Core Duo's USB bug applies to Dothan in a desktop.

Zero · Post by **Zero** » Sun Mar 19, 2006 5:07 pm

I've got a question about sleep states and the inability to get into C3 with a Dothan in a desktop. (Its probably not completely necessary, since a desktop has no battery to worry about, besides a UPS, but this desktop is more of an experiment in power consumption & heat dissipation, so it seems valid.)

There's been alot of talk about the "bug" that effects the Core Duo. Basically, when a USB2 device is plugged in, the battery life is substantially reduced because the processor is unable to get into a C3 or lower-power sleep state, as it was designed to, to conserve power. I wasn't very worried about it until I learned that it affects all Pentium M / Centrino systems as well, since its a windows problem. (I have both a laptop and a desktop with a Pentium M.)

I can reproduce the glitch, and watch the processor in the laptop stay out of C3 with a USB2 thumbdrive connected, and watch it regain its C3 sleep ability once the drive is disconnected. BUT, something else that I didn't expect.... when monitoring the laptop, C3 is also disabled when plugged in to AC power, (since battery conservation isn't necessary when plugged in). So when I tested the desktop Pentium-M, I wasn't suprised to see that it would never go into C3 at all.

Anyway, on to the question:
The USB2 glitch I can deal with, but now I'm confused, as to, how to allow the desktop Pentium-M 760 (which is on 24/7 and is usually idle), to be able to enter C3 or a lower power sleep state even if plugged in, to see how low I can get overall system power consumption. (It already scales between 800Mhz and 2.0Ghz with varying voltage, so I've done that part.)

System info, if it helps: P.M-760, i915GMm-HFS, Seagate Momentus 5400.2 120GB, 1GB DDR2, Phantom350, Zalman 7000AlCu w/ nexus fan, Plextor PX716-SA.

Thanks for any advice!!!

Heres the article if anyone's interested:
http://www.anandtech.com/mobile/showdoc.aspx?i=2693

Excerpt:

"Windows XP SP2 installs a USB 2.0 driver that initializes any connected USB device. However, the USB 2.0 driver leaves the asynchronous scheduler component continuously running. This problem causes continuous instances of memory access that prevent the computer from entering the deeper Advanced Configuration and Power Interface (ACPI) processor idle sleep states. These processor idle sleep states are also known as C states. For example, these include the C3 and C4 states. These sleep states are designed, in part, to save battery power. If an otherwise idle portable computer cannot enter or maintain the processor idle sleep states, the computer uses its battery power more quickly than you expect."

Tak · Post by **Tak** » Sun Apr 02, 2006 10:40 am

Interesting. I wasn't aware of this bug.

It already scales between 800Mhz and 2.0Ghz with varying voltage, so I've done that part.

Are you doing this with RM Clock?

If not, give it a try. It's an excellent utility which lets you configure power states and speedstep options. There's a section for C3 etc but it's greyed-out on my desktop machine, perhaps there's a way to force the pc into "battery mode" which should enable these options?

Zero · Post by **Zero** » Tue Apr 04, 2006 9:03 am

The speed management is done by windows, as long as you set the computer type as Laptop/portable, and select a power scheme that allows the scaling. To monitor the speed, I've been using NHC, http://www.pbus-167.com/chc.htm

Thank you for the link, I've downloaded that program and am trying to figure out how to use it.

I can get it to change the clock speeds and voltage, (same as with the windows scheme) but I'm still having problems figuring out how to enable C3 when plugged in.

I've set the profiles for AC and Battery both to "power saving", and under "advanced CPU settings", the boxes checked are;
"enable CPUSLP# in C3/C4", and
"Enable CPUSLP# in S1".

I'll keep playing with settings and buttons while monitoring C3 to see if any settings do the trick!

Tak · Post by **Tak** » Tue Apr 04, 2006 9:13 am

Ok, also you can try undervolting each speedstep state (I think NHC also does this but I prefer RMclock). You should be able to run the chip 100% stable at lower than default voltages.

I undervolt my laptop with it (not the desktop as it's overclocked!) and it makes a significant difference in temps, so the fan turns on less often, and it improves battery life too. For a desktop machine, it would reduce power consumption a little so it's worth trying

Here's a good guide:
http://www.notebookforums.com/showthread.php?t=70943

The forum for RMclock is also useful:
http://forum.rightmark.org/?id=6 (the main programmer, Dmitri, posts on there and may be able to answer your C3 question).

jojo4u · Post by **jojo4u** » Tue Apr 04, 2006 12:06 pm

Speedswitchxp let's you fine tune the Windows power schemes. I don't have a laptop so I can't try myself.

http://www.diefer.de/speedswitchxp/

Zero · Post by **Zero** » Tue Apr 04, 2006 6:48 pm

Tak wrote:Ok, also you can try undervolting each speedstep state (I think NHC also does this but I prefer RMclock). You should be able to run the chip 100% stable at lower than default voltages.

I undervolt my laptop with it (not the desktop as it's overclocked!) and it makes a significant difference in temps, so the fan turns on less often, and it improves battery life too. For a desktop machine, it would reduce power consumption a little so it's worth trying

That was my next step, before I got side tracked on this C3 thing. I'll be playing with the voltages next. The RMclock program seems like it has alot of useful options. I've been able to get it to do my clockspeed/voltage adjustments with it, but still no luck with C3. I'm thinking I may have to find some way to 'trick' windows into thinking its running on a battery? Also, thank you for your taking the time to help me out!

jojo4u wrote:Speedswitchxp let's you fine tune the Windows power schemes. I don't have a laptop so I can't try myself.

jojo4u, thanks for the link! I tried speedswitchXP, and it does have alot of options for controlling the Cstates of the processor. It only seems to have options for C0, C1, & C2. So maybe an update will come out soon for Dothan and Core/Duo. I set it all up, paying attention to the "AllowDemotion" and "AllowPromotion" settings in the C-state policies. Still, it never goes lower than C1. I just don't get it! It MUST be something to do with being on AC power, although even my laptop will go down to C2 with AC.