DV2 undervolting: get at least 25% more battery life!

Following the excellent flipfire's undervolting guide and inspired by Silvr6's experience with undervolting his dv2, I decided to give it a try and... I am speechless!! I was able to increase the battery lifetime from almost shameful 2h 25 minutes to 3h 30 minutes for mpeg 2 play, or even to 4h 20 minutes (light browsing and lots of reading). So, you get at least 25-30% more battery lifetime. This makes me like my dv2 a lot more!
Perhaps, it's the noted AMD's tendency to overvolt their CPUs, perhaps something else, but I was able to get the following 100% stable (in my case) voltages:
4.0x - 0.6875 V
4.5x - 0.7125 V
5.0x - 0.7250 V
5.5x - 0.7500 V
6.0x - 0.7750 V
6.5x - 0.8000 V
7.0x - 0.8000 V (yes, it’s 0.8V again)
7.5x - 0.8250 V
8.0x - 8.5000 V
Please note that these voltages are stable for my particular CPU; yours might be slightly lower or higher. For example, Slvr6 was able to get a bit lower with 4x – 0.675 V. Read the undervolting guide!
But I hope that the table above will encourage you to try the undervolting (btw, unlike the overclocking it’s 100% safe), since the benefits for dv2 and AMD’s “neo” are so dramatic!
Also note, that when in battery mode, I disabled all higher clocks, except 4x. And, similarly to Slvr6 findings, I did experience NO SKIPPED FRAMES even when playing an HD video. If I switch 8x to 4x dynamically, when there was some stuttering for about 2 seconds , but 10 seconds after the switch and further on – no skipped frames!
Finally, I was also able to compare the power consumption using a “kill-a-watt” meter. The findings were consistent with the battery lifetime numbers above. With standard factory voltages the laptop consumed about 29.5 W while playing an HD video (saved to the hard disk), and with RMclock “on” and only 4x (@0.6875 V) enabled I got 22 W – about 25% reduction in power. Since the video/audio playback quality was not affected at all, I think it’s worth trying!

 

Fantastic results, thanks for posting. I'll give this a shot sometimes this weekend. Now if we can just do something about the Radeon HD 3410, which is easily the biggest contributor to the heat and power consumption issues on this platform.

 

trying this right now.

 

You are welcome! And yes, it would be great to do something with the graphic card! I have Powerplay "on", but I don't see much of the difference. However, I'm using an ATI's beta driver for Windows 7 and perhaps it can be still tweaked.

 

Please let us know how did it work! I'm curious what the standard deviation is for allowed voltages. If it's large, that would explain AMD's choice to overvolt so much. But if it is small (I can only compare to Slvr6 data) - then, perhaps, we will see another BIOS update from HP?

 

Seems to be working great so far.

I basically just used your table, with the exception of running at 0.675 at 4.0x.

Haven't tried going any lower on anything else though.

 

Good stuff, i didnt even know it worked on newer Neo processors.

 

worked fine, running at .675 at 4.0x also, much cooler!

 

I actually started with .675 V at 4x too and it worked fine for about 2 hours, but then I got a bsod. So, if you will get it - just increase the voltage to .6875 V. I am using this laptop for about a week now with RMclock on - and had zero problems. And windows 7 is great.

 

What sort of temperatures are you getting from these new Neo cpus?

 

BTW, in case if you will wonder (like I did) how to make RMclock to run automatically at startup without causing UAC alert with Vista/W7, look here. It worked perfectly for me and RMclock now starts fully automatically every time without having to disable UAC.

 

Hi flipfire, and thanks so much for the wonderful guide!

After at least one hour of CPU stress test, I had the following temperatures:
8.0x: factory -> 64C (1.050 V), RMclock -> 59C (0.850 V)
4.0x: RMclock -> 51 C (0.6875 V)

 

I think I'll give this a try next time I pick up a dv3t - I'm all for a cooler and quieter notebook, and the extra battery time is good to have too.

 

hotter than my atom powered n10j that's for sure.

 

Well, it's hard to compare, since the temperatures above are not from CPU diode, but ACPI temperatures. For whatever reason CPU core temperatures reported by both Everest and HWmonitor were below the room temperature (i.e. something like 17 C - clearly wrong). So I had to go with ACPI.

Right now, when I'm typing it, the ACPI-reported CPU temperature is 47C. Not bad at all, I think.

 

my acpi temps are 52, was 60

 

here's something unexplainable: running my dv2 without undervolting, i get 1240 on 3dmark. running it with undervolting, not only is it cooler, but my new score is 1460... i did not change my drivers or uninstall any programs in b/w those tests. can anyone else confirm this?

 

I checked it with mine and got similar results: 1237 with RMclock off and 1390 with RMclock on Honestly, I'm not going to use this laptop as a gaming machine, so 3Dmark scores are not important (as long as Aero works fine and HD movies are playing well), but it is funny and strange.

 

yea, i just want to be able to play civ 4 every now and then, maybe sim city 4... upgraded my hd to a 7200rpm though also

 

Does undervolting decrease the performance of the PC??

 

No, it does not. Please read the guide.

 

Thanks for this thread. The dv2 is still $599 AR at staples until tomorrow. I might pick it up.

 

not that i've noticed. it stutters a little, but this thing stuttered without the undervolting anyway. i think you could use this program to run the cpu at full speed 100 percent of the time (no speedstepping) to avoid stutters, but that would be overkill. once a program starts running, i've had no problems with it running smoothly.

 

Set up profiles, when im on AC power its automatically locked to full throttle. When i unplug the AC and go on battery, speedstep kicks back in.

I used to lock my cpu to 1.2ghz on battery power, its still pretty powerful enough to run word, listen to music and chat.

 

i got 4.5 hours using the undervolt tutorial (with wifi on) but not really running any apps except steam and windows sync. Backlight on lowest setting. running windows 7 rc 7100. real world testing will be different, i know.

 

thanks for the table of values. i have undervolted my dv2-1044ca with the numbers and there is a noticeable improvement in terms of reduced CPU temperatures and battery life.

i hope there's a means to undervolt the ATI Mobility HD3410. that thing runs over 50C in my dv2.

 

well, i read the gpu bios but couldn't edit it. not brave enough to try flashing the thing. if you want to run it in xp, use the 9.4 xp catalyst drivers with mobility modder and maybe you can underclock it that way (i didn't try).

 

I have a dv2-1044ca and I seem to get random freezes with the following settings:

Code:

FID   VID
4.0x  0.6875 V
4.5x  0.7250 V
5.0x  0.7375 V
5.5x  0.7625 V
6.0x  0.7750 V
6.5x  0.8000 V
7.0x  0.8250 V
7.5x  0.8250 V
8.0x  0.8500 V

Basically, I set the VIDs for the first and last FIDs and checked "Auto-adjust intermediate states VIDs". I am using RMClock V 2.35.0. I bumped the 4.0x FID to 7 V to see if that makes a difference.

 

not all cpus are created equal, raise your voltage till you stop getting bsod

 

Most likely it is the lowest multiplier (4x) that causes the problem. So hopefully 0.7V will fix it. And don't worry, the difference in power saving between .6875 and .7 is still negligible: (.6875/.7)^2 ~ 3%.
If increasing 4x to .7V will not help, you will need to check every voltage with a real stress-test. It's time consuming, but it only takes a second of your attention each hour (it will take about 4-5 hours, to check all the multipliers). The benefit is that you will have voltages that are 100% stable for your CPU.

 

I had freezing again at 4x set to 0.7 V so I began stress testing each multiplier.

I have just finished stress testing each of the multipliers with the 4x VID at 0.7 V, the 8x VID at 0.85, and all the intermediate multipliers at the auto-adjusted voltages. I used StressPrime 2004 Orthos v0.41.110.18 and ran its Blend test once for each multiplier for 45-70 minutes each. I experienced no problems at all with any of the tests. I will now try using the computer normally again although I expect the freezing will return.

I want to note that when I experience freezing, I mean just that: all keyboard and mouse inputs do not produce any kind of response. Nothing on the screen updates as well. Unplugging the AC adapter dims the backlight to the power profile's brightness level. I do not get a BSOD. I must hold the power switch to force my computer to shut down.

Edited to add: I am now running 4x at 0.7125 V. So far, so good.

 

HW monitor is not seeing my CPU. Any suggestions?

 

that's normal, it won't detect the cpu (yet)

 

OK, neither 0.7125 V nor 0.7250 do the trick. I still get freezing, even when the system is idle. Is it possible for some other setting in RM Clock or elsewhere that could be causing the freezing?

 

hmm, did you follow the tutorial exactly as written?

 

Well, I haven't changed any settings unless the guide said so, aside from disabling RMClock's battery icon. Perhaps someone could post their Advanced CPU Settings or any other settings that may cause problems if changed?

Edited to add:
It has occurred to me that perhaps the freezing sometimes occur whenever the multiplier changes. I think this is because I never experience freezing whenever I run Orthos on a single multiplier.

 

So when will it detect it when i start running the Othros program?

 

anyone else having trouble getting the Orthos Tool? seems the URL in the undervolting guide is not working... Any where else to grab this tool from?

 

If you Google "orthos", the first result should be what you need.

 

Well, it looks like you need to check those "auto-adjusted" voltages.

And if that doesn't help when there are reports that RMClock has issues with windows 7. I didn't experience anything like that though.

 

I do not quite understand what you are telling me with the highlighted parts of my words. Like I said in the first quote, I did run stress tests at each of multipliers, whose VIDs were set by RMClock, and had no problems. Are you telling me to test multiplier transitions?

By the way, I am using Vista on my dv2, not Windows 7.

 

my god undervolting just works wonders.........
props to flipfire= :

 

If you did check all the intermediate multipliers for stability, and you feel that each of these tests was long enough, then it is very strange indeed. Is your system stable without RMclock?

BTW, what BIOS version do you have?

 

ok i'm having a problem opening RmLock. it's a .sys extension and it doesn't to let me open/unzip it. it tries using the internet to find the best program to open it with. any help would be great.

 

I updated the BIOS to F.07 soon after I got it.

Right now, I have the 4x VID set to 0.75 V. I have been running it for a couple hours without freezing. Maybe I got a CPU with poor tolerance for lower voltages.

Edited to add:

My computer froze again, the moment I clicked the submit button for a post. I'm going to stop using RM Clock for awhile to check my system for stability, although I don't think I had any problems before I started using RM Clock.

 

It sounds like you just downloaded the driver if all you have is a .sys file. You need to go here, scroll down to "RightMark CPU Clock Utility (RMClock)", and select either the .rar or .exe file to obtain the complete program. If you are running Vista x64, you will need the signed 64-bit driver linked in the undervolting guide.

 

it's still not working. it sayin it can't install driver. make sure it's not a read only file

 

Did you replace original RTCore64.exe from the install file with the one linked in the undervolting guide?

 

After reading this thread, I think it is possible that processor power management may be causing my freezing, since I usually use the HP Recommended power plan and I changed the minimum processor state from the default of 100% to 50%.

I created a custom power plan in RMClock, set both the minimum and maximum processor states to 100%, and assigned it to the "Performance on demand" profile.

Hopefully this will finally do the trick as it did for the last poster in the aforementioned thread.

Update

Froze again but I think I know why. In RMClock > Management, I had "Use OS load-based management" unchecked, which could be causing RMClock to conflict with Vista's power management. I have it checked now. This may explain why I do not have problems when running stress tests on single multipliers; no load-based management from either RMClock nor Vista come into play.

 

i still can't get rmclock to open damnit. i keep getting this message.