Sandy Bridge

[Sticky]

Sandy Bridge, anyone else interested in this? Im thinking of getting an i5 2500K (its the one with the unlocked multiplier) in january IF the price is not too high, should be perfect for DCS:A10, its time for an CPU upgrade for me, Im still on socket 775 :( So this should be a good upgrade for me :thumbup:

[EtherealN]

I'm actually very torn about the K series Sandy Bridge. The multiplier is unlocked, sure, but the base clock is locked instead. This means that as you move the multiplier up you cannot scale your memory controller with it, potentially introducing a whole new bottleneck.

 

So I'm on the fence in anticipation of some good tests of this factor, for now.

[Sticky]

FSB is still free to change, but they moved the memory bus clock away from that with 1156 and 1136, right? So changing the FSB wouldnt affect the memory bus?

 

Edit: Wrong facts, FSB is not free to change anymore, according to EtherealNs link

Edited December 24, 2010 by Sticky

[Sticky]

Sorry if my assumptions are wrong, I havent built anything since 775 and its not fresh in my memory.

[EtherealN]

Regarding "FSB"/bClock: http://en.inpai.com.cn/doc/enshowcont.asp?id=7947&pageid=7706

 

In theory, unlocked multiplier might outweigh the locked bClock, but this does require some real-life tests to verify, and since these simulators are a bit different to most games in their harwdare tasking it becomes a bit more difficult to infer from tests of 7zip compression or BC2 tests.

[Sticky]

K, thx, will look at that.

 

Anand says "Finally, if you focus on multiplier-only overclocking you lose the ability to increase memory bandwidth as you increase CPU clock speed. The faster your CPU, the more data it needs and thus the faster your memory subsystem needs to be in order to scale well. As a result, on P67 motherboards you’ll be able to adjust your memory ratios to support up to DDR3-2133."

 

Im thinking about that last line, would that solve the bottleneck problem you wrote about?