Sandy Bridge Turbo 2.0 technical questions

[RFish1]

I'm a system builder looking for information on Turbo 2.0 on the Sandy Bridge platform. I understand that new methods of implementing Turbo are upon us which can use the shock load of the heatsink along with rapid polling. My question is regarding how Turbo adapts to different amounts of thermal dissipation available to the processor by using aftermarket heatsinks.

If I build a system with a much larger heatsink would Turbo take advantage of this and stay clocked higher for longer as it has more thermal dissipation available to it? Would Turbo allow the processor to stay above TDP for longer as the heatsink will have a greater thermal capacitance?

Thank you.

[RGiff]

Don't build a SYS with a much larger heatsink. Go H2O , Cousair has Two units that work fine on 130 watt CPUs , Model H-50 and model H-70 , I have both very dependable , Keeps my QX9650 which stays clocked at 3.8 , right at 40C, all day and all night.

[idata]

I most likely won't be overclocking as I cannot find a ITX board avilable. I'm stuck with something like a Scythe Shuriken for my builds as nothing else will work other than stock. My question regards the stock Turbo 2.0 operation only but I thank you for your responce.

[DSilv11]

A generic answer OK?

Turbo Boost, as currently implement, is opportunistic and will engage if 1) Power (based on the processor current draw curve (Imon curve) built into the board by the designer), 2) thermal head room are within specifications. In practice the Imon curve is normally the limiting factor, not the thermals so adding a larger heat sink or more efficient heat sink usually will not gain you anything as related to Turbo Boost. In all the boards I have looked at Imon drops out of turbo boost mode with between 10 to 30 degs C remaining till Tcase max. It is after the system has left turbo boost mode were the heat sink design and chassis cooling become critical factors in staying out of a thermal throttling condition. A very rough rule of thumb \ quick check of Turbo Boost is to increment CPU stress one core at a time while monitoring the processor frequency.Depending on the processor under test, Turbo Boost will drop out completly sometime before the last core begins to be stressed. (~ 80 to 90% CPU loading) (The OS also complecates this since it thinks it should use the processor's rathar than just the controlled stress program so loading is difficult to judge. ie a rough rule of thumb)

This is how it functions in the current Westmere generation products. I don't know if the Sandy Bridge Turbo will change this.

(My New Year resolution is to finish reading the Sandy Bridge specifications.)

(If you need to develop a headache, you can read up on Imon as it is implemented in the current products here: http://www.intel.com/assets/PDF/designguide/321736.pdf http://www.intel.com/assets/PDF/designguide/321736.pdf)