My laptop is randomly overclocking itself......

The_Conductor

Just wondering what (if anything) I can do with a weird problem......

I put a T9600 engineering sample in my laptop- and was wondering why it was running so hot. I installed the NVidia control panel to get clock readings for the different components- and it showed the two cores running with with very different FSBs (and randomly varying FSBs).

The multiplier is randomly between 8 and 15 and the FSB ranges from 500Mhz all the way up to 900Mhz- meaning the laptop goes all the way up to and beyond 6Ghz

Its got normal stock cooling- which is why its red hot (its an Acer 8930- so its a big baby). I'm going to have to try to figure a way of better cooling it.......

I've attached a screenshot of the NVidia Control panel showing the CPU settings (took it at random- obviously it should be 2.8Ghz- not 5 or 6Ghz.......

Any suggestions would be appreciated.......

S.

The_Conductor

And here is a shot of the second core (at about 6.5Ghz........)

Its actually stable- rather surprisingly- but its incredibly hot.......

Voodu Child

You're seeing the effective FSB, the actual FSB is 1/4 of the reported figure. So your clockspeed is 1/4 of the reported value. There wouldn't have been a hope in hell of the CPU actually running at 6GHz, especially idle :pac:

Better off to use something like CPU-Z to be honest.

Alanstrainor

Haha, 6.5Ghz and a laptop with stock cooling had me thinking the whole thing was molten! CPU-Z is what i use anyway, i find it's the most accurate.

The_Conductor

Alanstrainor wrote: »

Haha, 6.5Ghz and a laptop with stock cooling had me thinking the whole thing was molten! CPU-Z is what i use anyway, i find it's the most accurate.

I tried that earlier- but it only gave me a reading for Core 0
It came in a 2.8Ghz (its stated speed- but the multiplier and FSB vary if you refresh the screen topping out.

Curiously enough- the core voltage on CPUZ is randomly varying between 1.006 and 1.900 volts (its changing several times a second). It was about 1.4 volts in the screenshot. The intel spec page says its parameters are 1.000 to 1.250 volts. Am I right in guessing this indicates a stability issue?

Pic of the two apps side by side reporting on the same core attached.

Voodu Child

Well, that's more like it (the CPU-Z readings). Your FSB is 266 and the effective FSB is 4x266=1066. I would just disregard the Nvidia utility for now.

The core voltage is supposed to change with increasing multipliers, that's quite normal. But from the sounds of it than your CPU is being fed with excessive voltage per multiplier (for whatever reason, probably something to do with being an ES) which is why you are experiencing high temps. (Again, this is all assuming you are being shown correct readings)

What you want to do now is install Rightmark CPU Clock utility and work on getting a stable low voltage for each multiplier, starting with the highest (which should be 10).

The_Conductor

Voodu Child wrote: »

Well, that's more like it (the CPU-Z readings). Your FSB is 266 and the effective FSB is 4x266=1066. I would just disregard the Nvidia utility for now.

The core voltage is supposed to change with increasing multipliers, that's quite normal. But from the sounds of it than your CPU is being fed with excessive voltage per multiplier (for whatever reason, probably something to do with being an ES) which is why you are experiencing high temps. (Again, this is all assuming you are being shown correct readings)

What you want to do now is install Rightmark CPU Clock utility and work on getting a stable low voltage for each multiplier, starting with the highest (which should be 10).

10.5? (its 2.8Ghz)

Thanks for the tip- I'll install Rightmark when I get home and try to get to the bottom of this.

Btw- when I say its hot- its actually melted the plastic on the bottom of the laptop- its that sort of hot......

Voodu Child

smccarrick wrote: »

10.5? (its 2.8Ghz)

Yep, but I don't think Rightmark supports .5 multipliers at the moment.

LOL at the melted plastic, that definitely sounds too hot. You'll need to stress-test the CPU to find a stable voltage, but i'd recommend bringing the 10.5 (or 10 or whatever) voltage way down before firing up the stress-tester. You're going to just have to estimate it (so you may bsod) but i'd probably just go all the way down to 1.25v for the first iteration. You definitely don't want to run it at 100% load on 1.9v or whatever it is at at the moment, that's just too risky.

The_Conductor

Voodu Child wrote: »

Yep, but I don't think Rightmark supports .5 multipliers at the moment.

LOL at the melted plastic, that definitely sounds too hot. You'll need to stress-test the CPU to find a stable voltage, but i'd recommend bringing the 10.5 (or 10 or whatever) voltage way down before firing up the stress-tester. You're going to just have to estimate it (so you may bsod) but i'd probably just go all the way down to 1.25v for the first iteration. You definitely don't want to run it at 100% load on 1.9v or whatever it is at at the moment, that's just too risky.

Thanks for the tip. It idles on less than 1 volt (0.98v) and the Intel spec sheets specifies operational parameters as 1.000 to 1.500 volts.

Its a pity Acer don't allow you play around with BIOS settings!