EVGA Teases Community with Upcoming LGA2011 Motherboard Prototype

[FordGT90Concept]

Because Memory IO pins are on one side of the processor package. With LGA2011, I guess there are MemIO pins on either sides.

Additionally, it should have less latency to the second pair that would otherwise be located farther away than the first pair. Unless...

They might only have 2 DIMMs per CPU so the other CPU is just outside of that picture. Then again, it could have something to do with those new quad-channel memory controllers that have yet to debut (excluding 771, of course). We'll have to wait and see as always...

 

[btarunr]

Additionally, it should have less latency to the second pair that would otherwise be located farther away than the first pair. Unless...

They might only have 2 DIMMs per CPU so the other CPU is just outside of that picture. Then again, it could have something to do with those new quad-channel memory controllers that have yet to debut (excluding 771, of course). We'll have to wait and see as always...

Latency is less of an the issue. The issue here is wiring. since the MemIO pins are on one side of the package, it's easy to wire them out to the DRAM slots. If you wanted DRAM slots on either sides of the socket (like on LGA2011), you would have had to add many more layers to the PCB, just to wire them under the socket and across to the other side of it, not to mention interference from the relatively "high voltage" zones of the socket. Even the slightest interference will result in hardware faults for the memory. With LGA2011, I think there are MemIO pins on either sides of the package, and VCC pins on the "top" side (which is where most boards are showing Vreg area).

 

[mastrdrver]

its ugly but lower latency for quad channel... even feeeeed of ram bandwidth to the cpu....

Just so fud doesn't get spread (though I know you're not referring to it). Quad channel does not have lower latency then triple or dual channel fwiw.

Also the latency issues that result from dimm slots being further then others is solved by making the wires to the closer slot longer then they need to be. How ever slight it may be.

 

[n-ster]

Additionally, it should have less latency to the second pair that would otherwise be located farther away than the first pair. Unless...

They might only have 2 DIMMs per CPU so the other CPU is just outside of that picture. Then again, it could have something to do with those new quad-channel memory controllers that have yet to debut (excluding 771, of course). We'll have to wait and see as always...

all 4 DIMM slots are for that 1 CPU, as you can clearly see in the pic on the top right that there is no socket and the DIMM slots wouldn't be further away from one CPU than another.

and I'm guessing that if the layout were to have 2 DIMM slots per CPU, the logical thing to do usually is to put the 2 pairs of slots next to each other like we see in some dual socket motherboards