2.8C and 3.0C

[cphowitzer]

After much surfing and researching, I am still unsure whether to get the 2.8C/E or the 3.0C/E... I would very much appreciate your feedback.

I currently just received the new DFI Lanparty i865pe mobo, and have the PC4000 Ultra 1gb, Dual Channeled (2x512). So now I'm looking for a good processor. I am willing to overclock at about 15-25% but preferrably with stock cooling.

Now the confusion comes from people saying that the 2.8 and 3.0 and other iVs are just same chips but overclocked and sold at different price. It makes sence, (businesses would not go through the trouble of overclocking), but is it true? Also Northwood and Presscot seem equally good, although currently I am leaning towards Norhwood, although still, Prescott is newer. I don't know.

Any recommendation would be greatly appreciated, thank you. Im planning to buy the chip tommorow, and be in time for the awesome game coming out... U know it... Doom3... (although im already shaking from fear... ouch).

Thx,
CPH

 

[CTho9305]

Originally posted by: cphowitzer
Now the confusion comes from people saying that the 2.8 and 3.0 and other iVs are just same chips but overclocked and sold at different price. It makes sence, (businesses would not go through the trouble of overclocking), but is it true? Also Northwood and Presscot seem equally good, although currently I am leaning towards Norhwood, although still, Prescott is newer. I don't know.

Welcome to the forums. This question doesn't belong in Highly Technical (try General Hardware or CPU/Processors), but I'll answer that part anyway.

Anyone who says Intel or AMD "overclock" is either ignorant, misinformed, or misrepresenting the truth. When a CPU is designed, a target frequency is chosen (say, 2GHz). Due to manufacturing variations, you get a near-gaussian (normal) distribution, with most parts running at, say, 2GHz, some that are only stable at 1.8, and some that are stable at 2.2GHz. The manufacturer tests all the parts to see how fast they can be clocked and remain stable using test patters and "scan". Note that when they run these tests, they have to do it at the worst-case conditions: something like 90 degrees celcius, and with the motherboard providing the minimum rated voltage (usually 5 or 10% below the ideal voltage). If a part runs at 2GHz, it will be sold as 2GHz.

When you overclock, you take advantage of a few things. First, you presumably don't run your CPU at the absolute max temperature, with the worst voltage supply. Second, if a CPU works at 1.99GHz, but not 2GHz, it has to be sold as a 1.9 or 1.8GHz CPU (since they only sell a few speed grades). Overclockers usually run at higher than rated voltages, and low temperatures.

Also, when a manufacturer tests a CPU for stability, they know where all the potential speed paths / critical paths are in the chip, and run specialized tests to make absolutely sure the CPU will not perform incorrectly at its rated speed. Overclockers generally run a few programs, and then consider a CPU stable. "Overclocker" stable and manufacturer stable are two different things.

Note that some people claim that when yeilds are better than the market demands (i.e. a large number of the CPUs work at 2.2 or 2.4 GHz), the manufacturer will sell some of the faster parts as a lower rated speed. I don't work in a department where I'd know if there's any truth to this claim.

Anyway, I would define overclocking as increasing frequency beyond the speed it passes manufacturers tests. Therefore, if Intel sells a 3GHz chip, and a 2.8GHz chip, it's likely the 2.8 did NOT pass their more-strignent tests at 3Ghz. Note that sometimes you'll see higher speed grades being rated at higher voltages (If I remember correctly, OLD athlons ran initially at 1.65 volts, but as the speed grades got higher they bumped it up to 1.7 volts). Some people might call this overclocking.