RMClock not controlling voltages

[soccerballtux]

Running performance on demand on Core2Duo with Abit Ip35-E motherboard.

Have the core2duo overclocked to 3.4Ghz at 1.4875.

at 6x mult; VID is 1.188. At full speed 10x, it jumps to 1.288 reported in RMClock.

In CPU-z it goes from 1.4v idle at 6x to 1.472v load 10x mult.

What setting am I not correctly applying? I've used this program before with my previous Sempron 3100+ overclocked and the VID changing worked no problem; was reported correctly in CPU-z no problem. I'd really like to have my chip running at a lower voltage when I'm not using it. As it stands, I can't seem to underclock it.

 

[Quiksilver]

Wouldn't a better question be why are you using software from a desktop environment?
The smart idea here would to simply enable C1E and Speedstep in your BIOS and do all overclocking changes from there rather than relying on desktop software to do it for you.

 

[soccerballtux]

Originally posted by: Quiksilver
Wouldn't a better question be why are you using software from a desktop environment?
The smart idea here would to simply enable C1E and Speedstep in your BIOS and do all overclocking changes from there rather than relying on desktop software to do it for you.

Those are both already on :/

 

[soccerballtux]

bump

 

[soccerballtux]

plz halp

 

[Binky]

I'm no math whiz, but it seems that the ratio between these two extremes is about what I'd expect. You have the voltage cranked up in bios, no? RMclock is reporting the stock voltages, and cpu-z is reporting actuals. Compare both extremes from rmclock to cpu-z;

1.188/1.4 =84.9%
1.288/1.472 = 87.5%

Seems reasonably close and about right to me.

I use RMclock on several core2 systems and love it.

Edit: if you want to test my theory, set your bus speed and cpu voltage to stock levels and see if rmclock and cpu-z report about the same figures. I'm guessing that they will...

 

[soccerballtux]

Originally posted by: Binky
I'm no math whiz, but it seems that the ratio between these two extremes is about what I'd expect. You have the voltage cranked up in bios, no? RMclock is reporting the stock voltages, and cpu-z is reporting actuals. Compare both extremes from rmclock to cpu-z;

1.188/1.4 =84.9%
1.288/1.472 = 87.5%

Seems reasonably close and about right to me.

I use RMclock on several core2 systems and love it.

Edit: if you want to test my theory, set your bus speed and cpu voltage to stock levels and see if rmclock and cpu-z report about the same figures. I'm guessing that they will...

Sure;

think is I know that at 6x mult I could undervolt it way more than 1.4v. I'm just not sure how to. I tried turning the 0 (6x mult) volt down but that never changed anything

 

[Binky]

Rmclock can't do that. You have an approximate +15% voltage boost kinda hard-wired through the bios setting (from RMclock's perspective). RMclock can only manipulate the VID/FID and these changes can't override the boosts to FSB and voltage that you have done through bios settings.

I think clockgen and softfsb used to make changes like this to voltage and fsb but I don't know of any similar utils that work on the core2.

Edit: check out this thread (posts 22 and 26 especially) - http://www.hardforum.com/showthread.php?t=916551&page=2

 

[GundamF91]

Binky's correct. RMclock will only allow you the VID based on the CPU's VID table. So that means it's only whatever the factory set it as. You can verify this by looking at RMclock tab that shows start up multi and VID. That'll match the lowest setting available.

One shortcoming of RMClock is that it will not allow you to beyond bounds of the factory setting. What you can do however is to artificially raise BIOS VID as you've done already. So if your BIOS VID is 1.4v, and your factory setting VID only goes up to 1.3v, then whatever you do, the gap will be 0.1v. So if RMClock drops VID to 1.2v, it really drops the BIOS VID down to 1.3v. There's no other way around it.

As for using Intel driver vs. RMclock, I much prefer RMclock. With the C1E, it would drop and increase multi and VID almost constantly, swinging back and forth between 6x and 11x. I dont' think that's healthy for a system. So I use RMclock to normalize it, so that it would only go up if the CPU usage is constantly high for a set time, ie. 3000msec. So this eases the back and forth of the throttling.

 

[soccerballtux]

I was able to push my Sempron 3100+ past the rated voltage (I'm pretty certain about this) when I overclocked it and RMClock managed to control it fine; what you're saying is giving me the idea that whatever voltage I had it at, it must not have been beyond the manufacturer's specifications, because I was able to control the voltage? I did notice a slight disparity between what RMClock told me I was running at and what CPU-z told me (CPU-z was higher than RMclock).

I agree about RMclock I noticed that swinging back and forth too.
~~~~~

I think I see what Binky is saying; when I run Orthos on my powersavings mode with forced 6x multiplier my voltage is 1.384-1.408 in CPU-z. When I jump to 10x it's 1.472v. This makes sense as I've got the Bios voltage at 1.5025(or perhaps .5075 I don't recall; this would make sense because 1.408+0.1 = 1.508)

So it seems like you're saying I'm not able to undervolt it? The CPU says "I can go between 1.18v and 1.28v and no matter what I can't go any below that"? On my and my friend's semprons in our laptops we've undervolted from ~1.15v to .975v; increased my battery life about an hour to 6.5h.

I guess that would mean there is something different about this Core2Duo and my previous Sempron and the Sempron in my (and my friend's laptop). I just thought all processors were capable of being undervolted in RMclock .

~~~~~~
My greatest concern over all this is with having a constant voltage that is higher than the recommended specification for this chip. On my previous OC'd desktop Sempron I felt 100% ok with the higher voltage because 99% of the time (or even more than that really) the processor was at 4x mult and a really low voltage; idle temps were great.

How much damage would 1.408v constant do to the chip?

 

[Binky]

Two posts up I suggested that you read another thread on this topic. Give it a look and read the part about the cpu voltage "auto" setting. You'll see that RMclock can't actually change voltages itself. Only the motherboard can do that. Rmclock only manipulates the FID/VID.

Rmclock behavior on an AMD platform may be completely different. I do recall having significantly more control of an Athlon Mobile cpu on an SIS chipset. I think I used CrystalCPUID back then.

Even if you don't get the voltage control that you are seeking, I still think that rmclock is very useful to fine tune the throttling of your CPU.

 

[soccerballtux]

Originally posted by: Binky
Two posts up I suggested that you read another thread on this topic. Give it a look and read the part about the cpu voltage "auto" setting. You'll see that RMclock can't actually change voltages itself. Only the motherboard can do that. Rmclock only manipulates the FID/VID.

Rmclock behavior on an AMD platform may be completely different. I do recall having significantly more control of an Athlon Mobile cpu on an SIS chipset. I think I used CrystalCPUID back then.

Even if you don't get the voltage control that you are seeking, I still think that rmclock is very useful to fine tune the throttling of your CPU.

Aye I'm looking at it now I'll post back in a sec~

thanks for the recommendation.

edit: So it looks like if I set my voltage to auto, RMclock should function correctly. I think I'll go try that and see if I BSOD on bootup. Probably will :/