Overclock.net banner

1 - 3 of 3 Posts

·
Registered
Joined
·
53 Posts
Discussion Starter #1
Hello! Could you please answer some questions? I just could not find any comprehensive answers to them on the internet.

1) I have seen some people saying that you should always go for 100:100 instead of 100:133 ratio for BCLK/DRAM when memory speeds and latencies are approximately the same. That a RAM with the 133 mode (2133, 2666 MHz) somehow lags behind a RAM with the 100 mode because it is not fully synced with the CPU bus speed. Is that true?
2) Do memory kits with speeds such as 2400 or 3200 MHz (basically any numbers divisible by both 133 and 100) run the 100 mode or the 133 mode? Or they can do both?
 

·
Iconoclast
Joined
·
30,430 Posts
1) Generally, no. There is no "CPU bus speed" and nothing to synchronize the memory clock with. There are reference clocks, but running a different or multiplier reference clock doesn't have any performance implications, in and of itself. Memory vs. reference clock ratios haven't had any intrinsic effect upon performance since front-side buses and off chip memory controllers. At the same clocks and timings performance will be the same irrespective of the reference clock used for the memory multiplier. I frequently use the 133MHz reference clock because some platforms do not like high memory multipliers.

However, some boards may have different auto settings depending on BCLK, memory reference clock, and memory multiplier, which can affect timings. Better boards generally don't do this, or allow the same manual settings irrespective of the auto settings.

2) The memory itself doesn't care where the final clocks come from and neither SPD nor XMP profiles typically specify the memory reference clock. This is normally determined by the board.
 

·
The last VRM burner
Joined
·
2,590 Posts
There should be no difference in performance between both multipliers. My sticks although, did not want to launch with 133 multiplier at their rated speed.
 
1 - 3 of 3 Posts
Top