Why does the T60p get such poor battery life?

[darrenf]

Tabook clearly states that the non-P T60 models get

6-cell: up to 6.0 hr; 9-cell: up to 9.0 hr; 9-cell plus Ultrabay battery: up to 11.5 hr

while the P models get

9-cell: up to 5.4 hr; 9-cell plus Ultrabay battery: up to 6.9 hr

When I read this before ordering, I was sure that it was a typo because it didn't make sense. Now that I have my T60p, I see that even 5.4 hours is optimistic. My T60p with 14" screen is lucky to get 4.5h hours with a 9-cell.

Where is the juice going??? The tabook makes no distinction in power consumption between 14" and 15" models (although it does use the all-encompassing "up to" prefix). It also doesn't distinguish between power consumtion at 1.66, 1.83 or 2.0GHz, single or dual core and it doesn't distinguish between power consumption of the GMA950, x1300 and x1400 video cards.

So why the big jump when the processor goes to 2.16GHz and the video controller is a V5200?

All the processors have a speed floor speed of 6x which is 1GHz on the 2.16GHz. Slower processors can therefore clock down a bit slower but I doubt that this accounts for much. Heck, I locked my processor at 1GHz and then at 2.16GHz and only noticed about 0.7watt difference in power draw at rest! Apparently the more subtle internal power saving states have more impact on consumption than they used to and the gross speed is not as large a factor as it once was.

So that leaves me to believe that the GPU is the culprit. I wondered if perhaps the V5200 didn't implement PowerPlay so I set PowerPlay on and ran a game. I can confirm that it sucked (about 1FPS ) so PowerPlay certainly seems to work. Still, the GPU is my guess as to why the power drain is so severe. It almost argues for having both the GMA and V5200 on the same computer and switching the V5200 on when needed.

So, is there a ThinkPad engineer out there who can tell me why the p-model is so power inefficient? Barring that, can anyone else share your details on power consumption? I'm using MobileMeter -- be sure to set CPU frequency method to "From Driver" (I think "From OS" works too) as "Measure Actual" always shows 2.20 for some reason

My 14" T60p model 2007-83U uses:

14.5W - CPU locked at 1GHz, HDD awake, PowerPlay on (no color depth or frequency reduction), brightness at minimum, Catalyst Control Center uninstalled (just in case), 1400x1050x32-bit, while writing this post.

15.3W - Same as above, CPU locked at 2.16GHz

20W - Same as above but at max brightness

BTW: I ran some tests with the second core turned off and the power usage was the same (or very close). One core stayed cool througout (27C I think) confirming that it was off.

-darren

[JHEM]

Where is the juice going???

Out the bottom drain hole!

Regards,

James

(Sorry, I couldn't resist!)

[sidn]

Did you try to reduce the GPU/memory-frequency to minimum?

[darrenf]

Did you try to reduce the GPU/memory-frequency to minimum?

No, that's a good idea. I just installed ATI Tools and it shows core and mem freqs of 0 (!!) -- I wonder if it doesn't work with this chipset.

-darren

[AssPenny]

Its worse because of the Video card first, then screen resolution next (uxga). Screen rez isnt much, but it does contribute.

[darrenf]

Its worse because of the Video card first, then screen resolution next (uxga). Screen rez isnt much, but it does contribute.

Agreed, but my basis for comparison is a T42p with the same screen res (sxga+)

-darren

[Sottozero]

Two reasons:

1. The GPU is very powerful, and thus draws more power when asked to perform at its highest levels.

2. The screen (at least the 15" FlexView model, 1600 x 1200). On any notebook, the screen is the element that draws the most power. The higher density and brightness, the more power it draws.

[darrenf]

Two reasons:

1. The GPU is very powerful, and thus draws more power when asked to perform at its highest levels.

2. The screen (at least the 15" FlexView model, 1600 x 1200). On any notebook, the screen is the element that draws the most power. The higher density and brightness, the more power it draws.

I appreciate the help, but please read my post

1. This is at idle with PowerPlay turned on.

2. The screen should be similar in power consumption to my T42p as both are 14" SXGA+

The difference in battery life between the T42 and T42p (utilizing the exact same display) was nothing like this. Tabook says 6.4hrs for the T42p and 7hrs for the T42 with identical 14" SXGA+ screen and 9-cell battery. That's a drop of 8.5%.

The stated difference betweed the T60 and T60p is 40%!!!

-darrenu

[darrenf]

Out the bottom drain hole!

Nope - I checked and the same amount of power is coming out the drain holes on the non-p T60.



-darren

[christopher_wolf]

Sounds like a power drian issue; I seem to recall that thise was predicted.

It was discovered that Windows XP bleeds power away through the USB on the new Intel Duo Core platforms. Do a search for it and you will find it on here.

There is also a solution for it in the original article.

Here is one of them; See

http://forum.thinkpads.com/viewtopic.php?t=19943

And here is a solution to that problem as well as a potential solution to yours; See

http://forum.thinkpads.com/viewtopic.php?t=20670

HTH

[darrenf]

I ruled that out too. The only internal USB device in the T60 is the fingerprint reader. I disable that but there was no improvement. I disabled the USB controller and still no change. I wasn't surprised because the original Anandtech article said that this wouldn't correct the problem. Next I installed Microsoft's "patch" with no improvement. Lastly, I disconnected the fingerprint reader from the ribon cable that connects it to the motherboard.

I re-enabled the USB controller that the fingerprint reader is connected through and indeed the device has disappeared, but still no change in power consumption.

Clearly Lenovo knows the cause because they factored it into the tabook battery life estimates. That's the most disappointing part -- if it was fixable I think Lenovo would have done so before releasing the laptop.

-darren

[sugo]

My T42p with 14" screen is lucky to get 4.5h hours with a 9-cell.

I am confused. I thought you said you could get 5+ hours on T60p and you implied T42p has longer battery life.

On my T42 (ati 9600), power drain is about 10.0W with minimum LCD brightness, WiFi off.

[darrenf]

My T42p with 14" screen is lucky to get 4.5h hours with a 9-cell.

I am confused. I thought you said you could get 5+ hours on T60p and you implied T42p has longer battery life.

On my T42 (ati 9600), power drain is about 10.0W with minimum LCD brightness, WiFi off.

Typo - thanks for catching that. I fixed it with an edit. I was referring to the T60p.

My experience was similar to your on the T42p. That's why I find the T60p power consumption so disappointing. I had hoped that the Core Duo and 20% better battery capacity would offer an improvement in battery life.

-darren

[darrenf]

2. The screen (at least the 15" FlexView model, 1600 x 1200). On any notebook, the screen is the element that draws the most power. The higher density and brightness, the more power it draws.

One more benchmark: With the screen off the machine pulls 9.5W. If I could only use the machine without video I could run for almost 9 hours!!

-darren

[astro]

Have I missed something, or have you not stated a comparison between the 6-cell and 9-cell? I presume you have the 6-cell as well, because that is the standard battery and the 9-cell was purchased extra? Is that how Lenovo works?

My point is, how do you know that the battery isn't faulty?

I think you really need to compare with other people's experience first instead of all this speculation. That said, the speculation will obviously continue, in which case I think the next thing you should consider is how many Watt hours the battery supports. You should be able to infer battery life from that and your stated power consumption.

[darrenf]

Astro, the p-models come with 9-cell batteries only. Since this is my first T60, I don't have a 6-cell to compare it to. I have 4 more T60's on order but they are all -p models. I will compare 15" to 14" when some more come in to see if there is an appreciable difference.

I think that I'm not being clear in my concern because it's really not about my specific copy of the T60p. *Lenovo* states in the tabook that the p-model takes 40% more juice than the non-p model. My experience bears that out, but my question is, what about the p-model takes so much more power? One would presume the video card or processor because those are the only two items not found in any non-p model, but it seems absurd that either would take *that* much more power.

The processor is only 8% faster than the 2GHz. Even so, I believe that since all the processors in this family have a 667FSB and the same power draw (31W - I guess that's running at 100%), they all have a minimum clock rate of 1GHz (can someone with a 1.66, 1.83 or 2GHz Core Duo confirm)? There should, therefore, be no difference in power draw between the -p and non-p processors in their most power efficient state (which is where the ideal battery life figures are calculated)

Unless I'm overlooking something, that leaves only the video chip. Why on earth would the V5200 be that much of a pig when it's just doing 2D work in PowerPlay mode?

Again, I'm hoping to get a reply from someone with inside info from the Lenovo engineering team (we see them post here from time to time). I would also love to know what power consumption rates non-p users are getting, since most of the figures posted in other threads have been generalizations up to this point.

For the record, my battery (Sanyo) has a design capacity of 84.24Wh and a full charge capacity of 86.17Wh. This is in line with the time estimate that Power Manager is giving me based on the consumption rate shown in Mobile Meter which increases my confidence in Mobile Meter's reading.

-darren

[darrenf]

A little more info on the V5200. According to THIS article, the V5200 is supposed to usher in a new era in power savings, specificially:

ATI's solution is to borrow some of the ideas used by processor manufacturers in their notebook lines. ATI Technologies has introduced a new BIOS that drops the clock speed down when 3D applications are not being used.

The company's engineers have also built in dynamic clock gating technology, which shuts down unused parts of graphics processor.

The first part sounds like fluff since that's what PowerPlay has always done. The second part sounds fantastic, but that makes me wonder even further where the juice is going.



-darren

[darrenf]

Did you try to reduce the GPU/memory-frequency to minimum?

I did some more research on this. ATI Tools and ATI Tray Tools don't recognize the V5200 as a supported chipset. I installed the Omega drivers and was able to run the Omega verion of Tray Tools. It supported most features, but not overclocking.

So, I removed the omega drivers and reinstalled the stock drivers with CCC.

This is where it gets interesting. The only place to monitor clock rate is in under "Information Center"->"Graphics Hardware" in CCC. Whether on AC or battery it shows a clock rate of 398MHz and a memory clock of 324MHz. I changed the powerplay settings to "Optimal Battery Life" for all three modes (AC, DC high and DC low) but the speed still reads the same. Next, I enabled color depth reduction and refresh rate reduction in the PowerPlay page. That page *shows* that the active settings are "Optimal Battery Life" but the clock rate is unchanged, the refresh rate is still at 60Hz and the color depth is still 32-bit.

I could understand that the info page might show the standard operating frewquencies even if they were reduced at the moment for power saving, but would the color depth and refresh rate change not be reflected in display properties?? Interesting..

On an unrelated note, Bluetooth draws 3Watts.

-darren

[darrenf]

Looks like I was wrong about PowerPlay -- it is working. I found another tool called HWiNFO32 (great proggie!) that shows the GPU freqs among other things. It shows the core at 209.3MHz and the memory at 135MHz.

Interestingly, it shows that the processor doesn't support the C1E Enhanced Halt State (or C4E). I don't know what C4E is, but I thought that C1E was the most gross adjustment (cutting the multiplier to 6x in this case) which appears to be working.

Odd..

-darren

[christopher_wolf]

Looks like I was wrong about PowerPlay -- it is working. I found another tool called HWiNFO32 (great proggie!) that shows the GPU freqs among other things. It shows the core at 209.3MHz and the memory at 135MHz.

Interestingly, it shows that the processor doesn't support the C1E Enhanced Halt State (or C4E). I don't know what C4E is, but I thought that C1E was the most gross adjustment (cutting the multiplier to 6x in this case) which appears to be working.

Odd..

-darren

C1E is the state that replaced the older C1 state in Intel processors; basically, whenever the OS traps a halt instruction, the processor goes into a halt state where no logic is performed; however, this isn't common for most applications to trigger.

The only reason why this is "Enhanced" is because it only solves half of the problem; that is, it only addresses the current draw of the processor and not the voltage and thereby halve the power consumption if voltage is constant. The enhanced part is where the C1E state comes in as it also underclocks the processor to the minimum level it can go to, and *then* it gets undervolted, for a short period of time, to maintain stability as it is forced by various control circuitry and logic to do this in the firmware. C4E is much the same.

If I recall correctly, Intel was stating that the C1E state gets you a far better drop in power consumption during a an idle/halt period of time rather than just the C1 state since you are addressing both the current draw and the voltage at the same time...Thereby reducing the power far more than you could if you just addressed either the current or the voltage.

Also, from what I have heard, there have been reports of similar issues on MacBook Pros. See;

http://www.red-sweater.com/blog/105/all ... acbook-pro

Note Cryptonome's Post as I suspect a similar thing might be occuring with the chipset in your T60p although to a far lesser extent

Also note that the C4E state is triggered when the system is unplugged; in certain cases, the fan and controller respond to this and shut off most of the time. Which I have noticed on various Thinkpads I have seen and heard.

[sugo]

On an unrelated note, Bluetooth draws 3Watts.

Is that active or idle? If you turn off bluetooth, does it reduce 14.5W to 11.5W at minimum screen brightness?

[darrenf]

On an unrelated note, Bluetooth draws 3Watts.

Is that active or idle? If you turn off bluetooth, does it reduce 14.5W to 11.5W at minimum screen brightness?

I don't think there is a performance difference between active and idle, but I haven't experimented with it much. I leave BT off unless I need it. All of my figures were with BT off, so at idle with minimum screen brightness and BT, power consumption is 17 to 17.5W.

-darren

[sugo]

Thanks for all the info, darrenf.

I hope other T60p owners can comment on battery life they are getting. 4.5 hours on a 9 cell doesn't sound very promising on a $3k laptop.

[darrenf]

christopher_wolf,

Thanks for the heads up and the info from Cryptonome!

I worked with RightMark to enable other power saving states to see what would happen. I don't know of an easy way to issue a HALT and I don't know if it can be done from a user-space program, so I can't repeat Cryptonome's exact experiment.

Interestingly, after setting the CPU power saving states back to their defaults I suspended and resumed only to find that one of the cores is running 20C cooler than the other. RightMark shows that both are awake and processing code, so I don't know what the cause would be.

I'll tinker some more -- until I fix it or break it. I think I need to read Intel's reference on power states too.

-darren

[jeremivw]

Thanks for all the info, darrenf.

I hope other T60p owners can comment on battery life they are getting. 4.5 hours on a 9 cell doesn't sound very promising on a $3k laptop.

I'm getting great battery times. I'm using no power man (except lcd off @ 5mins for when I step away) and max performance on all and have been doing things like copying 50 gig from a usb hdd, using flash keys, bt mouse, wwan, everything on lcd full bright cd installs yada yada yada.

I suspended at 3% in just a little under 4 hours, I think. Not sure about the exact minute but I'm close with my est. Additionally, the power drain has been very slow and consistent. It's a really nice batt.

BTW: I've got the PANAsonic 9cel

[darrenf]

For all the stress that you put on it, I would consider 4 hours great as well! A few questions for you:

1.) What is the battery life at minimum brightness with no load?

2.) What is the current draw in Watts under those conditions?

3.) What is the capacity in Wh of your battery?

You can get all this through Power Manager.

Thanks!

-darren

[jeremivw]

For all the stress that you put on it, I would consider 4 hours great as well! A few questions for you:

1.) What is the battery life at minimum brightness with no load?

2.) What is the current draw in Watts under those conditions?

3.) What is the capacity in Wh of your battery?

You can get all this through Power Manager.

Thanks!

-darren

OK. I'm still charging (so all those little numbers are still going up) but I'll try to answer in a bit, dude.

[jeremivw]

Got some numbers.

I'm currently downloading Office updates via WWAN and copy several gigs from a usb hdd.

drawing 2.34A w/92% remaining (76.11/82.95Wh). 2hrs 44mins remain.

now since I wasn't working constantly this morning (stepped away many times) and the lcd was turning off, I guess the lcd draws a lot of amps. Too bad I can't see what the draw is if I turn the lcd off.

[Kyocera]

What if you plugged in an external monitor and turned off the LCD?

[jeremivw]

What if you plugged in an external monitor and turned off the LCD?

No monitor to try that out.

Now I'm fully charged and booted up. I'm preset to MAX BATT LIFE (LCD min, etc.) and about to unplug...

Battery details...
Battery: 1-Main
Status: Discharging
Remaining percentage: 100%
Remaining time: 5:04
Remaining capacity: 82.92 Wh
Full charge capacity: 82.95 Wh
Current: 1.37 A
Voltage 12.36 V
Temp: 28 C
Cycle Count: 3
Manufacturer name: Panasonic
Manufacturer date: 2005-12-28
First used date: 2006-03
Serial number: 3xx
Bar-coding number: 1Zx#x#xx#xx
FRU part number: 92P1133
Device chemistry: Li-Ion
Design capacity 84.24 Wh
Design voltage: 10.80 V

After a few minutes (typing the stats above) the remaining time is still fluctuating between about 5:03 and 5:35 time remaining...and I've got all RF still on, WWAN up, BT mouse in use, etc!