Granola: Free, light, unobtrusive power-saving software for Windows and Linux

[Blue Protoman]

http://grano.la/


Thought it'd be worth checking out. Granola dynamically adjusts the power your computer uses so it draws only what it needs. You save almost 40% of CPU energy by just letting the damn thing run in the background! Your computer thus draws less energy, thus improving its friendliness towards the earth. It doesn't affect performance, either. On top of that, it tells you how much money, carbon, and power you've saved. And since it slows power consumption, it stands to reason that it would extend battery life on laptops. This is absolutely the case; people report an extra hour or so on their laptops. Did I mention that it's free? This is one of those things that you have no reason not to get.

 

[demonstar3]

thanks looks pretty cool

 

[Wrath Of Khan]

http://grano.la/


Thought it'd be worth checking out. Granola dynamically adjusts the power your computer uses so it draws only what it needs. You save almost 40% of CPU energy by just letting the damn thing run in the background! Your computer thus draws less energy, thus improving its friendliness towards the earth. It doesn't affect performance, either. On top of that, it tells you how much money, carbon, and power you've saved. And since it slows power consumption, it stands to reason that it would extend battery life on laptops. This is absolutely the case; people report an extra hour or so on their laptops. Did I mention that it's free? This is one of those things that you have no reason not to get.

Would it have to be pnd,d. Granola is the nameof a muesli cereal too. And theres a hemp variety too.

 

[DaMummy]

i have the sudden urge for hemp cereal...

 

[Wrath Of Khan]

i have the sudden urge for hemp cereal...

You and Woody Harrelson :rolleyes:

 

[Hoodoo]

Do not want. EULAs give me headaches.

 

[Coldbird]

Gave it a try... lets see if I see a miracle on my next electricity bill but I somewhat doubt it...


If it does then hell yes I will keep using it.

 

[Cloudef]

Seems quite dodgy application, I'll keep my powertop, lightweight setup, kernel settings and custom schedulers.

 

[Blue Protoman]

Gave it a try... lets see if I see a miracle on my next electricity bill but I somewhat doubt it...


If it does then hell yes I will keep using it.

The idea is long-term savings. And remember that obviously the more computers you own, the more this will benefit you.

Seems quite dodgy application, I'll keep my powertop, lightweight setup, kernel settings and custom schedulers.

I use it myself, and it's fine. On my netbook I can safely confirm the extra battery life.

 

[god_at_hell]

I do not see the advantages over normal speedstep/powernow with an ondemand governor. The governor will adjust the processor speed in respect to the system needs too.

 

[Blue Protoman]

When it's back up, check the site to see how it works.

 

[Alec]

It doesn't really explain how it works.

 

[Blue Protoman]

Really? I must be thinking of another page. Let me look a bit.


EDIT: http://grano.la/blog/how-granola-works/ and http://grano.la/blog/faq/ provide some decent information, especially the latter.

 

[god_at_hell]

Again ... I can not see the difference between this and a normal ondemand governor. Both use the same technique. The only big difference for me is the flashy "trees saved" windows. Reminds me of other software which promises something big ... like those windows tweakers (don't forget to defragment your registry every second day ) or memory boosters. Or am I missing anything big?

 

[demonstar3]

erm, i think it is a cool idea, though since i installed it my computer has crashed around 2-3x and the weird thing was I think it was in hibernation mode :\


running windows 7 professional 64 bit.

 

[Blue Protoman]

Weird.


god_at_hell: I don't know about those so I couldn't tell you. But this is free, so why not give it a shot?

 

[god_at_hell]

Ok ... you're right ... a test would be the way to go. I think I will do a test of granola on my laptop. I will sample the power consumption or energy status every minute under different conditions (movie viewing, working, surfing, compiling). The same test I will do with the standard ondemand governor. Since I am at a buisness meeting for this week, I have to wait for the weekend to do this and will report back afterwards.

 

[god_at_hell]

So, here is my test of Granola. It took a little bit longer, because I had not free time I wanted to burn for this (I had to read Robert Jordans "Eye of the World" ).


About the Test:


I wrote a simple Python program which sampled the power consumption of my notebook every 20 seconds and dumped them to a file. Sincere the laptop is a little bit special in the way it gives the power consumption, I had to cut the first 5 samples from every data set. What remains are 25 samples for each measurement. Exept for the "work/surf" test. Here there are 40 samples. Some tests I did several times. Here you will get a weighted mean.


The error stated for each test is the statistic error. The systematic error should be around +-0.3 watt and is given by the spread of means of the power consumption at idle. The different idle tests were done at different times and different battery states. Systematics come from background jobs, dependence between read out value and battery state, system temperature and more.


Tests and results:


Idle:


ondemand: (7.23 +- 0.03) W


Granola: (6.96 +- 0.04) W


watching a movie:


ondemand: (14.81 +- 0.02) W


Granola: (14.95 +- 0.02) W


transcoding a movie:


ondemand: (32.53 +- 0.09) W


Granola: (32.50 +- 0.09) W


kernel compile (-j1):


ondemand: (20.46 +- 0.03) W


Granola: (20.05 +- 0.09) W


kernel compile (-j4):


ondemand: (32.15 +- 0.04) W


Granola: (32.22 +- 0.04) W


working and surfing (no wlan):


ondemand: (12.01 +- 0.54) W


Granola: (12.57 +- 0.73) W


For the working/surfing scenario I tried to produce the same enviroment for both tests. I wrote something for the same amount of minutes, loaded the same webpages and watched the same flash movies. To get a lower statistical error in this test it would be better, if the test time is longer (eg. one hour instead of 15mins).


End result:


For the applied tests both governors perform the same (keep in mind, there is a systematic error of +-0.3 W). If there is a better power consumption for Granola, it has to be in the range of the stated errors and with this far below the 15-30% energy savings stated on the Granola webpage.


What was not tested is the reaction time of the system. The webpage states, that Granola is a fast governor, which does not decrease the system resonsiveness. Since I did not notice any negative impact of the ondemand governor in the past, I see no real gain from using Granola (but no harm either).


PS:


For the tests I checked If Granola was running like it should. The governor was reported by cpu-freq-info to be "userspace" and frequency scaling was obvious. This means Granola worked like it should.


NOTE: If you test Granola and deinstall it afterwards, you have to reinstall or rewrite the init-script which starts the ondemand governor. Granola deletes the entry on install and does not rewrite it when it is deinstalled.

 

[Alec]

Color me unsurprised :rolleyes:


They're probably after computer usage statistics, since apparently it sends back data ("By logging into your Granola account, you can track savings across systems through the Granola website").