Help building a computer

[Ewhenn]

Here's the result of my OC to my 2500K, with 12 hour Prime 95 Stability test.

(CLICK FOR BIGGER PICTURE)

Overclocked from 3300 MHz to 4400MHz (33% overclock). If I wanted to I could push it farther as my temperatures are still good and my voltages still have some room. I'd imagine I could get 4600 MHz stable before going past the safety zone on voltages/temperatures.

There are two voltage methods for Sandy Bridge:

Fixed Voltage - The voltage stays the same all the time no matter what speed the chip is going.

Offset Mode - The voltage changes depending on how fast the chip is going.

By default sandy bridge CPU's "slow down" when you aren't using the processor (ie. sitting there reading a loaded up website).

My advice is to overclock with a fixed voltage to start, and find a decent overclock, then consider moving that over to an offset voltage.

Example, if by default the processor requests these voltages at these speeds:

1600 MHz, 0.988v

2700 MHz, 1.102v

3300 MHz, 1.246v

4400 MHz, 1.410v

If I set an offset of: -0.090v, that would mean the processor would get:

1600 MHz, 0.898v

2700 MHz, 1.012v

3300 MHz, 1.156v

4400 MHz, 1.320v

Basically the offset voltage gets applied to every speedbump's voltage request along the way.

The part that makes offset overclocking harder to get stable is that while it may work out so that the voltage change is stable at the top speed, your lower speeds might have a problem, for example, 0.898v might not be enough for your particular chip to run at 1600 MHz. So when you overclock this way, you need to make sure every step is stable.

Also, if you use offset overclocking, you will generally be using negative offsets, since the chips usually request way more voltage than they actually need to run at a given speed.

If you set a manual voltage of 1.320v, it would look like this:

1600 MHz, 1.320v

2700 MHz, 1.320v

3300 MHz, 1.320v

4400 MHz, 1.320v

The benefit of manual voltage overclocking is you only need to worry about stabilizing the top speed, which makes it very easy to do. The negative is that at the lower speeds, it still gets full power (power it doesn't really need). As long as your voltages aren't really high this shouldn't matter, but it's something to keep in mind.

On the topic of voltages:

For safe overclocking of Sandy Bridge chips it's generally accepted to not go over 1.380v for 24/7 daily operation. Personally, I'm even a bit more cautious, I won't go over 1.340v.

You want to try limit voltage to what is as low as possible to run at your desired speeds, since more volts = more wear and tear on your chip, and this ramps up exponentially. Also more volts = more heat (This is the reason for getting a better heatsink when you overclock, they handle the extra heat).

I'm going to be blunt: Yes, you will reduce the life of your CPU, but it won't be meaningful. CPUs are engineered to have a lifetime of 15 years or more. If you decrease the lifetime of your CPU from 15 years to 10 years by overclocking, it's not a big deal. You simply aren't going to be using the same PC 10 years from now. As long as you stick to those guidelines with voltages, you shouldn't ever have a problem. I've overclocked all of my PCs for 10+ years and have never had one fail prematurely (I still use my overclocked 7 year old Athlon as a file server), the key is just not to over do it.

If you end up buying an AsRock board, I can tell you exactly what to do to overclock it (safely).

Edited September 4, 2011 by Ewhenn

[DragonRabbit]

wow temperature so low at 4400MHz

[Deathstar]

I'm going to be blunt: Yes, you will reduce the life of your CPU, but it won't be meaningful. CPUs are engineered to have a lifetime of 15 years or more. If you decrease the lifetime of your CPU from 15 years to 10 years by overclocking, it's not a big deal. You simply aren't going to be using the same PC 10 years from now. As long as you stick to those guidelines with voltages, you shouldn't ever have a problem. I've overclocked all of my PCs for 10+ years and have never had one fail prematurely (I still use my overclocked 7 year old Athlon as a file server), the key is just not to over do it.

mine just had a tenth birthday.

damn your screen is huge.

[Ewhenn]

wow temperature so low at 4400MHz

Combination of a couple things:

-Lowish room temperatures, my PC room is about 75F in the summer.

-A decent Heatsink, CM 212+, not great, but good.

-Very good case airflow. 2x 200mm fans and 4x 120mm fans. In conjunction with below....

-Very good cable management. There are literally no wires that restrict airflow, everything is ran behind the motherboard/along edges.

I have lots of room temperature wise to overclock higher, I just don't want to push more voltage through my CPU to do it. My chip is actually slightly below average, most can get around 4500-4600 MHz at 1.340v or lower. Every chip is different, just the luck of the draw. /shrug

mine just had a tenth birthday.

Your case isn't normal, most people don't use the same PC for 10 years.

damn your screen is huge.

Not that big, 25" @ 1920x1080 resolution.

Edited September 4, 2011 by Ewhenn

[Hess]

I have a friend who is telling me that the h80 with push/pull has a 4 degree difference as far as cooling goes compared to the h100 with just push.

I'm really liking the h100 though I was able to get it for a relatively good deal, the same price as an h80 so I hopped on it, but the return is free so if I had to I could always return it and order the h80, however as you said surface area > slightly more airflow.

[Ewhenn]

I have a friend who is telling me that the h80 with push/pull has a 4 degree difference as far as cooling goes compared to the h100 with just push.

I'm really liking the h100 though I was able to get it for a relatively good deal, the same price as an h80 so I hopped on it, but the return is free so if I had to I could always return it and order the h80, however as you said surface area > slightly more airflow.

That would surprise me if it was right. Surface area is pretty much king when it comes to water cooling.

[Deathstar]

I have a friend who is telling me that the h80 with push/pull has a 4 degree difference as far as cooling goes compared to the h100 with just push.

I'm really liking the h100 though I was able to get it for a relatively good deal, the same price as an h80 so I hopped on it, but the return is free so if I had to I could always return it and order the h80, however as you said surface area > slightly more airflow.

That would surprise me if it was right. Surface area is pretty much king when it comes to water cooling.

so let's say we have 1 gram of water (for reference that's 1/1000 of a liter)

then 1 gram of air ( .775 liters)

Our 1 gram of water can support 4.2 joules of energy before its temperature is raised by 1 degree celcius

our 1 gram of air can support about 1 joule of energy before its temperate is raised by 1 degree celcius

or the same volume of water can support 775*4.2 = 3255 more times the energy than the same volume of air.

I also believe water is about 50-100 times better at transferring heat than air.

[Ewhenn]

I have a friend who is telling me that the h80 with push/pull has a 4 degree difference as far as cooling goes compared to the h100 with just push.

I'm really liking the h100 though I was able to get it for a relatively good deal, the same price as an h80 so I hopped on it, but the return is free so if I had to I could always return it and order the h80, however as you said surface area > slightly more airflow.

That would surprise me if it was right. Surface area is pretty much king when it comes to water cooling.

so let's say we have 1 gram of water (for reference that's 1/1000 of a liter)

then 1 gram of air ( .775 liters)

Our 1 gram of water can support 4.2 joules of energy before its temperature is raised by 1 degree celcius

our 1 gram of air can support about 1 joule of energy before its temperate is raised by 1 degree celcius

or the same volume of water can support 775*4.2 = 3255 more times the energy than the same volume of air.

I also believe water is about 50-100 times better at transferring heat than air.

err.. it's about 2 different water coolers

One with a smaller radiator, but with 2 fans total, a fan pushing air in AND sucking air out (lots more airflow). 120mm x 120mm size.

Another with 2 fans pushing in, no fans sucking out, twice the surface area, 120mm x 240mm.

As long as there aren't any factors compounding it (like the first one having a better CPU contact solution), the second option should be better.

[Hess]

I feel as if the air flow from my case will be blocked by the bigger radiator to where the h80 will fit nicer.

If you want to look at the case I'm getting its herehttp://www.corsair.com/cases/carbide-series/carbide-series-400r-mid-tower-case.html

[Hess]

went ahead and bought the asrock motherboard

[Res]

I've heard asrock is a shitty brand.

[Ewhenn]

I've heard asrock is a shitty brand.

Was a shitty brand.

I wouldn't have advised buying their products 2 years ago. They've stepped up their quality since then, it's now on par with Asus, Gigabyte, and MSI.

[Res]

Oh, I still wouldnt buy from them then. I'd rather buy from someone that has a longer track record of 'not being shitty'.

Just my personal opinion

[Sephiore]

Oh, I still wouldnt buy from them then. I'd rather buy from someone that has a longer track record of 'not being shitty'.

Just my personal opinion

I'd listen to anything Ewhenn ever says about computers man.

[Muse]

Now posting from my just-built-3-hours-ago SCREAMING fast computer.

Thanks for the help guys, especially Ewhenn.

[Erador]

No pics no proof!