the settings are 1T-2T and 1T should be the faster one, but it gives me less than 1% increase in PCMark2002 Ram Bench and in SiSoft Sandra.
Also, when I enable the "turbo" setting my PC blocks during post, despite the fact that this setting seems to do nothing more than setting the Command Rate at 1T.
Why is that? Maybe because of RAS settings that are not vissible in my BIOS?
i can choose 2 way or 4 way interleave, yet the benchmark differences are very negligible.
How many sticks do you have? To perform 4 way you need at least 4 single-sided or 2 double-sided sticks. Otherwise, you're just doing 2 way (1 double-sided or 2 single-sided) or no interleave (1 single-sided).
I thought a thought, but the thought I thought wasn't the thought I thought I had thought.
Imagine memory is organized in rows and columns. To access memory the chipset need to send the row, then wait (RAS-TO-CAS Delay), Then send the colum and wait (CAS Delay). Then it can read the data. If the next read is also the next address it goes into burst mode and reads data at a rate of once per clock cycle. So the latency settings only affect the begining of a transfer. The longer the transfer the less lattency matters. In most applications 95% of the needed memory remains in the cpu's cache further reducing the improtance of latency. This is why you only see single digit increases in performace when moving form CAS 3 to CAS 2.
This only explains two of the three settings and is likely over simplifed. Unfortunately all the information I have found has been doubed down for the benifit of those who have a four year degree in CIS.
I am guessing that the commad rate is the rate at which commands are given once the memory is in burst mode. 1T = every clock cycle, 2T - Every other. I am probably wrong but the important thing with each of these number is lower is better.
The only memory I have seen that need to be run at 2T is PC3300. If you play a lot of games you should set your memory at the highest speed at which you memory runs stable at 2 2 2 1T.