Hacker News Reader

Comment by devx | Hacker News Reader

Surface Pro actually has a worse GPU than MBA, too (Intel HD 4600 vs Intel HD 5000), which contributes to that, plus a higher clocked CPU. And we already know that Windows' battery life is terrible [1].

It's all a perfect storm of bad decisions the reason why Surface itself has bad battery life.

[1] - http://www.codinghorror.com/blog/2013/10/why-does-windows-ha...

View on HN

Windows doesn't have the best power consumption--the facts are irrefutable--it's empirically visible.

In the Surface's defense, it is running a wacom digitizer over a 1080p IPS display. Not only is it pushing 2x the pixels as the 11" MBA, but it's doing it over IPS (up to 15% greater power consumption according to Wiki).

But yes, Windows has much poorer battery life, my speculation is that since Apple targets their hardware to a few small (comparatively) number of hardware permutations, they can take this to their advantage, and power-optimize the shit out of everything.

The most effective optimizations will probably be on the driver level, which microsoft would have little say over (in terms of the surface, they should but they're weak in the hardware game).

Here's a question complaining about 1/2 battery life for ubuntu, and an answerer said they were able to match the battery life by using proprietary drivers. http://askubuntu.com/questions/158778/why-does-ubuntu-have-a...

Remember how everyone used to give Microsoft a free pass on its awful installation process (driver hell) because it supported so many options, but then Knopper (one guy!) created a Linux distro that auto-installed on almost anything. Microsoft makes tons of money on Windows and has tons of engineers, it could power-optimize for a heck of a lot of configurations. Given that a fairly small number of chipsets account for a huge proportion of PCs, I think it would be easier than we think.

Actually, that's not really true re: the GPUs. They're pretty roughly comparable, and trade off different specs (execution units vs clock speed) with the result that the HD 4600 is actually better in some benchmarks and worse in others.

See http://www.notebookcheck.net/Intel-HD-Graphics-4600.86106.0.... and http://www.notebookcheck.net/Intel-HD-Graphics-5000.91978.0....

Although, that said the Surface Pro 2 actually has an HD 4400 which is definitely inferior to the HD 5000, though the difference is not that really huge (probably roughly 20% overall), which in most cases is not enough to make a difference in playability of games - if you're getting 25fps average on the HD 4400, 30fps on the HD 5000 will be noticeably better but still bad IMO, especially in high load scenarios when the framerate dips. Personally I prefer to play with over a 60fps average so that even in more intense scenes it doesn't drop to below 30fps.

And if you're not playing games, it's really not even relevant.

I've seen it suggested that the higher EU count of the HD 5000 allows it to be clocked lower and thus operated at a lower voltage, adding up to a net power savings for the same level of performance. I imagine this could be particularly significant at moderate GPU loads, rather than fully loading the GPU with a game.

Reply on news.ycombinator.com