OK...I don't know jack about RAM timings. At first I thought I knew, I figured the lower the numbers, the better right? Then I read other articles and crap saying to move them higher...I'm confused. Can someone give me a real brief "dumb man"'s explanation of what the timings do, and is it "the lower the better" or not? Some one be gentle and enlighten me :-)
In general the tighter (lower) the timings the better, all the different acronyms corrospond to how quickly the ram transfers data. Sometimes though tight timings are not possible, especially when you are overclocking and you may want to loosen (higher) the timings to get stable.
Simple but I don't think we need an entire discussion on ras, cas, command rate... etc.
OK so if you're not OCing the proc, then try to tighten the timings via CPU-Z. If you are OCing, loosen them up? So just keep using CPU-Z or Memtest to test the stability as you fiddle? That about sum it up?
Well actually I'm not OCing anything right now...I'm just asking for future reference for when I build a new comp with the CONROE thank ya very much. My current comp is too 'frankenstein-esque' to OC w/ stability. I'll have to find an OCing guide at some point and struggle through when the time comes. Thanks
RAM runs at a certain clock rate / frequency (MHz). As well, the data in the RAM is layed out in a grid like fashion of rows and columns. Thus, to select some data to read or write to the RAM, first a signal is sent about which row to seelct, then which column, then the actual data operation happens (read/write).
Each of these operations take some time to complete, but they aren't all the same length. So rather than each operation take one clock cycle a piece, forcing the clockrate to be whatever the slowest operation is (and having hte faster operations jsut sit idle once their done, wasting time), you can instead break the clock down to much smaller rates, and have each operation take multiple, but varying, clock cycles to complete. These are what the RAM timings are: the number of clock cycles that the controller should wait to expect that an operation is done.
Now, if the timings are too high, you'll end up wasting time, as the controller will sit idle waiting for an operation to finish, even though it's been done for a clockcycle or two already. But converserely, if you set the timings too low, the controller will issue new operations before the previous ones finish, causing data corruption.
Now, as you increase the clockspeed of the RAM by overclocking, what may have taken 3 clockcycles before may have to take 4 since the length of a clockcycle has decreased.
Hopefully that'll give you a general idea of how things work. Just google the definitions of how it all works if you want a finer detailed explination.
i remember reading a 5 or 6 page review on ram timings on various makes of ram here on toms hardware, but cant find it now (. The max benefit they managed to get out of it was 3%. I decided i could not be bothered with the headache of the whole thing and so left it.
I got it down now thanks a lot boys. Anyone got anything to argue against the +3% benefit? If its only 3% I agree...too much trouble to worry about it. The only things I really am pressing to OC is my proc and the video card because thats the biggest improvment I've had with my current computer. I have NVidia 6200 (yes crappy I know, but at the time it was good) w/ 300 core and 500 mem core...I OC'ed it to 430 core and 650 mem core. That helped a crap load...and I could only get my 2.6 Ghz P4 to 2.8 with total stability because I my mobo won't let me adjust the VCore. So yeah, maybe I'll just get the fastest ram I can (non-OCing) and live with it. Thanks for all your input boyos!