I recently opened up few old boxes and put together the best configuration I could master with what I had on hand. Nothing fancy: just good old Pentium CPU and modest motherboard – an older P4P800-E. My question to all of you clock gurus is the missing connection between the real world meaning what one holds in one’s hand and what the software says once those physical pieces are put together.
Of course I did not expect to see similarity between what the memory label said on the DIMM and what the mem86test reports but I hope there is a rational explanation to all of it.
I offer a screen shot of what mem86 says and seek a good explanation of how that came to be. The CPU is 2.8GHz 800FSB chip and the DIMMs are DDR400.
My questions is this: why does BIOS say FSB is 800MHz and memtest86 says it’s 200MHz?
3200MB/s is only a theoretical number. The 2240MB/s reported by Memtest86 is the speed at which the software is actually capable to access the memory.
The reason why this figure is seemingly low is because the P4 uses the GTL+ bus to communicate with the chipset and then the chipset talks to the RAM. The North Bridge adds latency and makes memory accesses much less efficient.
Modern CPUs starting with AMD's K7 and Intel's first-gen i5/i7 integrate the memory controller into the CPU which drastically reduces latency and enables software to achieve memory transfer rates much closer to theoretical.