The difference in OCing a CPU with unlocked multiplier

[Papagaj]

Hello everyone

I might sound stupid and lame.... but anyways:)

Please what is the difference in OCing lets say 975 EE on 4GHz

A:) 133*30 = 3990 Mhz
B:) 148 *27 = 3996 Mhz

I mean.... both would result in 4Ghz final OC. So why in some reviews of 975EE.... they are raising just multiplier like guru3d and in some they are raising bus speed with multiplier?

Why do they do that with unlocked CPU?

Thanx for explanation....

[jcniest5]

If I'm not mistaken, there's advantage to OCing with Bus Speed. When you OC with Bus Speed, your memory also benefits (increase in speed) whereas if you just raise the multiplier, only the CPU gets the benefit, not the memory. Having said that, there are some boards that just simply can't OC with bus speed...thus when raising multiplier comes in handy. So, if your board can OC FSB (referred to a bclck on i7) really well, bumping up the multiplier will benefit the overall speed even further. In other words, you get the best of the both worlds comparing to just one.

[unclewebb]

There's a 101 variables that all interact with each other. The only way to find out what's better is to run some benches of whatever software is important to you at each setting.

[Leeghoofd]

in your case : the higher FSB will yield a higher performance of the rams ( depending on the divider you choose ofc and the speed ya rams can take ) , same goes for the uncore speed which is also linked to dividers

[SoulsCollective]

OCing with multiplier means you can leave RAM, uncore and QPI speeds all at a certain level - which means that if you're going for very high CPU clocks, you aren't limited by the other components in your system.

[LinkinParkBoy]

200 x 20 = 4000Mhz CPU
200 x 8 = 1600Mhz Ram

[Papagaj]

200 x 20 = 4000Mhz CPU
200 x 8 = 1600Mhz Ram


My 975 suppose to arrive tomorrow so if i understood that correct... i just change multi to 30 and memory multi to 12 ( it should be unlocked too... the highest memory multi for locked CPU is 8 i think)

So in final.... we have 133*30 ( 4000Mhz) for CPU and 133*12 ( 1600MHz) for Memory.... is that right? Or Am I just noob ? :)

[Xello]

So in final.... we have 133*30 ( 4000Mhz) for CPU and 133*12 ( 1600MHz) for Memory.... is that right?

Yes.

Every clock is based on the bclk - that includes CPU clock, memory clock, QPI clock, and uncore clock, and uncore especially can increase memory speed and L3 cache performance.

Basically the point is that since changing bclk changes every one of these clocks based on their respective multipliers, it's beneficial to have the choice of changing the cpu clock through either its own multiplier or the bclk.

You say you want 133x30 for 4000mhz, but what if you wanted to take it higher, say 133x32 for 4.25ghz? But what if you got a chip that couldn't manage to clock that high (unlikely, yes) the next lowest would be 133x31 = 4.12ghz. Maybe your chip could manage 4.2ghz though? Then you would change the bclk to, say, 140, for 140x30 = 4.2ghz. Of course an extra 10mhz on the bclk would raise your other clocks slightly - your memory from 1600 to 1680mhz, which is fine assuming your memory can handle it, if not you could pump up its volts and see if that does the trick. A simpler way for 4.2ghz would be a bclk of 200, with x21 on the cpu, and x8 on the memory for 4.2ghz and 1600mhz exactly. That's what people with locked processors do because we can't pump the multi up to 31 :)

As you can see it's quite a complex process to talk about in text, but it's quite simple when you're doing it by yourself and after a day or two of tweaking the bios you should get the hang of it :up: