Contribute
Register

P55 mobo 1156 memory optimization (4x2 vs 2x4)

Status
Not open for further replies.
Joined
Feb 23, 2010
Messages
269
Motherboard
GIGABYTE P55A-UD4P
CPU
i7-875K
Graphics
Zotac GTS250 eco
Mac
  1. MacBook Air
  2. MacBook Pro
Classic Mac
  1. 0
Mobile Phone
  1. iOS
I don't know how the cpu communicates with the RAM on these motherboards. With dual channel motherboards and 8GB DDR3, with all things equal in the memory speed and latency:

would FOUR 2GB sticks be faster than TWO 4GB sticks?
 
carbine said:
I don't know how the cpu communicates with the RAM on these motherboards.
The "big news" about the Core i7 family was the inclusion of an Integrated Memory Controller...

See the P55 vs X58 architecture block diagrams below (and how it has nothing to do with RAM).
p55_diagram.gif

X58_blockdiagram.gif


With dual channel motherboards and 8GB DDR3, with all things equal in the memory speed and latency:

would FOUR 2GB sticks be faster than TWO 4GB sticks?
Great question...
It all comes down to whether the memory controller interleaves the banks (not channels) or not...
The answer can probably be found in Intel's Core i7 Integrated Memory Controller spec sheet...

Yet, another consideration is upgradability...
Having 2x4GB will leave room for another couple sticks.

Out of curiosity, if you find the answer, please post back.

Cheers,
Lnx2Mac
 
From latency tests and PI runs I've done in the past, results were the same with either setup. The only big difference between 2x4 vs 4x2, other than the expandability options Lnx2Mac mentioned, would be overclocking potential. Having 4x2 puts more stress on the IMC (internal memory controller) and can sometimes limit your overclocking ceiling. For example, when I owned my UD4P I could hit 4.0 with an i5 750 using two 4GB sticks with a 1.25v Vcore. When I switched to four 2GB sticks, I needed 1.36v to hit 4.0.
 
Thanks Lnx2Mac and Jocelyn84 for the thoughtful responses. I had emailed Gigabyte tech support with this question and will post their reply, but I don't imagine it could be any better than the actual data from Jocelyn84's experiment. Thanks again. I love this forum. :headbang:
 
Jocelyn84 said:
From latency tests and PI runs I've done in the past, results were the same with either setup. The only big difference between 2x4 vs 4x2, other than the expandability options Lnx2Mac mentioned, would be overclocking potential. Having 4x2 puts more stress on the IMC (internal memory controller) and can sometimes limit your overclocking ceiling.
Bravo ! I love the collective-IQ of this forum ! ;)

For example, when I owned my UD4P I could hit 4.0 with an i5 750 using two 4GB sticks with a 1.25v Vcore. When I switched to four 2GB sticks, I needed 1.36v to hit 4.0.
Are you absolutely sure this is from the IMC ?
or could that be related to the quality of the RAM sticks themselves ?
 
For example, when I owned my UD4P I could hit 4.0 with an i5 750 using two 4GB sticks with a 1.25v Vcore. When I switched to four 2GB sticks, I needed 1.36v to hit 4.0.
Are you absolutely sure this is from the IMC ?
or could that be related to the quality of the RAM sticks themselves ?[/quote]

Its the IMC (on the chip with Nehalem) and I was running 2000Mhz ram at 1600 back when I had the UD4P setup. Friends of mine were actually surprised I hit 4.0 on air with all dimms filled up. I was told I'd be lucky to even hit 3.8 using 4 sticks without going over 1.4 Vcore. Another thing worth nothing is that my temps were a few degrees warmer (5-6) using 4 sticks vs. 2 sticks, even if I used 1.36v for both.
 
Status
Not open for further replies.
Back
Top