Take a train, the speed the train goes at would be the RAM frequency and the time it takes for passengers to get aboard/leave the train would be the timings. (The time it takes the memory to access/read/write data )
You may have the fastest train, but if the passengers takes a long time getting in there, the slower train with quick passenger in/outs might just arrive first!
Obviously it's more complicated then that, but it gave me a good a while back and hope it'll help you too.
There's plenty of info on the net about mem timings, including on AMD's website :
HyperTransport this is how your CPU transfer information from/to your northbridge and then from northbridge to/from southbridge.
In other words, it replaces the Front Side Bus, you can calculate your resulting HT speed by multiplying your CPU clock by your HT Link, you will want to keep that around 1000 since there's only instability over that and no improvement in performances.
Allow me to add my 2 cents here about RAM timings and speed:
It is going to be best to run at a 1:1 divider. If you're talking benchmarking, then you can play with it and sometimes having a higher bandwidth will make things faster. Just keep in mind that is theoretical. You are never going to see that increase really. Take a game for example, like F.E.A.R., and grab a digital stopwatch. Time a specific load time at a higher bandwidth at a ratio of like 1:2 or 2:3 etc... then time the same load time with a 1:1 ratio and tighter timings. Since this is reality you are measuring, you will see that the 1:1 ratio is performing better. Anything over 1:1 is really just a waste because your CPU can only go as fast as your FSB will let it. the RAM has no say...
A forum community dedicated to overclocking enthusiasts and testing the limits of computing. Come join the discussion about computing, builds, collections, displays, models, styles, scales, specifications, reviews, accessories, classifieds, and more!