I can see from the test data that absic has found issue in the relationship between the CPU
memory controller and DDR3 frequency when it goes above 1333 MHz, but what I am not understanding what is the precise reason for the problem is. In all of the tests run by absic, as the memory controller is a bit unknown or is a black box whose workings are not measurable or transparent.
In a research paper from Corsair (
AN806: Memory Upgrade Resource Guide), they suggest that running 4-modules of RAM puts "extra load" on the memory controller, but what that "load" is comprised of is not explained.
When the memory controller speed is increased for overclocking, it simply cannot manage the same amount of memory at that overclocked speed for proper stable communication with the CPU.
Here are some possible reasons (speculations really) for the problems that higher RAM frequency causes (1333 vs. 1600 MHz):
- the voltage is increased beyond the capacity of the memory controller;
- more bandwidth is used for data throughput than the memory controller can handle;
- more heat is generated, causing the memory controller to fail;
- a combination of all these causes "interference".
I am sceptical that the memory controller is the problem if
it can't be explained why it is the problem. The reason why I am sceptical is that my CPU has
Dual-channel architecture, and in any case, the integrated memory controller (IMC) should enable the CPU to handle higher RAM frequencies anyway.
Basically, I just don't understand what is the precise reason for CPU/RAM bottleneck. It seems to me that if we could understand what is the precise cause of the bottleneck (rather than speculate), then perhaps we could adjust the Bios to achieve a theoretical RAM frequency maximum. Any ideas what causes the bottleneck?
