A Look At The Windows 10 vs. Linux Power Consumption On A Dell XPS 13 Laptop

[phoronix]

Phoronix: A Look At The Windows 10 vs. Linux Power Consumption On A Dell XPS 13 Laptop

With the current-generation Dell XPS 13 XPS9370-7002SLV currently being tested at Phoronix, one of the areas I was most anxious to benchmark was the power consumption... For years it has been a problem of Linux on laptops generally leading to less battery life than on Windows, but in the past ~2+ years there has been some nice improvements within the Linux kernel and a renewed effort by developers at Red Hat and elsewhere on improving the Linux laptop battery life. Here are some initial power consumption numbers for this Dell XPS 13 under Windows 10 and then various Linux distributions...

http://www.phoronix.com/scan.php?page=news_item&px=Windows-Linux-Power-Dell-XPS

[tildearrow]

Typo:

Also visivble from the graph

Why not a MacBook? macOS is extremely good at power consumption.

[grigi]

Michael, on my skylake notebook I got some more power savings by enabling the following kernel options (added to the kernel boot parameters):

pcie_aspm=force
i915.i915_enable_rc6=1	(needed to disable VT-d for it to work)
i915.i915_enable_fbc=1	(can cause graphical glitches)
i915.i915_ldvs_downclock=1
drm.vblankoffdelay=1	(not sure if this was for a workaround?)

Also changing SATA power management options helped a lot, but I think that notebook has NVMe, so anything I did is probably not applicable.

[stingray454]

Thanks for the testing, been closing in on buying a linux laptop for a while now but the power consumption while on battery power has been something I was worrying about. Glad to see it's not a big issue anymore.

[numacross]

Michael, on my skylake notebook I got some more power savings by enabling the following kernel options (added to the kernel boot parameters):

pcie_aspm=force
i915.i915_enable_rc6=1	(needed to disable VT-d for it to work)
i915.i915_enable_fbc=1	(can cause graphical glitches)
i915.i915_ldvs_downclock=1
drm.vblankoffdelay=1	(not sure if this was for a workaround?)

Also changing SATA power management options helped a lot, but I think that notebook has NVMe, so anything I did is probably not applicable.

Some more options tested for my Skylake XPS 13 with glossy display on 4.16: i915.enable_psr=2 (1 causes glitches) and i915.fastboot=1

Loading GUC firmware works fine as well (i915.enable_guc_loading=1, i915.enable_guc_submission=1).

I also found out that disabling the built-in SD card reader in the BIOS makes a measurable difference in power draw, which probably indicates a failure in PCIe power saving features for that particular Realtek part.

[browseria]

I have a 9360 (the prior year's version of this laptop) and a Haswell 2013 Macbook Air...side-by-side doing the same tasks, they drain the battery at the same rate (within a few percent of each other). If you're not doing a lot of compiling or video conferencing, this means 8+ hrs on a full charge. Although the Macbook Air is slightly better built, all things being equal, I prefer the Dell (Fedora 27, stock, but using PowerTOP). If you were waiting for Linux to meet/exceed Mac battery life...I'd say we're pretty much there.

[JeansenVaars]

Finally! Been waiting for these kind of benchmarks for long now. So much room to expand on the details here though. One comparing Desktop Environments would be good as well. And there's no mention of any powertop/tlp optimizations.

I've been using Fedora on my 2-1 Lenovo for long (KDE) and with powertop + tlp consumption is quite low and super stable.

[treba]

Wow, this is great. So we have a common base line now and we can start to really compare energy efficiency under normal load. I guess tool-kits and common programs like browsers will be the next big things where the linux desktop has some disadvantages. Maybe Webrender and video acceleration in Firefox will help. And Wayland, when used with its full potential (there's lots of optimization potential in gnome-shell, for example usage of hardware planes, AFAIK). Audio, thanks to Pulseaudio, seems to be already in quite good shape.

[numacross]

Wow, this is great. So we have a common base line now and we can start to really compare energy efficiency under normal load. I guess tool-kits and common programs like browsers will be the next big things where the linux desktop has some disadvantages. Maybe Webrender and video acceleration in Firefox will help. And Wayland, when used with its full potential (there's lots of optimization potential in gnome-shell, for example usage of hardware planes, AFAIK). Audio, thanks to Pulseaudio, seems to be already in quite good shape.

In the case of web browsers on laptops like the XPS we really need to get on the level of... Microsoft Edge. From my tests (about half a year ago) on a 6200U it was the only browser under Win10 capable of not making the fans spin fast while watching videos, especially 60 FPS ones (on youtube I forced h264 to get HW decode in other browsers). Chrome was worse while Firefox was the worst in that regard. I suspect it's a case of Edge using native DirectX/DWM stuff while not having to use ANGLE.

I don't have Win10 on my XPS any more so I can't test the latest changes to Chrome and Firefox sadly.

[galatians]

Using powertop to tune my Dell Precision 5510 (Skylake) decreased it's idle power usage from 7w to 3w. I didn't see any mention of powertop in the article.