Linux 4.17 Offers Some Promising Power-Savings Improvements

[castlefox]

phoronix If possible please make the same item the same color in all of the different charts.
It was confusing me for a while for 4.17git changed color a couple times from chart to chart.

[Michael]

phoronix If possible please make the same item the same color in all of the different charts.
It was confusing me for a while for 4.17git changed color a couple times from chart to chart.

PTS/OpenBenchmarking.org does keep them the items the same color when within the same result file, but with this testing doing different tests on different systems, they end up in different result files and thus are not correlated.

[tessio]

That is sad news (for me at least).

Let's just hope that the good people at canonical decide to backport it to their upcoming 18.04 LTS release, providing this significant power saving to people not riding the HWE stack...

They don't need to do that. 18.04.2 will have this kernel or the next..

[asoltesz]

This is brilliant.

I am already fairly satisfied with the battery life of my Dell XPS 13 (9360) under Ubuntu 16.04.

If this improves another 10-20%, I will be happy as a lark.

[tmpdir]

Image how much impact this update will have on the worldwide power usage

Would be curious about any estimations ont this, wouldn't that be an interesting article on phoronix ^^

Also looking forward to related work like suspending USB controllers, Thunderbolt controllers, Bluetooth controllers etc. Wasn't Hans de Goede working on that?

[trek]

Image how much impact this update will have on the worldwide power usage

Would be curious about any estimations ont this, wouldn't that be an interesting article on phoronix ^^

very interesting question, but not simple to answer


how many cpus running linux are there worldwide?

I would not count desktop (less than 1%) and mobile (they are not updated), so starting from cloud computing, in the 2014 it was estimated[1] that amazon owns 2millions of server with a 27% of the market share, this mean nearly 7,5millions of server dedicated to cloud computing worldwide

assuming their cpus have similar TDP and they all runs an updated linux kernel, if the mean saving is about 15W, the worldwide power saving would be 110MW


how many watts are using the cpus running linux worldwide?

in the 2015 it was estimated[2] that amazon cloud computing was consuming 500MW with a 30% of the market share, this results in 1666MW worldwide

assuming a 8.4% of mean saving, the worldwide power saving would be 140MW

[starshipeleven]

assuming their cpus have similar TDP and they all runs an updated linux kernel, if the mean saving is about 15W, the worldwide power saving would be 110MW

For datacenters you also need to take into account the heat management systems that keep constant the temperature of the server room.

That's a big cost on the bill in most cases.

[morydris]

This is what poor Greg K.H. is always on about. For some reason, people can't wrap their head around testing and shipping up to date stable and LTS kernels, so they rely on backporting. The problem is, backporting is so labour intensive that very little of it is ever done, in proportion to the huge number of upstream patches which people would like to see on their devices.

With Android and SoC vendors, it has become so severe that Greg says vendors which are trying honestly to fix the upstream gap are five or six years away from achieving that.

This right there is why I'm weary of many of the Arm based SBC (I have many of them) - often (or always?) they need a custom kernel that always lag behind or stalls if the company dies (R.I.P. NTC). I don't think they "can't wrap their head", but it simply takes more resources than most of these small companies [can] dedicate.

[morydris]

phoronix If possible please make the same item the same color in all of the different charts.
It was confusing me for a while for 4.17git changed color a couple times from chart to chart.

On a related note (and a pet peeve #1): please please please make all agree on which side is good and which is bad. Having it change on a per-chart basis make it way harder to parse. I'd suggest that since bars grow from left to right, other to the right should indicate good. For benchmarks where you are measuring time t, this can be fixed by changing to rate (1/t).

[tmpdir]

very interesting question, but not simple to answer


how many cpus running linux are there worldwide?

I would not count desktop (less than 1%) and mobile (they are not updated), so starting from cloud computing, in the 2014 it was estimated[1] that amazon owns 2millions of server with a 27% of the market share, this mean nearly 7,5millions of server dedicated to cloud computing worldwide

assuming their cpus have similar TDP and they all runs an updated linux kernel, if the mean saving is about 15W, the worldwide power saving would be 110MW


how many watts are using the cpus running linux worldwide?

in the 2015 it was estimated[2] that amazon cloud computing was consuming 500MW with a 30% of the market share, this results in 1666MW worldwide

assuming a 8.4% of mean saving, the worldwide power saving would be 140MW

Been thinking about this several times today, finally forget about it when reading your reply...
Your numbers sound realistic. After reading your reply I googled a bit, but couldn't find concrete numbers, only seen a few articles like:

https://www.theguardian.com/environment/2017/dec/11/tsunami-of-data-could-consume-fifth-global-electricity-by-2025

Based on the figures in that article (Guessing it takes in account the factors startshipeleven mentioned), the actually number could be much higher to the degree of adding extra zero's. But that is when ballparking the ammount of linux usage (guessing it mostly is). The expected carbon emissions in that article are scary to .