Jump to content

Understanding Prime95 (results)


yonyz

Recommended Posts

Hi,

 

I've overclocked my Intel Q6600 to 3.0Ghz (9x333, because I have 667Mhz DDR2 sticks).

I ran a Torture, In-place large FFTs Test overnight, and in the morning, I saw that the program was unresponsive, with a 'Prime95 is not responding' Windows error, with the two familiar options of closing to waiting for the program. This was the first try.

Another night, I ran the same test, and this time, in the morning, the computer was simply frozen. Even the mouse pointer was unresponsive, so I had to hit the Reset button.

On the third try, in the morning the program was again unresponsive.

 

What do you think it means? Not enough voltage for the CPU? BTW, looking at the unresponsive program window, I couldn't see any errors both times.

 

Thanks in advance.

Share this post


Link to post
Share on other sites

Large FFT's are designed so that the data for them is too large to fit in the processor cache, so it must be stored in the RAM. So it could be either your processor lacking enough voltage, or a problem with your RAM. If the same happens with the small FFT's, then its for sure the voltage.

Share this post


Link to post
Share on other sites

Alright then, I'll run the small FFTs test tonight and post the results tomorrow.

Yup, once again, the application crashed. By how much should I increase the voltage of the CPU?

Share this post


Link to post
Share on other sites

Prime 95 Errors:

 

Freeze: Increase the vcore

 

Other errors can indicate instability with the chip if they are during small fft (increase vcore by .125) or instability with ram large ftt (Try raising the ioh and/or running memtest).

 

CPU errors show in large and small, RAM only shows in large or blend.

Source: OCN i7 OC guide

Edited by alexonfyre

Share this post


Link to post
Share on other sites

Complete system specs and current voltage settings would be nice :)

GIGABYTE GA-P35C-DS3R

Kingston 2x1GB DDR2 667Mhz CL5-5-5-15

GeForce GTS 250 1GB

Intel C2Q Q6600 @3.0Ghz (9x333Mhz)

Corsair VX-450

 

CPU VCore as set in BIOS: 1.25V

CPU Normal VCore as stated in BIOS: 1.2625V

CPU VCore as shown in Hardware Monitor: 1.20V (Minimum: 1.18V)

Share this post


Link to post
Share on other sites

Are you using the stock oem cpu heatsink and cooler or aftermarket? Do you know what stepping the processor is?

 

If you're temps are in check you can try the following;

 

CPU Vcore = 1.35v

Northbridge Voltage = 1.45v

CPU PLL Voltage = 1.60v

 

Retest after changes - but make sure to keep an eye on cpu temps................................................

Share this post


Link to post
Share on other sites

Are you using the stock oem cpu heatsink and cooler or aftermarket? Do you know what stepping the processor is?

 

If you're temps are in check you can try the following;

 

CPU Vcore = 1.35v

Northbridge Voltage = 1.45v

CPU PLL Voltage = 1.60v

 

Retest after changes - but make sure to keep an eye on cpu temps................................................

I am using a CoolerMaster Hyper 212 Plus CPU cooler.

Isn't 1.35 a big jump from 1.25, in terms of voltage?

How do I change the northbridge voltage, and why should I do that?

How do I change the CPU PLL voltage? Also, Intel recommends not passing 1.5 voltage, so why set it to 1.6?

 

As for the CPU stepping, it is G0.

Share this post


Link to post
Share on other sites

I am using a CoolerMaster Hyper 212 Plus CPU cooler.

Isn't 1.35 a big jump from 1.25, in terms of voltage?

How do I change the northbridge voltage, and why should I do that?

How do I change the CPU PLL voltage? Also, Intel recommends not passing 1.5 voltage, so why set it to 1.6?

As for the CPU stepping, it is G0.

 

#1 Ok - that's much better than the stock OEM cpu heatsink/fan

#2 In the grand scheme of things not so much so - you'd actually be safe up to about 1.425v for 24/7 use with good air cooling

#3 In your BIOS northbridge voltage is referred to as G(MCH) Overvoltage Control - you'll want to add +.2v (we might be able to drop to +.1v later)

#4 Not sure what revision board you have - but CPU PLL voltage may not be an option in your BIOS

Share this post


Link to post
Share on other sites

I am using a CoolerMaster Hyper 212 Plus CPU cooler.

Isn't 1.35 a big jump from 1.25, in terms of voltage?

How do I change the northbridge voltage, and why should I do that?

How do I change the CPU PLL voltage? Also, Intel recommends not passing 1.5 voltage, so why set it to 1.6?

 

As for the CPU stepping, it is G0.

 

To answer each of your questions about voltage:

The more voltage, the stronger the current through your chip (I=V/R). When your proc returns errors, that generally means it doesn't have the electrical power to handle the frequency that it is trying to run. So increasing voltage may make your unstable system stable by increasing the power to the chip. This holds true for just about all electronic components. Naturally there is a limit to how much current the components can physically handle without failing, and that is why we must be very careful when increasing voltage because if that failure is a critical failure then you can flush your $300 processor down the toilet.

Now:

A) .1 is a large jump, but 1.35v should be safe for that processor. As wev said, make sure your temps are okay (sub-65 at load, I believe, if someone out there is more experienced with C2Q procs feel free to correct me.) That cooler is a champion though, so if it is seated correctly and your airflow is good, you should be okay.

 

B) Your northbridge or NB or chipset or QPI (in Intel, HyperTransport for AMD) is the means by which your CPU communicates with the rest of the system (aside from memory, but more on that later). It is possible that instead of your proc not having the power to handle its clock, the chipset is not able to move the data fast enough back and forth, causing an asynchronism and an error. Increasing voltage to increase power allows the data to move more quickly.

 

C) The PLL or Phase Lock Loop, is essentially the part of the proc that holds the BCLK, the whole system relies on this creating a consistent signal. If it doesn't have enough potential across it (voltage) then it may not be able to make a clean, consistent signal, which would could also create a CPU error. There is some debate on the PLL voltage and how much is safe, I don't have much experience with it, so I can't say one way or another, but if these guys say it is okay then I would trust them.

 

Here is a pic I found at pureOC, that might help illustrate things:

diagra.jpg

Edited by alexonfyre

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...