X200s - amazing power consumption

[lordt8]

well i did something silly yesterday, got ubuntu, installed it on a SD card, took out the HDD, boot into it to see the power draw, well it was around 10 W with wifi on, and about 8 W, with out WiFi, mind you that is before any kernel tweaking
so right now i can rule out the HDD as the power hog, am almost certain that my biggest problem is the SL9400, but then again i also think that lenovo didn't work out the bios cpu power management properly

[lordt8]

well i decided to do something else, kill lenovo power management, install cpu genie 64, now am testing cpu voltages and am going to create my own undervolted battery plan. i'll post the results before and after

[ThinkRob]

so right now i can rule out the HDD as the power hog, am almost certain that my biggest problem is the SL9400, but then again i also think that lenovo didn't work out the bios cpu power management properly

No, you can't rule out the HDD.

Ubuntu's default power management leaves a *lot* to be desired. I don't use Windows, but I'm positive that Lenovo's preload undoubtedly supports WiFi powersaving and likely supports other things such as SATA link power management, audio chipset power management, USB autosuspend, GPU powersaving, etc. Linux has support for all of those, but no distros that I know of have them all configured out of the box. That makes sense, really, considering that almost all distros try to be hardware-agnostic, whereas the default install is targeting a specific set of hardware.

Given that all of those things that I mentioned can add up to a lot of power usage, I wouldn't draw any conclusions from that test. Just my 0.02 USD.

[lordt8]

i got new clues,

i got a ssd, so did a clean install, no drivers, wwan and fingerprint were the only things without a driver, so idle power consumption was around 6.9W, with everything but the power management driver and wwan card installed i was at 6.9-7.2W idle and under 9W during wifi surfing. So i installed the wwan driver, but couldn't get the Fn+F5 working without the power management driver, but after the install power consumption rose up by 2W. Am betting that the soft on/off for the radios isn't working properly... so that's it am over investigating this topic if I find out something I'll post it here

[andykrej]

I've found out that playing an mp3 file on my X61s jumps the power consumption from 6,7W to 10-12W. This seems a bit wild to me, but it is consistent whether i change drivers or operating system (tried Win2008R2 and XP).

Can any of you with the X200s do a mp3 playback test and check your wattage?

I'm thinking it's some kind of ineficiency in the CPU or chipset.

[ThinkRob]

I'm thinking it's some kind of ineficiency in the CPU or chipset.

It's not an inefficiency. You're simply... well... using the hardware. When idle, the CPU will drop to C4 or C6 -- fairly deep sleep states. Since waking up from these states incurs a bit of latency, it won't drop into those states if it's being used (since it would have to wake back up shortly thereafter to do the requested work.) The level of usage governs what state it will scale to: for med-heavy usage it will spend most of its time in C0 (active) or C1, since it can't afford the performance hit taken by the deeper states, with light usage it will drop into C2 or even C3, C4 depending -- but it won't spend all of its time there since it still has to do some useful work.

Playing an MP3 is a fairly light task, but it's enough of one that it will wake up the CPU enough to keep it from spending all of its time in C4 or lower, thus the increased power usage.

Pardon the horrid explanation -- I'm a bit decaffeinated at the moment -- but that's the general gist of it. Executive summary: that sounds like totally normal, understandable behavior for a modern CPU.

[harrigan]

I don't have an mp3 at hand but playing a flac off the harddrive with mplayer in an xterm, my power goes up 1.3W.

(This is for an X200s. You wrote X61s, not sure if that was a typo or not).

[ThinkRob]

I don't have an mp3 at hand but playing a flac off the harddrive with mplayer in an xterm, my power goes up 1.3W.

(This is for an X200s. You wrote X61s, not sure if that was a typo or not).

FLAC is probably going to result in a bigger spike than an MP3, if only because the larger filesize will likely exceed the read-ahead size, thereby preventing the OS from simply reading the entire thing into memory prior to playback (i.e. it will keep the drive active, whereas the MP3 would simply be read all at once, allowing the drive to spin down/idle.)

[Mugrunty]

Just thought I'd add my power consumption here. I have an X200t with the multitouch screen (capacitive) and simply took a screenshot of the power manager's window. I remember a few reviews saying that the multitouch version had worse battery life than the outdoor viewable model. IIRC, they stated that it had around 10W or so...or that it only lasted 6 hours with the extended battery.

Here's the Screen shot. My Wifi was on, and the screen brightness was set to 2 bars. I got around 5.16W. Obviously, this was just idling. Even with the brightness set to 10, it till only draws about 5.70W.



EDIT: I only have the 4-cell battery

[lordt8]

i have something new, and finally a real clue,
i ran powercfg.exe -energy and well what i saw was a confirmation of my worries,

USB Suspend:USB Device not Entering Suspend
The USB device did not enter the Suspend state. Processor power management may be prevented if a USB device does not enter the Suspend state when not in use.
Device Name USB Root Hub
Host Controller ID PCI\VEN_8086&DEV_293A
Host Controller Location PCI bus 0, device 29, function 7
Device ID USB\VID_8086&PID_293A
Port Path
USB Suspend:USB Device not Entering Suspend
The USB device did not enter the Suspend state. Processor power management may be prevented if a USB device does not enter the Suspend state when not in use.
Device Name F3507g Mobile Broadband Device
Host Controller ID PCI\VEN_8086&DEV_293A
Host Controller Location PCI bus 0, device 29, function 7
Device ID USB\VID_0BDB&PID_1900
Port Path 4


just to say, the software radio switch was off for the mobile broadband! so now we have it, an error in the drivers not killing the card and sucking my battery for no good reason!

[andykrej]

Here's the Screen shot. My Wifi was on, and the screen brightness was set to 2 bars. I got around 5.16W. Obviously, this was just idling. Even with the brightness set to 10, it till only draws about 5.70W.

That's really impressive. The X200s should be able to beat that. And i see you don't even have the ULV chip. Have you made any changes to the setup to get it that low? You should theoretically get more than 16 hours idling with the 9 cell

I'm curious how the X201s compares in all this . Anyone?

[Mugrunty]

I don't think I made any changes. But, AFAIK...I can only get a 4 cell, or an 8 cell for the tablet. I'd love to have 16 hours though...lol. With the 4-cell I can squeeze almost 4 hours out of it while taking notes. I have two classes in a row totaling 3 hrs and 20 minutes. The first class I'm simply typing, the second I'm writting. When using the pen, the power consumption rises a couple of watts whenever it's in range. I'll have to do some experiements.
I can leave the second class with 20% - 30% of battery remaining. I might eventually go for the 8-cell battery.
I'm still amazed at how much cooler/quieter this laptop is compared to my T60! I could never get the T60 below 10W... and that was with the screen on zero! I guess the SSD, low voltage processor, LED backlight, and integrated graphics helps quite a bit.

Just out of curiosity, how much power does the ULV processor save? I checked the voltage of mine with CPU-Z while plugged in (set to adaptive), and it read 0.925V @ 1596MHz. That doesn't even sound like much of a reduction in speed from the 1.86GHz max.

[andykrej]

I think CPU-Z isn't reporting the true speed. I read somewhere that the speed is halfed internally in the CPU, which CPU-Z can't detect. So it's probably going 800 MHz.

Am also curious about the ULV processor. The fastest is SU9600 running at 1.6 GHz and FSB is only 800MHz. Could be interesting to hear from somebody who knows both models.

[harrigan]

I don't know what the theoretical difference should be but ThinkRob and I have very similar systems, me with an SL9600 and him with an SU9300. Based on the posts in this thread, my system idles about .2-.3 watts higher.

Actually doing work would be an interesting comparison because the SL9600 should use more juice while working but finish faster which means more time at idle over a fixed timeslice.