How does Windows keep all the threads on the turbo'd cores?

[soccerballtux]

Wondering if the bulldozer benchmarks might go any different.

 

[RobDickinson]

It does by default I think? Just loads up threads sequentially.

What it cant do as yet is spread them around to make use of 'modules' , which I assume means lower clock speeds but hopefully (for AMD) spreading the load across shared parts of each module.

 

[soccerballtux]

It does by default I think? Just loads up threads sequentially.

What it cant do as yet is spread them around to make use of 'modules' , which I assume means lower clock speeds but hopefully (for AMD) spreading the load across shared parts of each module.

So if you do a 32m superPi run without manually setting the core affinity, it sticks to the turbo'd cores? I thought Windows moved the thread around at high speed to even it out across the chip? Thermal reasons or something.

 

[RobDickinson]

Dunno , I'm just guessing. Should I care?

AMD have had 5 years of development to look at getting win7 patched in its favour.