Overclock my Core 2 Duo E8200

[AddRAM]

Just wondering if overclocking an E8200 to 3 GHz is ok ? All my specs are below.

My current cpu temp averages 94 F , Thanks All

 

[TwoCables]

Yeah, it's ok, but the stability of your system has to be tested now. While testing the stability, watch the temps using Real Temp (it will go by Celsius which is the standard). For the sake of avoiding instability caused by heat, try to keep the temps out of the 70s and 80s.

Also, watch CPU-Z for the core voltage. Keep that under 1.45V as shown by CPU-Z.

You can also seek more help at www.Overclock.net.

 

[AddRAM]

Where did you get your info from ?

 

[TwoCables]

Where did you get your info from ?

Over 3 years of daily exposure on Overclock.net (12+ hours per day, sadly, sometimes 18+ hours - but I love that place). There was an official Intel PDF posted somewhere that shows that 1.45V is their absolute maximum recommendation for a core voltage. The temperature-related information is a combination of my experience with an E8400 and my experience on Overclock.net including all of the help I received overclocking mine to 4 GHz.

I also learned along the way that 95 to 100°C is the temperature where the CPU will automatically throttle itself down (it's the Tj. Max, or "Thermaljunction Maximum"), so that's essentially the maximum safe temperature. However, instability is far more likely to occur before it ever gets that high, especially if the CPU is under load. That usually happens in the 70s and 80s.

I know that Intel has a VID Voltage Range stated on their site of 0.85V to 1.3625V, but the VID Voltage Range is just the voltage that the CPU can request from the motherboard when the voltage is set to Auto and it's not related to the minimum and maximum safe voltages. I was using 1.368V for my E8400 when I had it in an EVGA 680i SLI for about a year, but that's not how I learned that going above 1.3625V is safe.

I also know that they show a T Case temperature of 72.4°C, but that's measured at the "geometric center on the topside of the CPU's integrated heat spreader", according to Intel. This is the same place that makes contact with the heatsink. Usually, the Tj. Max is about 20-25°C higher than the T Case.

As we say on OCN a lot, "OCN FTW". The knowledge one can gain over there about this stuff is just as amazing as the knowledge one can gain over here about Windows 7! If it weren't for them, then I'd still have a cheap pre-built from the Windows 98/2000 era and that's the era where my knowledge would be strongest in too.

 

[AddRAM]

OK here`s a shot of the bios, what changes do you recommend ?

 

[TwoCables]

Oh, for that small of an overclock (3.0 GHz), the voltage can be set so that when you're in Windows looking at CPU-Z, it shows about 1.25 to maybe 1.27-1.28V. That should be enough for an overclock this small (if anything, perhaps more than necessary). Due to a combination of vDrop and vDroop, you may need to enter a BIOS setting of like 1.35V to start with and work your way down from there in order to achieve this voltage (to see it in CPU-Z).

The VTT FSB Voltage should be ok at 1.20V, but you may also need to increase it one increment up (that is, if the next increment is 1.25V). It's a little bit similar to the core voltage because it's powering the FSB in the CPU, so stay under 1.45V for this too.

Everything else seen in the photo should be ok left where it is.

When I had my E8400 in my EVGA 680i SLi, I got it up to 4.0 GHz and I had to increase the FSB voltage to 1.4 too. So, I had a core voltage at 1.368V with a FSB voltage at 1.4V, but that's nForce boards for ya. They love the voltage. I even remember having to increase my NB voltage a bit, but the SB voltage didn't need to change because the SB doesn't control much that's related to overclocking the CPU.

 

[AddRAM]

So don`t touch the FSB Clock that`s highlighted ? I thought you would adjust that too. And gimme a tool to check the stability. I`ve never had an issue with this pc since day 1, after all I did build it

It`s been a great board.

We can go past 3 GHz, I just wanna do it slow...

 

[TwoCables]

So don`t touch the FSB Clock that`s highlighted ? I thought you would adjust that too.

I thought you did and this was an older photo from before you did.

For 3 GHz, 1333 would be changed to 1500. The FSB can be thought of as having a multiplier of x4.0 at all times (Quad Data Rate, or "QDR"). The E8200's highest multiplier is x8.0. So, 3000 (3 GHz) divided by 8 is 375W. 375 MHz is the actual Bus Speed. 375 x 4 is 1500, so that's the FSB Clock Speed you'd need to enter for 3 GHz.

Just to make it clearer if need be (if even just for someone else reading this), the stock speed of the E8200 is 2.66 GHz which is 2660 MHz. So 2660 MHz divided by 8 is 333 MHz. 333 MHz x 4 is 1333 MHz. Using a calculator, the numbers are slightly different, so I'm using the advertised numbers.

Edit: When you increase the FSB Clock, the memory will go up along with it. I'm not sure about the 750i, but look for an option that might be called "FSB - Memory Clock Mode". One of the options for the setting I'm looking for is Unlinked, and this can help keep the memory closer to 800 MHz, and sometimes even exactly 800.

 

[AddRAM]

No, No, No I haven`t made a change yet. All that is stock.

The 1st change to the voltage I can make is 0.0125V. Should I start at that, Or should I go higher

I only see an " Advanced DRAM Configuration " either enable or disable.
The memory didn`t change in the bios, at least not yet.

 

[TwoCables]

I made an edit to add that if you increase the FSB Clock, then the memory clock will go up with it. If this board has any option to choose a setting that will be called "Unlinked", then use that. My EVGA 680i SLi called it "FSB - Memory Clock Mode", and I think it had the same name in 780i boards too.

 

[AddRAM]

Windows is booting and here`s what we have now. How do the volts look ?
Another question, why do the core speeds and multipliers constantly change ? It does this on all 3 of my Core 2 Duos, I guess that`s normal ?

 

[TwoCables]

Windows is booting and here`s what we have now. How do the volts look ?
Another question, why do the core speeds and multipliers constantly change ? It does this on all 3 of my Core 2 Duos, I guess that`s normal ?

Oh, nice!

I admit that 1.208V is lower than I was expecting. What is it set to in the BIOS?

Anyway, yeah, the constant fluctuation is very normal. If CPU-Z could show this in true real-time, then it would look like the CPU is having a major fit after taking way too much caffeine. Older versions of CPU-Z didn't show these fluctuations enough because they believed no one wanted to see them. I guess they were wrong, but now lots of people are like "Omg, my CPU is going nuts!" hehe

 

[AddRAM]

I only went up 1 notch on the vlotage ( as it were ) in the bios from 0.000V to 0.125V. It changes in incraments of .0125V should I leave it there ?

In the bios It reads

CPU Voltage 0.0125 ( that`s what I can change )
Memory 1.90
VTT FSB 1.200
NB 1.24
SB 1.50

Also, the memory clock didn`t change, it stayed at 800 MHz, the box says it supports 800 MHz so should I change that ? I wrote XFX asking them if it will support 1066 as Tigerdirect claimed it did but the XFX specs don`t say that, that`s why I got PC2-6400.

Oh, and by the way.... You`ve been Repped

So if I wanted to go to 3.2 GHz I would change the clock speed to 1600, is that right ?
Just tried and it wouldn`t boot. I guess it will have to stay at 3 GHz.

 

[TwoCables]

I only went up 1 notch on the vlotage ( as it were ) in the bios from 0.000V to 0.125V. It changes in incraments of .0125V should I leave it there ?

In the bios It reads

CPU Voltage 0.0125 ( that`s what I can change )

Oh, you can try to leave it there, but the 1.208V voltage that it's resulting in is a bit low. Of course, if it turns out to be stable, then that would be nice.

Speaking of which, Prime95 is the most-recommended program for stability testing. For the Core2 series, Prime95's "Small FFTs" test is the best test because it focuses just on the CPU. Most people recommended to me in my days of overclocking the E8400 to tweak until it can run Prime95's Small FFTs test for 12 hours! So, I did and it was worth it. I never had any stability issues. Although, I also added 12 hours of Prime95's "Blend" test and the "In place large FFTs" test too. lol I'm crazy, but I also had my E8400 @ 4.0 GHz!

Prime95 will show you two "Worker" windows, one for each core, underneath a main window that shows the start time of the test and some other useless info. hehe The titlebar of the two Worker windows will show you if the Worker is either Running, or Stopped. If it's Stopped, then it failed. Any failure like that is instability and it just means that you have to do some more tweaking, such as increasing the core voltage or something.

When you go to their site to download Prime95, you might find it a little bit difficult to find the the download link. You might laugh when you see their site. Anyway, I recommend just doing a Ctrl+F for "p95v277.win64.zip (4.6MB)" because it will land you right on the download link so that all you have to do after finding it is press Enter.

Memory 1.90
VTT FSB 1.200

As you go higher and higher on the overclock, the FSB will probably have to be increased to help with the higher FSB clock.

NB 1.24
SB 1.50

Also, the memory clock didn`t change, it stayed at 800 MHz, the box says it supports 800 MHz so should I change that ? I wrote XFX asking them if it will support 1066 as Tigerdirect claimed it did but the XFX specs don`t say that, that`s why I got PC2-6400.

Oh, and by the way.... You`ve been Repped

lol thank you.

When I had my 680i, I was told that I could overclock above 800 MHz if I wanted, but finding the tightest (lowest number) timings would be the most beneficial for nForce chipsets. I was also told that if I could get the Command Rate down from 2T to 1T, then I would feel a difference. My memory at the time didn't want to do that, so I looked around on Newegg for memory that could by searching through customer reviews using Ctrl+F for "1T". Fortunately, I found one fairly quickly. I ordered it, switched to 1T, but it failed. I got it to work after some tweaking though.

So in the end for 1T, I had to increase the memory voltage a lot, and I had to take the main timing, the one that matters the most (other than Command Rate) called CAS # Latency, and loosen it from 4 to 5! Argh. lol It was worth it though. hehe That change from 2T to 1T made a slightly noticeable improvement for general things, like browsing through my hard drive, opening programs, and even surfing the internet. So, any time my memory was used I noticed it a little. That little bit was worth it because I was looking for some tangible difference!

So I guess I can summarize what I learned by saying that it's not the memory's clock speed so much with the nForce chipset, but the timings. Although, thanks to that, I never learned how to overclock memory and I still don't care for it. lol I guess things would have been different if I had started out with an Intel chipset like the P35. The P35 guys were overclocking their memory!

So if I wanted to go to 3.2 GHz I would change the clock speed to 1600, is that right ?

Yep! This should require a slightly higher core voltage and probably even a little bump on the FSB voltage.

Oh, I almost forgot about temperature monitoring: Real Temp!

Real Temp download page: Download Real Temp 3.70 | techPowerUp

You'll notice a "Distance to TJ Max". This doesn't really have to be watched, but it's kind of neat to know what the distance is. The core temps are what matter, of course.

Yeah, this is like riding a bike. I haven't talked about overclocking a Core2 Duo on an nForce board in about 2 or 3 years. I could almost shed a tear. I haven't used that EVGA 680i SLi in about that long. I used to think it was awesome. Well, it was when it came out. So yeah, I'm having some fun here!

 

[AddRAM]

Running Tests I`m doing the Torture Test !!!!

What the hell is Average Distance to TJMax LOL

 

[TwoCables]

Running Tests

What the hell is Average Distance to TJMax LOL

The tests in Real Temp aren't really stability tests.

Anyway, the Average Distance to TJMax is the average temperature spread between the core temperature and the Tj Max temp which is 100°C for your CPU. So in the screenshot, the Distance to TJ Max was 49 and 56. That means it's 47 and 56 degrees away from 100°C. The Tj. Max is the temperature where the CPU is supposed to automatically throttle itself down to cool off. It's kind of like it goes, "Yeah, I don't think so. This is too much. I'm takin' a break to cool off. Sorry."

Real Temp GT is for 6-core CPUs. That's why it's wider. Other than that, it's the same program. The reason for the temperature variations between Real Temp GT and the plain Real Temp is they were polling the CPU at different intervals and caught the CPU at different times.

 

[TwoCables]

Just a quick update: I am looking at going to bed sometime soon, but I don't know when. So I may need to soon begin thinking about finding a point where we can pause and continue tomorrow. Of course, there's always Overclock.net - An Overclocking Community. hehe

 

[AddRAM]

Sleep on a Saturday ? Unheard of LOL !! Alright, thanks alot for your help. I`ll keep playn with it and I`ll seeya in the forum

 

[TwoCables]

Sleep on a Saturday ? Unheard of LOL !! Alright, thanks alot for your help. I`ll keep playn with it and I`ll seeya in the forum

You're welcome. This is fun stuff! When I joined SF, I didn't expect to be seeing threads like this. I love it.

Anyway, I should be back tomorrow morning for more overclocking goodness. hehe

 

[AddRAM]

Got it up to 3.1 still stable

Had to put the volts back down to defaults.