# What causes memory instability?

Tags:
Last response: in Overclocking
Share
July 13, 2011 12:05:53 AM

Hello all,

I've been doing a fairly large amount of reading on overclocking, as I will soon join the club and would like to jump into my first experience with somewhat of a clue. My focus is currently on RAM.

Here is my understanding thus far:

Countrary to what I think a lot of people believe, RAM performance isn't completely dependant on RAM frequency OR RAM timings, but rather the combination of both. RAM frequency is determined by multiplying the base clock by the RAM multiplier. The result is a number in htz, or in slang-ier words, ticks per second. Ram timings look like <4-4-4-12 or 6-8-6-21>, where each number represents the amount of 'ticks' it takes (or rather, the RAM is given) to accomplish certain tasks. The first seems to be of most interest: The CAS latency.

The actual performance of the RAM is determined by combining the frequency and timing to attain a value known as Latency, typically in nanoseconds. RAM operating at 1333mhz is operating at 1333 millions ticks per second, which translates to one tick every 1,333,000,000th of a second, or 1 tick per 0.750 nanoseconds. Assuming a CAS of 6, the actual latency is 0.750ns * 6 * 2 (I assume this is because latency refers to a there-and-back length of time) which comes to 9.002 nanoseconds.

Some people will say that timings aren't nearly as important as frequency and will forgo 'tight' timings to achieve higher stable frequencies and may end up with something (exaggerated) like a CAS of 12 @ 2200mhz => 21.818 nanoseconds. On the other hand, people may try for a CAS of 4 @ 500mhz and get an actual latency of 16 nanoseconds.

My question actually begins here:

To achieve higher frequencies, the RAM needs higher voltage, which results in higher temperatures and lesser stability.

To achieve 'tighter' (lower) timings, the RAM needs higher voltage, which results in bla bla bla.

Compare, for example, CAS 6 @ 1066 to CAS 9 @ 1600. The latencies are identical, but certain sticks will be less stable (or fail entirely) at one of these two setups than at the other. This leads me to believe that it isn't the end result that determines the amount of strain on the RAM, but one (or both) of the elements multiplied together.

CAS 2 @ 533.33mhz has the same latency (9) as CAS 9 @ 2400mhz, but it intuitively seems to me that the latter is a much more impossible setup than the former.

Which element, then, of the RAM puts more strain on the sticks and has the largest impact on stability?

(While you're all at it, please feel free to contribute to my understanding and/or correct any fallacies)