High Turbo Ratio vs Overclocking Sandybridge

87ninefiveone

Distinguished
Oct 16, 2011
449
0
18,860
So, I've just discovered that my motherboard (MSI Z68AGD80) allows adjusting the turbo ratio that the CPU (2600K) applies when 4, 3, 2, or 1 core is in use. So, my question is why even bother overclocking when you can just adjust the turbo ratio to get scalable performance.

I've successfully overclocked my CPU to 4.5GHz, but I've also been able to set my turbo ratio to the following and have it work just as well...

4 Cores = 39x
3 Cores = 41x
2 Cores = 43x
1 Core = 45x

I realize that by doing this I don't get a consistant 4.5GHz, but really who needs it? I've run Prime95 testing with differing numbers of threads and adjusting the turbo multipliers does seem to work as intended. So, why bother with traditional overclocking?
 

chesteracorgi

Distinguished
The benefit of OC'ing is only seen in extreme apps like CPU intensive games (e.g. flight simulators). For the great majority of uses OC'ing makes no difference.

Unless you are doing untimate gaming with AA and rez turned to ultra settings (and even then OC'ing the GPU gives better results than CPU OC'ing) the i5 & i7 processors can do the job without OC'ing in 95% + situations.
 

87ninefiveone

Distinguished
Oct 16, 2011
449
0
18,860


Right, I would agree completely. There's really no need to OC at all except that it's fun and easy. I just don't understand why there isn't more discussion out there about adjusting the turbo multipliers to achieve a high dynamic overclock as opposed to a static overclock (especially in light of what you've said, since no one "needs" the OC performance at all times).
 

kAzure

Distinguished
Dec 3, 2011
9
0
18,510
from what you've described
( 4c @ 39x or 3c @ 41x or 2c @ 43x or 1c @ 45x)

you've oc'd the core frequency to what? 3.7Ghz, with turbo settings of +2, +4, +6, +8?

I'd assume that this would be better for the longevity of the processor, due to lower temps @ idol, and better efficency.
Plus if you were able to utalise faster speeds under load, then the processor would provide this within limits. Sounds good to me.

What i'd like to know is how 'determined' the processor is in regards to achieving the (+2, +4, +6, +8). as in, what temp/voltage will the chip decide is too much and drop back a level?

Also in the case you have got it working, does it choose either (+0 OR +8) with a single core, or does it move in increments, and if so, increments of what, 200Mhz or 100 Mhz in the case of (+2, +4, +6, +8)?

( i hate to be doing something where it ''switches" back and forward from 3.7Ghz to 4.5Ghz, be it preformance testing, or a game that relys on that performance factor. )

I would think that this would be a good option to save power while doing less intensive things, while being able to jump straight into doing something 'intensive', without having to change settings. But if you're willing to go to a certain temperature/voltage( or whatever the attribute change that accompanies the performance increase ), but the automated turbo process wont go that far( or say heats up in the process and has to scale back ), then you're faced with a problem and a choice...

To be honest I would like to know more about this as well.
Sorry to add more questions, but somebody out there would probibly know the answers...
 

87ninefiveone

Distinguished
Oct 16, 2011
449
0
18,860
My base clock is set to the normal 3.4GHz. From what I can tell using both CPU-Z and Prime95 the CPU does a lot of switching between 1.6GHz (idle), and up to 3.4GHz during light tasks and typically 4.2GHz under load. After watching it for a while and playing with it a bit it seems pretty rare to see it hit 4.5Ghz. I kind of wonder if the Turbo frequency has an unspecified maximum number of bins that it will clock up to. It's been difficult to determine since the frequency jumps around so quickly.

As for voltage, I've got my MB set to auto and it ranges anywhere from 1.15 to 1.32V which seem pretty reasonable to me given that it's on auto.