x220 Idling at 4.95w - What About You?

I believe* I have finished tweaking this machine to get it to use the least possible power. On Battery in power manager it idles right at 5w. Obviously this increases with usage.

Figured this thread could be a compilation of things people are doing to reduce their power usage. Share tweaks and help others.

For me, this is what I've found to give the best battery life.


In Rig
i5-2520
Intel 310 80gb
320gb 7200 hdd
4gb RAM
IPS
W7 64 Pro

Whittled down factory image. Clean install actually did idle almost a watt higher. Vinuneuro kept telling me that, but I didn't want to believe! You can whittle down the factory install to pretty much the same as a clean install anyways. I just did a factory install then went through and gutted it with YourUninstaller. Came out to be almost the exact same size as my clean install.


My Drivers

I personally try to keep things as minimal as possible.

Bios 1.16
Hotkey 3.52.0000
Intel Chipset 9.2.0.1030
Audio 8.32.23.0
Power Management Driver 1.62
Power Manager 3.61
Intel HD Graphics 8.15.10.2342 (Intel 2361 increases power draw - higher brightness)
Intel PRO 1000 Lan 11.10.84.0
Monitor 4.36
Integrated Camera 1.1.0.1147
UltraNav Driver 15.3.8.0
ThinkVantage Fingerprint Software 5.9.4.6882
Ricoh Multi Card Reader 2.14.18.01
Intel Centrino 6205 14.00.1000


In Bios
I just disabled stuff I don't use or have on my rig like bluetooth. I didn't find any advantage power wise to disabling the other stuff. I thought there would be frankly, but nothing changed.

Ethernet - Enab
Wireless - Enab (I disable it in Network Connections. If I disable in bios my watt usage increases by 1w)
Wimax - Disa
WWAN - Disa
Bluetooth - Disa
USB Port - Enab
Ultra/HDD - Enab
Microphone - Enab
Expresscard - Enab
Card Slot - Enab
Camera - Enab
Fingerprint - Disa


In Power Manager

All default settings. Oddly enough I saw an increase in power draw if I edited any of the settings. Weird... I'm still testing that to find out exactly what is causing the increase after change. I'd like to edit the profiles to my exact specs.


MSConfig

Under services after hiding all microsoft services I have four things with checks.

ThinkPad PM Service
Lenovo Auto Scroll
Lenovo Hotkey Client Loader
On Screen Display

Under startup the only thing checked is ThinkPad Power Manager.


Services

Once I hit the desktop after boot I have 38 processes running. I pretty much used Black Vipers guide with a few changes to my personal situation. Some have said Windows 7 services make no difference on power draw... well I've found on my machine that isn't true. I dropped about a half watt by disabling services, which is a big difference in battery life.


I believe that is about it as far as battery life goes. I've probably forgotten something, but I will add to it if needed.

 

Impressive. What, if any, programs did you have open when you saw that power consumption? What display brightness? Was anything connected to USB?

John

 

Is this for 6-cell battery?

 

Must be this is the specs for 6 cell "11,1V - 5.6Ah - 63Wh" so it matches the WH on his screenshot

 

This is good information for everyone, but idle consumption isn't very useful. Not many people set their notebook aside waiting for the power saving mechanisms to kick in. Off a brand new reinstall of the factory image my machine goes down to 3-4W, but that's not realistic. I'm assuming you're using BatteryBar, what's your average power consumption while using the machine like browsing, etc? What screen brightness? Which profile, Max Battery?

In a fully functional installation mine averages 6.5-6.8W while browsing on a medium brightness without doing anything exotic other than just turning off stuff I don't use in the BIOS. I haven't sought more since at least for the time being 9+hrs is far more than I need out of the machine. I'm turning my attention to power consumption in Linux which is rather poor compared to Lenovo's highly optimized Windows setup. With vsync conquered, this is the only thing left before migrating to Ubuntu or OpenSuse.

 

That is just idling with no programs open. That is at 4 display brightness, for me is enough inside. No usb devices connected.

Correct.

Yah obviously while people use their comps they aren't going to have a use for this number. That wasn't really the point. I personally find the number pretty useful since I know what certain things add as far as wattage. So if someone is idling at 4 instead of my 5 I'd like to know what they are doing. Given that browsing on mine generally adds a watt... I know I could be browsing at 5 instead of 6. That would add a good bit more life. So if someone is idling theirs at 4, I want to know how!

In the original post... it is on Battery profile. I forgot to add my brightness though, which is at 4. Really a factory install idles at 3-4? I never saw my numbers go near that when I first did the factory install. They were at around 6.

 

I partially disagree here. I find idle consumption a very indicative number. When the CPU is in use it's in use, and there isn't all that much you can do about it. Where you gain battery time is by having idle cycles. The lower your measured wattage is at idle from eliminating processes and such, the more idle cycles you have. This applies even when you are doing miscellaneous tasks.

Example: Machine A idles at 5 watts, Machine B idles at 7 watts. Machine A has more idle cycles, so when both machines are browsing the internet or editing a word document, Machine A will continue to have more idle cycle.

Even when you are working, your machine has idle cycles, and you want to increase this count as much as possible.

Now, I said I partially disagreed, because what you said is absolutely correct if you are doing something that eliminates ALL idle cycles (ie: rendering, video, etc....).

 

I haven't found much correlation between idle consumption and when the machine is in light use and one core is likely parked, like during browsing. When the installation is brand new the idle consumption is extremely low and goes up at least 1-1.5W just from installing standard applications. However, average consumption is always very similar. BatteryBar is useful for checking this since it isn't as volatile as other meters.

 

A great way to calculate power consumption is with a Kill A Watt. That way you could get idle and load power levels. Anyone want to be a guinea pig?

 

Hmm, I didn't see that at all. Right after the fresh factory install (nothing new installed) it was idling higher than it is now. Ha, only reason I know that is because I monitored it very closely. It was around 5.90-6.10, quite odd.

That sounded like you volunteering

 

I agree that's it's useful to know how low one get. I could only get down to 6.25W on my T420s and that was with the backlight on an unusable minimum brightness. Assuming that people are all achieving similar CPU utilisation then the main differences in idle power consumption are likely to be (i) differences in the CPU idle power (ii) HDD / SSD power; (iii) number of RAM modules; (iv) whether wireless devices are disabled. Interestingly, my idle power consumption went up when I enabled the Lenovo battery stretch function.

The trouble with that approach is knowing how much power is lost at the PSU. I've got one of those meters and a quick check showed about 11W measured by the meter and when I switch over to battery it dropped to about 8W. You can also measure power consumption under load by running off battery with a power profile without power savings implemented. HWiNFO32 also makes a fairly good estimate of the CPU power comsumption (switch on the Sensors).

John

 

Batteries will definitely skew the results. I have no idea how power efficient those bricks are. If anything, it might be interesting to run with both a Kill A Watt and check hwinfo results to compare.
I see what you're saying though, power consumption measured at the power cord versus at the processor. I just question the accuracy of software and sensors to measure it accurately.

 

wow, sigh. I guess I'm still getting about 6 hours and 20 minutes of battery. I dont know how much more I want to screw around with my install.


Ok, I just did this. Here is my factory install, NO optimization, no messing with startup services, no bios disabling.

i7
8gb ram (this prob adds like another .3 watts or so)
4 brightness (lenovo drivers)
intel 320 SSD
WIFI ON!

As you can see in batterybar, I'm discharging at about 5.2watts.

I'd like to know how some are getting more than 7 hours of normal usage. I mean, 5 tabs open, watching flash video, instant messaging clients open, etc etc etc. I can easily get 6 hours, doing all of those things, USING CHROME.

And, I use the brightness usually at at least 6 (lenovo drivers).

 

Do you have an mSata/SSD in that? Or just HDD. When I just had a HDD in my x220 I found it idled around 6.5 or so. That was also on a clean install so I assume it would be a little lower on the factory (from what I've seen at least).

Isn't that funny? Mine jumps up almost 4 watts when I activate battery stretch. It still claims it will add 30 minutes to my battery before I activate it. After activating it I watch my battery dive a few hours.

Which is pretty damn good frankly. Ha, this thread wasn't meant to frustrate!

Do you find batterybar to be higher or lower estimating the drain than power manager? So 5.18 in BB, what is it in PM? Just curious.

Could probably squeeze out more by disabling services. I found disabling things like security center, updates and a few others dropped my idle, YMMV. Since the 310 and 320 are very similar in watt usage we shouldn't be far off from each other. You have 8gb vs my 4gb and you have an i7 vs my i5. We are pretty close.

Have you tested firefox 3 vs chrome? Vin suggested switching back to firefox 3 and I found it to use almost a whole watt less than ff4.

 

Well, there's your problem.

 

Doesn't the latest version of Chrome (13.0.782.107 beta-m) use less watts? Maybe Floz is using it.

 

I've got the Intel 320 on board. I've not gone out of my way to maximise power savings. I know that, from experience, working offline in Word with a usable display brightness my T420s is using 8 to 9W and I can get at least 8 hours with main + bay batteries.

John

 

Yes I am. It does appear to use less cpu.

By my own testing, firefox4 is a greater battery hog than chrome 13 beta. At least in my mind, ie9 is not an option for me.

I think the battery circuitry has provisions to measure the actual discharge rate of the battery, this is the figure that batterybar is providing. I find that the lenovo power manager tends to estimate higher.

As for firefox3. It was firefox3 that caused me to migrate to chrome, I won't go back to firefox3, but to maybe firefox 6 or 7, when I can have clear evidence of cpu usage reduction.

 

Just my 2 cents on browsers. Firefox, not matter which version, has always been the most responsive web browser for me. Chrome on my desktop is crapping out with white blocks all over the place when I scroll or play a Flash video inside the window. Not to mention, my computer feels like it is running Linpack when Chrome is open.

 

LOL, I don't know about that.......

 

Try Firefox Aurora, or for the bleeding edge, Nightly. Nightly has occasional issues (Session Manager addon broke for two weeks, but was fixed). Surprisingly uncrashy for it's nightly status.

Firefox 4 and 5 are still pretty bad in terms of CPU usage.

IE9 is really good this time around, except for the interface.

For more information: http://blogs.msdn.com/b/ie/archive/...ng-the-industry-with-internet-explorer-9.aspx

Just looking at the Chrome graphs make me go WHAT IS THIS I DON'T EVEN.

 

That was from chrome10. As I've mentioned in other places here, chrome13 beta appears to have improved power consumption.

Microsoft's test was concluded from 3 tests. Idle at a blank page, a news site, and an html5 test. I'll note also, that the html5 test was provided by microsoft.... no wonder their browser scored so well on that test.

Finally, chrome continues to be one of the fastest web browsers overall, and it appears that you will trade some power consumption for speed.

 

Chrome will be Chrome.

Not really? Firefox 8 and SRWare Iron (a Big Brother free Chromium build) are pretty much equal in terms of speed for me. IE9 seems to be a smidge faster, but at this speed, does it really matter?

I'm personally indifferent to the HTML5 test. The News figure is more relevant.

 

http://www.betanews.com/article/Bro...-fastest-most-standards-compliant-benchmarks/

Chrome12 wins. 3rd party testing.

 

What browser you use is going to have minimal impact on your battery life. The types of sites you visit have a much greater impact. For example, if you visit sites that rely heavily on flash, those will burn much more battery than a simple forum like this.

On topic, I have no idea how you get your prcoesses down to the 30s on a laptop. My desktop gets around low 40s on boot and its pretty bare bone. Do you not use any of the thinkpad utilities?

I don't own a x220, but I do have a t420 and on boot I get 65 processes on boot and around 7.8watts on idle, but obviously its higher b/c of the larger screen.

 

Wooooo synthetic JavaScript benchmarks.
Jagermonkey outdoes V8 in one notable example:
Run Linux on... JavaScript?!
"Oddly enough, he found that Linux ran twice as fast on Jaeger Monkey as it did on V8. He's not sure why."

Better yet, go NoScript and not run unnecessary JavaScript at all. AdBlock Plus + NoScript = Efficient. Add Ghostery for privacy, but with negligible performance impact.

As for cutting power consumption, avoiding third-party programs that are less than thrifty with CPU cycles helps a lot. Same with avoiding programs that push Platform Timer Resolution down (use powercfg -energy to check; Chrome, Skype, and SRS HD Audio Labs all do this AFAIK). Optimizing disk access by using a good defragger like UltimateDefrag would help with a platter drive too.

 

I actually only use Power Manager and Fingerprint Reader. Those are the only TV tools I use since I have other tools that do the exact same things as the others. I try to keep things as minimal as possible. So using windows network connections instead of thinkvantage is what I strive for. The only reason I use PM is because it seems to have functionality that the windows one does not (battery threshold etc).

I do disable like windows update, security center, defender, all the homegroup/home network garbage. I just don't use any of that stuff. I don't find it useful (for me). Ha, on xp I had it down to 13 services at boot. I was quite disappointed with the 38 on 7.

I have all the functionality that I personally use with the trimmed down services. I mainly used Black Vipers guide, with a few changes here and there.

 

Wait, you do use some type of anti-virus software, right? I mean, riding bare is fun and all, but...

 

I submit this article for everyone's inspection. I found it to be VERY interesting, and provided greater insight into all this.

Chrome: Cranking Up The Clock Mike's Lookout

Also, see comment 13. I think some of the info provided in the article is a little outdated, in regards to how chrome functions today.

 

ClockRes 2.0
http://technet.microsoft.com/en-us/sysinternals/bb897568

This little program can be used to show you the minimum clock resolution that Windows is using. When I run Google Chrome, it sets this minimum to 1 ms which can interfere with the amount of time a CPU can spend in the power saving C States.

You will need to open up a command line window to run ClockRes. Run it before and after you start Chrome or any other browser and you will be able to see if it is adjusting your system timer. I haven't used IE for a long time but I don't think it does this to the system timer so you might get better battery performance if you can find a program that doesn't change this.



Here's a little program I wrote so you can monitor the percentage of time your Core i CPU spends in various C States.
Improving C State residency time can help reduce power consumption and improve battery run time.

CState 1.2
http://www.mediafire.com/?axqc0138wxxwv9h

 

In all my years of owning a computer I've never had anti-virus. Call me crazy or lucky, but I've never had a problem. I do random checks with Malwarebytes and Superantispyware, but that is about it. My buddy always tells me to get some active protection, but I just don't feel the need. I do backups pretty much every day since my work revolves around my computer. So really if I ever had a big issue I'd just wipe the drive and insert my clean backup. I like to live dangerously!

 

I understand totally. I've never had a virus either. But, microsoft security essentials is free and helps prevent "drive by" infections, which are becoming increasingly more prevalent.

 

Yes, I find that both chrome13beta and the googletalk desktop client are both changing the system timer to 1ms.

On your CState utility, I want the cpu to be at the c7 state for as much as possible, right?

Much respect, and many thanks.

 

IE9 actually adds an entry to Windows' advanced Power Options for Javascript Timer Resolution Frequency on AC and on Battery, where it is intuitively set to maximum battery life on battery.

 

Is there a way of enforcing the resolution? IE: Force 15.6ms resolution on battery?

 

I may check it out

Watch me come back in here after all these years and I get infected!

 

Since that would detract from the performance and experience of Chrome, Google's not going to let you do that so easily. Apple-style.

 

floz23: Generally speaking, the higher the C State number, the greater the power savings will be so a CPU in C7 should consume less power than the same CPU in C3 or C6.

I was at my desktop when checking out Google Chrome. I'm not sure if it is smart enough to increase the timer resolution when on battery power.

That is up to each program you use to include this option. I like Chrome but I don't know if it has this option.

If anyone wants to do some testing, it would be interesting to see the C State program I posted with and without Google Chrome running. You would need to find a very static web page when testing, preferably text only, so that doesn't interfere with your results. I did some testing a long time ago but it wasn't on a Sandy Bridge CPU. It did interfere with C State residency time but maybe Google thought it was worth it for increased performance.

 

Mine uses 7-9W idle.

Not sure how to get it lower..

Disabled startup items and things I dont use from BIOS.

Using a Crucial M4 SSD with i5

 

has anybody tested power consumption with linux?

 

Got it down to 6W idle, wireless off, 4 brightness.

I'll play around with services / processes. Maybe uninstall some stuff.

 

See the Linux on X220 thread.

Mine idles down to 7.2W with brightness 9 and wifi on (i7, 8 GB RAM, Intel 320 SSD, kernel 3.0).

Note though that this power draw estimate is inaccurate. I'm guessing most values quoted in this thread are overly optimistic. The battery reports the "immediate" power draw and I'm guessing that's what most tools display. A more realistic estimate can be obtained, however, by sampling the decreasing remaining battery charge over a period of time. That's what PowerTop does under Linux if you let it run for a few minutes, and the "long term average" estimate I get then is around 8.8-9.0W.

 

Hmm, for me the default power setting in Control Panel was Windows High performance plan. None of the Lenovo plans, listed underneath as additional plans (or indeed in Power Manager 3) were in use. Not sure why mine would be different to yours if we both used recovery disks of the recovery partition? Anyhow, I've changed over to Power source optimised now.

Under Idle timers, do you set the HDD to stop rotating after 30 seconds for both battery and AC?

 

Which PM are you using?

You are accessing those plans through control panel and not power manager? All power plans I use are accessed through power manager. For this post I used the Max Battery profile in PM. I renamed the profile to Battery so that is what I'm referencing.

Every setting in the individual plans is set to default. I'm usually always on Battery setting when unplugged and Performance when plugged. So whatever the HDD idle timers are on those, that is what I use. I think it is 30sec/never.

 

Okay, I understand you now- just a bit of ambiguity in the wording in your first post. Yes, I'm accessing my plans through Power Manager of course.

I just stay in Power source optimised permanently because I'm mainly at home and when I go out, I would probably forget to change the plans and am not out for more than a few hours anyway.

Must admit, I'm at a loss to know what programs and drivers to delete (as per your first post)? The only thing I've deleted so far has been Microsoft Office!

The safe to remove services on Black Viper, sure that's easy enough.

 

Ok good stuff.

Yes, I pretty much shaved it down to my first clean install. Found about 10 extra drivers installed on the factory install vs clean. I don't know if those helped with battery life or what? You don't have to trim down the install if you don't want. Some of the things I deleted were really necessary to delete. More of a hd space/ocd type of thing, but I like to say it was for the space...

The drivers on the first page were ones that I found I needed, I just uninstalled all the rest. The only ones I kept from the factory that weren't installed on my clean were the 8 "Windows Driver Package ..." drivers and the "Registry Patch to Enable Max PS".

 

Good stuff, man. Thanks

 

geez, lowest I can get it down to is 7-8w idle

even after doing most of your tweaks..and matching drivers.

 

Is your processor an i3-2310? I think they might run slightly warmer than the i5's (based on experience of both).

 

I've got the i7

any chance you could list what services you disabled/manualed off black's list?