Overclocking i5 2500k + p8p67, am I doing this right?

TheGuitarist

Distinguished
Jan 29, 2009
97
0
18,630
I read the overclocking guide - for the most part - and realized it said that the settings were for an MSI board, so I stopped and started searching threads on other tech forums for my particular CPU + MB combo. For a mild overclock, I read countless suggestions all stating that all I really needed to make sure of is make sure the internalpll was enabled and change my core muliplier. I ended up doing that and ran a 44 multipler for each core and verified through CPU Z what my current speed is. Is it really that easy? At what point do I start changing voltages? Obviously, I want to be MUCH more careful with my research and application if I'm going to do that, but I was thinking that a 4.6 OC using my same methods would be OK without changing voltages.

Furthermore, how long should I run prime95 for? Also, how the *@&*(&!$ do I install memtest? I read the directions, but do I really have to burn it to a CD and run it from there? Guess I'm just confused as to why that is and if it's really all I have to do?

Temps before OC were as low as 22C and averaged 28-30C. After the OC it's running 32-34C.

Any other "HEY! STOP THERE BEFORE YOU *(%$(*&$# STUFF UP" suggestions or stress test/system monitors to use would be appreciated. Thanks!


Oh, the relevant info is here:

i5 2500K
Asus P8P67
G.Skill Ripjaws 8GB 1600
Hypermaster 212+
Samsung Spinpoint f3 1tb
that 18 dollar ASUS DVD burner from newegg
Sapphire 6950 flashed to 6970

 

whitey_rolls

Distinguished
Oct 29, 2010
135
0
18,690
From what I understand (and I'm kind of in the same position you are) but until about 4.4 ghz you can essentially just change the multiplier.

Currently I have a 2600k and am at 4.3ghz buy just changing the multiplier. In regards to voltage, it really doesn't need to be changed.

I've monitored the voltage and 1.3v is where my CPU will cap out if I just leave everything as auto. If you look at the people doing the higher overclocking they aren't going much past that anyways. When I run prime 95 the voltage is around 1.26v at 4.3 ghz and will drop as the CPU isn't used.

Based on that information I would assume everything is fine as is, Intel just took the difficulty out of overclocking and is kind of throwing people off lol.


Oh and in regards to Prime95, people say run it overnight. I just left mine on for a few hours with no errors, I've been gaming etc no crashes so I'm going with it being fine. If you really want to be sure leave it for 12 hours.
As far as temps go, Just to give you a benchmark mine runs 55C max at 4.3 ghz while on prime 95 for an hour. My cpu cooler is the Corsair H70.
 
Here's a good guide for your board: Official ASUS P8P67 Series Overclocking Guide and Information

And yes, it really is that easy overclocking Sandy Bridge.

Memtest runs from the CD so that an operating system doesn't get in the way of the testing.

For a truly complete stability test:

24 hours of Prime95 set to Small
24 hours of Prime95 set to In-place large
24 hours of Prime95 set to Blend

Why all three? Because each test stresses a different area of the CPU. If you want to cut that time down a bit, just do the 24 hours set to Blend -- it will test all three settings for a smaller amount of time.

Why 24 hours? Because many people have had failures after 15 or 18 or 20 hours, showing that it wasn't truly stable.

Guidelines for keeping your Sandy Bridge happy:
Keep CPU voltage below 1.4v for an everyday overclock.
Keep temps below 75ºC peak for an everyday overclock. Use a program like Core Temp that gives temps for individual cores.

If you keep the CPU voltage on Auto, it will increase and decrease depending on load. Sometimes, it will give your CPU more voltage than it really needs, increasing the temps. I recommend setting the voltage manually to 1.2v (the default for Sandy Bridge) and seeing how high you can set the multiplier. If you get instability (Prime95 fails right away or within an hour or so) then increase it in 0.025v increments until it's stable again.
 

th1nker

Distinguished
Mar 12, 2011
16
0
18,510
Wow, this is exactly the thread I was looking for, I'm in the same boat and trying to look for good threads and vids about overclocking i5 2500 + p8p67. Currently in the research phase. Question though, do you have to run prime95 if somebody else used the same settings with the same processor/mobo combo without failure - or will it work the same in every computer with the same chip/mobo?
 
Each individual board and CPU will yield different overclocking results. You need to do your own discovery as to what your particular combo will do. We have several forumites here that have really crappy Sandy Bridge chips for overclocking -- they can only get to 4.2GHz stable. Others are like me -- 2500K CPU running at 4.8GHz currently. A few get higher than that. The average is about 4.5GHz or so.

Prime95 is a really good stability test because the "Small FFTs" and "In-place large FFTs" settings use different portions of the CPU, and the "Blend" setting tests them both while also testing the memory controller. It requires the CPU to report back with an exact calculation and if it doesn't you know right away. If your CPU isn't stable it will give a bad result and Prime95 will notify you immediately so you can take steps to correct it. If an unstable CPU throws out enough bad results, it can result in data corruption in real-life programs.
 

Joeypav

Distinguished
Apr 12, 2011
1
0
18,510


I am at 4.8 max temp 59 c and only 1.26v 2500k. Question is since I am on manual voltage will running at 1.26 24/7 damage the cpu