The 15 hours of battery was only for the FHD (1920x1080) display. I scanned The Verge review but wasn't clear if they tested battery life on the FHD or QHD+ display.

Toss up some terminus and vim and the FHD model sounds like a great all day coding machine.

On my Samsung ATIV (QHD+), I found that running the display at half the native resolution dramatically increased battery life and didn't seem to look any worse than a normal panel with a ~72 DPI native resolution.

How very odd... is it integrated graphics? I wonder if it's the display or the gpu/(v(ram)/cpu that makes consumes less resources.

Well....1920x1080@3 bytes per pixel@60Hz = 336MB/s, while 3200x1800@3 bytes per pixel@60Hz = 989MB/s. In QHD mode your laptop is sending nearly a gigabyte of data to the display...per second. It might not be very power hungry to compute those pixels with a modern GPU, but even simply sending that much data requires a lot more power than sending "only" 330MB/s.

I'm not so sure. Given this random google hit (on 10gbps ethernet nics) -- I'm not sure it makes much difference: http://www.cl.cam.ac.uk/~acr31/pubs/sohan-10gbpower.pdf

I couldn't find anything on hdmi/4k, but this seems to suggest it doesn't make much difference (compared to what I assume the backlight and screen draws): http://en.wikipedia.org/wiki/Thunderbolt_(interface)

At any rate, if we assume it takes 4 times as much power, and we know that phones do just fine with 1080p+ displays -- I doubt the signal is the problem. Maybe it's the RAM used for the videoframes?

I always assumed it was the screen that consumed power with HDPI displays (flipping more pixels) -- but maybe it's something else.

I was always under the impression that the backlight is a lot of the power that displays consume, which is why LED displays use so much less power than CCFC displays. That would remain constant with resolution, right? If that's the case, I'd think that the actual GPU work (or the RAM, like you say) is where the power difference was for me.

I could be wrong about how significant the backlight is on an LED display though.

I also thought that modern video cards (as in from the last 15-20 years) optimized 2D graphics so that regular desktop use didn't require pushing a screen full of pixels to the GPU on every frame. That would seem to diminish the significance of the actual number of pixels at QHD+.

I don't have a good answer.

That's a clever battery hack. I'll be using it on my sp3. Nice.

The Verge and most other reviewers simply run the tests with the display at some percent brightness - I forget what the Verge uses, but I think it's 50% or 65%.

This is a fairly useless metric. Since different displays have a different max brightness (and perhaps even different brightness curves, I feel like the "percents" aren't always linear), testing at a certain brightness value is the only way to go.

Thankfully AnandTech uses 200 nits for all laptops so we can actually compare these numbers.

Apparently simply disabling Windows Indexing increased battery life from 7.5 hours to 9.5 hours in some tests.

If you have a Pro or Enterprise edition of Windows, you can configure this:

1) Launch mmc (Win+R, "mmc")

2) File -> Add/Remove Snap-In

3) Group Policy Object Editor -> Add -> Local Computer -> Ok

4) Expand Local Computer Policy -> Computer Configuration -> Administrative Templates -> Windows Components -> Search

5) Find "Prevent indexing when running on battery power to conserve energy"

6) Enable -> OK -> Close

If you're on Home Edition then make a new *.reg file with the following content:

     Windows Registry Editor Version 5.00

     [HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows\Windows Search]
     "PreventIndexOnBattery"=dword:00000001

And run it.

You can change steps 1-3 to just "Win+R, gpmc.msc"... which will take you straight to the group policy editor.

You have to install Group Policy Tools for Windows in order to have gpmc.msc on Windows 7 work from the run box.

Going through MMC works even on virgin Windows 7 pro/ent without installing anything.

1-2 steps, Windows key, type gpmc.msc and it will be the first result in the prog search. I think this assumes the index is enabled, but since we are here ...

I'm giving this a try, and I'm having trouble following step 4. Once I add the Group Policy Object Editor to the right side of the GUI, I can click on it and get a pop-up, but I can't seem to find the expansion option.

Are you still in the "Add and Remove Snap-Ins" dialog by any chance? If you click ok on that dialog after having added the snap-in, it should appear in the main mmc UI on the left hand side.

https://i.imgur.com/ZHzEDrC.png

I've updated the original instructions a little to be more clear.

I feel very silly -- thank you very much for the help!

Do you have a link to the tests or the data?

I work on the indexer and am always looking for more data about what we are doing wrong.

Check out the top comment of this article: http://www.forbes.com/sites/jasonevangelho/2015/01/26/dell-x...

A commenter pointed out to the author that Windows Indexing was responsible for significant battery drain and the author confirmed by running the test again with it disabled.

Thanks, not great to see, but good to know what we have to fix

Perhaps a sane default would be to pause Windows Indexing by default when on battery? Defaults matter.

That's not really ideal though... for example, Windows Indexing is responsible for indexing email in Outlook. If you've been mobile for the last two days, that's potentially a lot of email you can't search.

Reminds me of a somewhat unrelated scenario in Android world. Auto backup of photos in Google+ Photos has a setting where you can have backup only happen when connected to power (and/or on Wi-Fi). Perhaps Microsoft could add something like this in their settings as well?

Yes, I understand this could be seen as feature creep but I think if the savings in battery life are as much as the grandfather says then it is worth the extra complication of a setting. What do you think?

Laptops can't really stay mobile for 2 days. At some point, it needs to power up, and your indexer can do the deed then. Too bad Windows doesn't have something like PowerNap that allows the system to run services while the clamshell is closed.

How often do users search in Outlook? I'd say maybe 5-10 times a day, and if it's already indexed, I won't notice.

Ars found their data to be roughly in line with Dell's claim as well, though they only tested the high-res notebook and didn't check the FHD one:

> Dell claims that the battery life of the XPS 13 is an impressive 15 hours for the 1920×1080 model and 11 hours for the 3200×1800 version. Our own battery testing is a little more punishing than Dell's, but the company's numbers seem to be in the right ballpark. In our test of light Web browsing, the XPS 13 came in just shy of nine hours

Usage pattern is definitely a huge factor though:

> in a more intensive WebGL test, it hit five hours on the dot.

and so is software running (using Chrome is a terrible idea if you want battery life, for instance)

Did you mean Anand, or did Ars do their own test?

If you meant Anand, they tested both. Look for "Dell XPS 13 (9343) QHD+" and "Dell XPS 13 (9343) FHD" values in the charts.

He's talking about the Ars review: http://arstechnica.com/gadgets/2015/02/review-the-dell-xps-1...

I have never met a laptop review where dividing by 3 is not a more credible estimation of actual battery time.

TheVerge is more like a lifestyle blog than an actual tech blog. They have this belief that Apple hardware is the best and any review - or any post really - will reference Apple if they get a chance. And if they don't get that chance they will work hard to create it. For a very specific cult of people thadt is OK, but it wouldn't work for an independent tech blog.

Isn't that because Apple is by far the biggest laptop manufacturer? Comparisons with the market leader are often useful.

I didn't think they were? I thought they had a very niche market and are outsold massively by Windows laptops? Unless that position has changed? I couldn't find any figures on the Internet (admittedly I didn't do an exhaustive search); someone else might do better.

Though the analysis in this article I'm linking is disingenuous ( it compares Apple laptop+iphone+ipad+ipod sales to Windows laptop sales...) its graphs tell the story pretty clearly: http://ben-evans.com/benedictevans/2014/2/12/apple-passes-mi...

Look at the little sliver of black that represents Mac computer sales (looks like ~5m sales/quarter) and compare with the orange line that represents Windows computer sales (just under 80m sales/quarter).

Even if 100% of the Macs are laptops and only 10% (as a pessimistic guess) of the Windows machines are laptops, sales are still heavily in the Windows territory.