Linux 3.11 May Lower Intel Power Consumption

[phoronix]

Phoronix: Linux 3.11 May Lower Intel Power Consumption

It's still being investigated, but early indications are that the Linux 3.11 kernel is consuming less power at least for Intel CPUs...

http://www.phoronix.com/vr.php?view=MTQyMTM

[synaptix]

Why is it measured in milliwatts and not kilowatts?

[Yorgos]

Why is it measured in milliwatts and not kilowatts?

I won't sleep until this gets answered

[Ericg]

I won't sleep until this gets answered

Then you are gonna die tired my friend :P

[flecht]

Values are the same and considering energy usage of modern laptops, it's easier to show it in mW.

Edit: I guess, I have been trolled. (-_-)'

[jrch2k8]

Why is it measured in milliwatts and not kilowatts?

i think you mean Watts

Kilo = 1000
Mili = 1/1000

Edit: yeah me too (-_-)'

[liam]

So, it uses less power, but takes longer to complete x264 test (relative to 3.9). Perhaps integrating the curve for total energy used would be a nice idea?
What we should be seeing, especially with haswell, is more dynamic power states. I realise this was with ivb, but I'd IMAGINE that intel is looking to make the kernel work best with their latest chips.
Also need to see how the work towards overall platform power reduction is going. Hopefully enough time has passed since the new framework (around 3.6, IIRC) for querying and setting component states for at least some of the drivers to have updated.

[jakubo]

So, it uses less power, but takes longer to complete x264 test (relative to 3.9). Perhaps integrating the curve for total energy used would be a nice idea?
What we should be seeing, especially with haswell, is more dynamic power states. I realise this was with ivb, but I'd IMAGINE that intel is looking to make the kernel work best with their latest chips.
Also need to see how the work towards overall platform power reduction is going. Hopefully enough time has passed since the new framework (around 3.6, IIRC) for querying and setting component states for at least some of the drivers to have updated.

total power consumption (per task, like frames per Joule for the games and maybe per many tasks in parallel to see where the offset is) and idle consumption would be nice... Will the offset from many graphs and the idle pc be the same? and some kind of explanation what the "O" is on the "time graph". is it time for the x axis? is it the same scale for all of the lines?
i know you are still doing tests and its not gonna be christmas soon. Its just what intrigues me.

[DanL]

Why is it measured in milliwatts and not kilowatts?

Hey, it's not an AMD FX-9000 series CPU...

[synaptix]

i think you mean Watts

Kilo = 1000
Mili = 1/1000

I blame the alcohol.