Undervolting W3J

Ok.. I'm pretty knowledgeable when it comes to undervolting the Pentium M.. but undervolting the Core Duo is a bit confusing to me.

1) I'm confused as to why NHC keeps reporting my 2.0GHz CPU to be 1GHz when in full speed.. Is this because it's only reporting the speed for one core?

2) In the voltage control tab in NHC, the 12x multiplier shows up twice.. why is that?

3) Prime95 doesn't seem to work. I've plugged in my new voltages, and I wanted to stress test it overnight, but it won't start.

Any advice is appreciated. Thanks in advance!

 

I tried NHC for dual core, it didn't really do anything but exhibit a few bugs like those you're mentioning: wrong speed reported, multipliers all wrong etc.

I switched to RMclock and it's working w/o problems.

I don't know Prime95.

If you really like NHC, make sure you're using the latest version, in my opinion the developer(s) should have done something to address the dual core issues by now, it's been half a year.

One note: you are not going to be able to undervolt your CPU when on lowest multiplier... with any currently available software. You can undervolt it at higher multiplier values, and reduce heat output, but not at the lowest speed.

 

I think NHC is simply filling in all the available slots in the higher multipliers - I have 13x twice on my PM 740. I just check one as active and leave the other turned off. Plus, it hasn't been updated since right before the Duo was released.

 

Then, I was wrong with the update and NHC is almost certainly buggy for Dual Core. My advice: go for RMClock. Bug-free for Dual Core, and more efficient at what it does since it's focused on it.

 

Oops, my bad - just checked the NHC site for what all was in 1.10 B3 and found:

1.10 News and fixes in the Notebook Hardware Control BETA 03
- add new faster and better nhc driver
- add new faster and accurater clock detection function (+- 0.2%)
- improved stability and fixed the blue screen problem on some core duo and P4 notebooks
- nhc is now faster and needs less cpu time
- Todo: -> BETA 01
- (01.04.2006)
News and fixes in the Notebook Hardware Control BETA 02
- add support for Intel Core Duo / Solo
- add multiprocessor support
- add new system tray icons (CPU Load, HD Temperature)
- add icons for profiles
- add clock modulation
- improved the monitoring function
- removed standby bug on HP Notebooks
- removed a lot of other bugs

So it still hasn't been updated since the Duo systems became widely available, but it was in B3. Just little leftover bugs that need to be worked out in the next release I guess.....

 

Cool.. yea NHC has been Beta 3 for the longest time now.. I wonder if the author is still working on it. I'll check out RMClock then. Thanks!

 

When I use RM clock it doesnt display the right cpu frequency. It says I am always at around 987 MHz. When I have RM clock set to maximal performance and windows power options to always on- the cpu still is 997 MHz but the throttle is around 1430 MHz. When I run super pi tho everything is 1830 MHz. I thought there was some way that I could have it set to 1830 all the time so it doesnt have to switch. Anybody else find this?
thanks
Dan

edit: do I have to turn off the intel speedstep for it to work?

 

That's the same thing with NHC! Hmm.. I wonder if these programs are only reporting the speed for just one of the cores?

 

earthdan you are right. I hadn't noticed this before, as I've almost never run Max Performance.

So, after some testing with RMClock 2.05:
- on Power Saving, RMClock succeeds in keeping the CPU at the desired speed (i.e., it won't clock it higher than 6x).
- on Perf on Demand, RMClock works as it should: multiplier changes dynamically, and voltages follow the custom settings. I'm unsure if they correctly correspond to the multiplier settings though, graphs cannot show that level of detail.
- on Max Perf, RMClock does not keep the clock at 11x as it should. Instead, dynamic switching still applies. However, the voltage is set to its 11x value!!

My conclusions:
1. RMClock is not bug free ( :takes stick and bangs himself on head for not noticing before: )
2. RMClock is only in partial control of the CPU frequency: can keep it to low values, cannot force it to high values. RMClock is in control of voltages. Unsure if voltages correctly follow multiplier settings. Unsure if RMClock is dynamically switching the multipliers, or Intel SpeedStep.
3. Unrecommended to use RMClock on MaxPerf, as it changes the voltage to the CPU without changing the multiplier... Unknown what effects this has are on the CPU, most likely negative.

These conclusions are drawn on the basis of RMClock readings. As far as I can tell, they are consistent with benchmark results and temperature readings (e.g. when on MaxPerf temperature sharply increases due to voltage increase).

New version of RMClock available (2.1), will install and check if these things still happen.

 

E.B.E.-
I just tried to install RMclock 2.1 and it kept my multiplier on 6 and voltage on 1.262 for every setting. I quickly shut this off because you said it might not be recommended. I will play around with it for a bit but I dont really wanna screw anything up. ANy ideas?

edit: interesting- I put the power settings in windows to home/office and its back to 6x and .95. I guess this means that when setting to "always on" it just changes the voltage. Says my cpu load is around 35% when idling.

 

It's not gonna burn the processor or anything, don't worry. I've kept it on MaxPerf for hours at a time, with these bugs in place, and nothing happenned. So you can test without fear. Just don't leave it like that for normal use.

I'm trying RMCLock 2.1 right now.

 

I haven't had the chance to try it out yet.. will do after work. Keep us updated though. Thanks!

 

how did RMclock 2.1 go? It seems very different than the last one - All the default settings are a little screwy.

 

A dump of benchmarking results + CPU logs for various RMClock and Power4Gear settings. Benchmarks were ran 4 times to estimate how consistent results are, not only their magnitude.

V6J T2400, 1.83 GHz

Part 1: Maximum Performance.
RMClock ver. 2.1, RMClock ver. 2.05, P4G SuperPerf (21_MP, 205_MP, SuperPerf.jpg)

Here and in the sequel, when I refer to P4G tests, I use RMClock only for monitoring, with management off.

 

Part 2: Performance on Demand.
RMClock ver. 2.1, RMClock ver. 2.05, P4G HighPerformance (21_PoD, 205_PoD, HighPerf.jpg)

 

Part 3: Power Saving.
RMClock ver. 2.1, RMClock ver. 2.05, P4G Quiet Office (21_PS, 205_PS, QuietOffice.jpg)


[PS: sorry for multiple posts, can only attach 5 images to a post so I had to separate them in some meaningful way...]