will disabling bulldozer cores help?

[Greenlepricon]

I'm pretty limited in my cpu knowledge, so I'm sorry if this is a silly question. I have a 8120 cpu and am wondering if disabling half of the cores (one for each pair) will help me. I only use my computer for gaming and normal internet/program use. Nothing too intense, and so I normally don't have a need for so many cores. It seems like the available cache should help, and I might be able to achieve a higher overclock with less power, but this is speculation for me. I would really appreciate any input. Thanks guys.

 

[OCGuy]

Absolutely not.

 

[T_Yamamoto]

No

 

[Accord99]

I'm pretty limited in my cpu knowledge, so I'm sorry if this is a silly question. I have a 8120 cpu and am wondering if disabling half of the cores (one for each pair) will help me. I only use my computer for gaming and normal internet/program use. Nothing too intense, and so I normally don't have a need for so many cores. It seems like the available cache should help, and I might be able to achieve a higher overclock with less power, but this is speculation for me. I would really appreciate any input. Thanks guys.

Maybe a little, on release Bulldozer set to 4 module/4 core mode showed a few % higher performance in games versus the default 4 module/8 core mode:

http://www.behardware.com/articles/842-9/amd-fx-8150-and-fx-6100-bulldozer-arrives-on-am3.html

However, the updated Windows 7 patch may have already fixed this.

 

[Greenlepricon]

Alright thanks. I had read somewhere where someone was claiming it helped. Thanks for the responses

 

[Ferzerp]

There was some anecdotal evidence suggesting a minor single threaded increase in performance because the original scheduling treated a module as 2 cores (which it really isn't, two threads on 2 modules runs faster than 2 threads on the same module). A patch was released that, as I understand it, treats each module the same as each HT Intel core is treated, so if there was a benefit before, it is now gone.

 

[Greenlepricon]

There was some anecdotal evidence suggesting a minor single threaded increase in performance because the original scheduling treated a module as 2 cores (which it really isn't, two threads on 2 modules runs faster than 2 threads on the same module). A patch was released that, as I understand it, treats each module the same as each HT Intel core is treated, so if there was a benefit before, it is now gone.

Huh interesting. So that's probably what I had read. Either way if that's fixed then there's no point to trying I guess.

 

[StokesMSFT]

Make sure you apply http://support.microsoft.com/kb/2645594

As I think previously stated.

 

[Ferzerp]

That's the one I'm talking about!

edit: Amusingly, some folks heard rumors of this patch and assumed super-duper massive performance increases, when in fact, it was more making the scheduler match the real world (instead of marketing) so that you only use both execution paths on a module when all other modules are already busy. It works better than HT (which we would hope, given its much higher xtor cost), but at its core, they share much. Sure, HT is more about unused cycles, but they are both about resource sharing, and any scheduler that treats each logical core the same as a discrete physical core can kill performance if it stacks threads on the same core/module instead of spreading them across all available cores/modules.

 

[Greenlepricon]

Oh cool thanks guys! I really appreciate the help.