it's about one Watt when browsing. Browsing for 20 hours at this rate, the IE user would save enough power to make a cup of tea, compared with Firefox and Chrome users. That Microsoft commissioned and published the report seems to indicate a certain desperation to Microsoft's IE marketing efforts."

Comments

It adds up

[Joce640k]

...a certain desperation to Microsoft's IE marketing efforts

Not at all. If you run a company with 10,000 PCs then it's a significant saving.

Re:It adds up

[crutchy]

What company is stupid enough to pay for 10,000 Windows licences?

Re:It adds up

[aglider]

Yes, only if all of them are just browsing the internet all the time.
But if they are making real work, maybe the results would not be that good.

Re:It adds up

[Joce640k]

Well..certainly not one that would allow their employees to have two extra cups of tea per week.

Re:It adds up

[wvmarle]

The difference is about 5% in power use of the computers only. Which may translate to 1% or less overall savings.

However IE is also slower in rendering pages, causing productivity loss (a few seconds a page of employee time eaten up) which easily costs more than the energy cost saved.

Re:It adds up

[jimshatt]

And besides that, you'd really need to calculate energy cost per rendered page. I might be able to make a browser that uses half the energy competing browsers use, but renders 2.5 times slower. But yeah, productivity loss is probably even more significant.

Re:It adds up

[LordThyGod]

Yes, only if all of them are just browsing the internet all the time. But if they are making real work, maybe the results would not be that good.

You have to consider the source here too ... its Microsoft. It was "sponsored" research, which translates to "rigged" test with rigged results. So it is indeed done for marketing purposes, or why else do it. Probably a simple web page with little css or js. You can't take anything they say at face value.

Re:It adds up

[Anonymous Coward]

And no mention of test conditions... with the actual report containing niceties like this:

In addition, at the request of Microsoft we set the JavaScript timer frequency to “conserve power” in
the Windows power options. We found, however, that the default Javascript time frequency for all
computers tested was set to “maximum performance.” We did not investigate the impact of this setting
upon browser power draw.

Re:It adds up

[DeathToBill]

No it's bloody not. Honestly people have some weird ideas about arithmetic. If your average PC uses 100W then the difference in browser power usage, 1W, is 1% of that usage. It doesn't somehow magically become 50% if you're running 10,000 PCs - it's still 1%.

But, for the same of argument, let's do the math. A company with 10,000 PCs each consuming 100W during work hours is using 1,950 MWhr (100W * 37.5 hours per week * 52 weeks per year * 10000 PCs / 1000000 W per MW = 1950) per year to power those PCs. Retail electricity is around 12.86 p/kWhr, so they're spending £250k (1950 kWhr * 12.86 p/kWhr * 1000 kW per MW / 100 p per £) on electricity to power their PCs.

A company with 10,000 PCs presumably has 10,000 employees to use those PCs. Suppose they all earn the minimum wage full time, costing £12,000 each (£6.19 per hour x 37.5 hours per week * 52 weeks = £12,070.50 per year - call it £12k). Those 10,000 employees cost the company £120 million per year.

So our company with 10,000 PCs is spending £250k on electricity and £120m on wages. But wait! All those savings will add up! Suppose those users spend every working hour browsing the web. That means they would each save 1950 Whr per year (37.5 hours per week * 52 weeks per year * 1W). Retail electricity is 12.86 p/kWhr, so each employee saves a whacking great... um... 25p per year (1950 Whr * 12.86 p/kWhr / 1000 W per kW). Yes, all those savings add up to £2,500 across the whole company. That's 0.0021% of your combined staff and electricity costs.

Now suppose you live in the real world and not all your employees work in front of a PC all the time and they only spend about 75% of their time browsing the web when they do and some of them, God forbid, take a holiday every now and then. How much do those savings add up to? Sweet. Bugger. All.

Re:It adds up

[1s44c]

You have to consider the source here too ... its Microsoft. It was "sponsored" research, which translates to "rigged" test with rigged results. So it is indeed done for marketing purposes, or why else do it. Probably a simple web page with little css or js. You can't take anything they say at face value.

I don't think they go as far as rigging the research. What I think they do is pay for thousands of very specific research topics and publish the ones that show them favorably and bury all the others.

If this is the best they could come up with they really are losing the browser war.

Re:It adds up

[Anonymous Coward]

Or an executive a £250,000 bonus for saving power!

Browser energy?

[aglider]

I'd like to know how can they tell whether the energy has been ate by the browser, the scheduler, the idle process or whatever else is in a Windows OS!!!
And I bet that IE v1 (not v10) would eat much less power as it supports a tiny slice of HTML and other web related technologies.

Disable Flash

[ninjanissan]

Now what would really save some energy on many computers would be to disable Flash. Flash commercials on some sites really waste many CPU cycles (energy). On my poor old computer it is clearly visible on the CPU load :) If you are using a laptop it will also make your battery last longer as a bonus!

On the other hand...

[gishzida]

Microsoft failed to mention the amount of power wasted cleaning up malware infections brought about because IE is not able to block malware 'mouse over' attacks. "Ad Block Plus" and "No Script" kill crapware attacks before they happen... unfortunately IE is part of the problem rather than the solution.

OS comparison

[TheP4st]

I wonder if initially it were meant to be a OS comparison but the outcome were not the one wished for so they had to settle for a browser comparison.

1 Watt is HUGE on mobile!

[bemymonkey]

As I type this in Firefox, Lenovo's Power Manager is showing power usage of about 6W. 1W less would be a 17% decrease! With the 9-cell battery currently attached, that's a 2h20m jump in battery life.

Of course, I've already dropped FIrefox's power consumption significantly using Adblock, Noscript and so on, so it's unlikely I'll see a full Watt of improvement by switching to IE, but for others, this could be huge.

Test procedures

[Anonymous Coward]

Let's look at the test procedure in the actual report...

Measure the true root-mean squared (rms) current, power, and voltage for each UUT over a six (6)-
minute period at 1Hz (averaging over 1s period) for the following test conditions:
a) Baseline: No browsers or other windows open
i) First perform a preliminary measurement of power draw in this mode for the UUT, to
ensure that the lowest suitable current range has been selected on the power meter to
maximize measurement accuracy
(1) Record the current range selected for testing the UUT
(2) Record at least 6 minutes of ‘Baseline’ UUT operation with no browsers.
(3) Move the mouse/trackpad once a minute to prevent the unit from going idle
b) Static Website Test: Three different browsers (Internet Explorer, Google Chrome, and Mozilla
Firefox) will be used. Each browser will be tested for the Top 10 U.S. websites as of March 25,
2013 (listed below, from Alexa 2013). The UUT will then be rebooted after all ten websites have
been tested. In all cases, the browser will have two ‘background tabs’ open to
cse.fraunhofer.org and cfvsolar.com, both static landing pages.
i) Each browser will be directed to the following websites, with all cookies accepted. Data
logging will begin immediately when changing the target website to capture transitional
power draw.
(1) Google.com
(2) Yahoo.com
(3) Live.com
(4) Youtube.com
(5) Facebook.com
(6) Wikipedia.org
(7) Ebay.com
(8) Amazon.com
(9) Craigslist.org
(10)Bing.com
ii) Record all power, current, and voltage measurements in a database. Each test will take
place for at least 6 minutes.
iii) Move the mouse/trackpad once a minute to prevent the unit from going idle

Notice the "at least 6 minutes" part...
So if we change sites every 6 minutes with one browser and every 30 minutes with another, that's still perfectly valid.

And then this gem:

In addition, at the request of Microsoft we set the JavaScript timer frequency to “conserve power” in
the Windows power options. We found, however, that the default Javascript time frequency for all
computers tested was set to “maximum performance.” We did not investigate the impact of this setting
upon browser power draw.