g5insider :
After reviewing the parts for your pc build i have just a couple of questions to ask...
Why sparkle for your PSU?
Why G.Skill for Memory?
Why Sapphire For GPU?
Why MSI for your mobo?
If you want to game spend money on the GPU and leave the CPU at the recommended clock speed which is 3.3GHz.
By over-clocking your chips you significantly decrease the life of you components with little gains
Water cooling would be optimal to produce the gains you want from over-clocking but your GPU is gonna do all the major calculations so make sure you get a good GPU
I've never used G.skill so i dont have much to say about memory
Oh, ok let me clarify a few things for you and the OP.
1. Sparkle is not a bad PSU company. Their OEM is great wall which has made many good units. Jonnyguru reviewed the 1200W unit in that line.
http://www.jonnyguru.com/modules.php?name=NDReviewGs&op=Story5&reid=212
However, for the money, might as well grab a seasonic x-750
http://www.newegg.com/Product/Product.aspx?Item=N82E16817151087&Tpk=seasonic%20x750
Seasonic is the top PSU OEM right now, the Corsair AX750 is identical to that x750 cept more money.
2. G Skill is the brand of choice on these forums. Low voltage, overclocks extremely well, excellent customer service and some of the best prices.
3. Overclocking doesn't actually decrease system life as long as you keep voltages and temps under control. Using an aftermarket cooler and ocing to w/o a v could actually increase the system life expectancy simply because the aftermarket HSF is keeping temps lower than the stock HSF. Regardless, CPU's are designed to last over a decade so it's not like you've got much to worry about. Detailed explanation of the ECE stuff behind this below.
How OCing can damage a CPU
Normally, electrons stay around their atom's and don't go wandering
off. So in a CPU, they'll stay in one transistor and not move to
others. However, if you've learnt quantum tunneling principles, you'll
know it's actually possible for electrons to escape from energy wells,
even infinitely deep ones, it's just very uncommon.
Now, a transistor in a CPU is made from alternating + and - doped and
undoped silicon. Once in a while, an electron will escape and bury a
couple atoms into an adjourning transistor, and if this happens enough
times, eventually all the way through to the adjourning transistor
before coming back to it's orbit.
Keep doing this and eventually an electron doesn't come back, but
stays attached to an atom in the adjourning undoped section of
silicon. Over time (usually years), this tunneling causes a hole to be
formed between two adjourning transistors and allows free electron
flow. This bypasses the "gates" between the transistors and as a
result, the computer will misread this resulting in an error.
This process is called silicon degradation and eventually results in a
complete CPU failure.
Now, as to where overclocking comes in.
If you know about electron orbital theory, the more energy an electron
has, the more likely it is to leave it's orbit and tunnel. IE if yur
CPU is running hot, or has a considerably higher voltage going through
it, electrons tunnel in much higher numbers. As a result, the more you
OC, the faster you make those tunnel which cause silicon degradation.
In addition, if you increase the voltage enough, you can actually
physically destroy the silicon lattice of the gates within a
processor. This is VERY VERY high though, much more than you'll probably be able to boot on, so I wouldn't worry about this.
OC and Heat
Boosting F, has a very minor, almost insignificant heat increase.
It's v increase that dramatically increases heat.
Power Dissipation = PD in Watt
Voltage = Volt
Freq = Hz
C= Capacitance in Farads
Total PD in Watt = C x F x V^2
As C doesn't change (ok it technically does, but for the sake of
keeping the math simply we can assume it doesn't)
If you actually plug in numbers and graph the function, the heat
increase due to a freq increase is minute compared to the heat
increase from a v increase, as one increases exponentially, the other
linearly.
Indeed, the more you increase the V, the less the F part of the
equation is relevant to the total temp.
Looking at real world data, look at the power usage increase in Tom's
i5 efficiency article.
http://www.tomshardware.com/review [...] 500-7.html
Each bump was a constant 10mhz clock speed increase, but due to the
exponential nature of the voltage increase contribution to PD, the
graph is not linear, and power usage does not increase until you start
seeing large v increases.
Power usage directly translates into heat.
As for actual temps, it's more complicated than purely based on power
dissipation
Cpu temperature = (Total PD in Watt) x (HSF's Thermal Resistance in
C/W) + (Ambient Temp in Celcius)
For comparison purposes the resistance and ambient can be considered
constant (technically not true once again, as resistance changes
slightly with temp, and ambient increases with more heat output).
Intel also has a nice presentation on how they determine/balance heat, noise and fan control here:
ftp://81.30.226.136/skoleni/mb,%20cpu,%20mem/Tcontrol%20IDF.pdf