phenom ii and ddr3 speed vs timings

so my hardware arrives tomorrow and i get to start playing around with it so i guess im just tryin to get the basics sorted. so the question speed or timings? i know amd in the past has always shown performance gains off of timings, but yesterday i ran across this link http://www.extremetech.com/article2/0,2845,2343573,00.asp that shows speed is a huge factor for phenom ii's.

i couldnt really find anything that benched for example 1600 at cas 8 vs 1333 at cas 7. if i were to go off this article alone it looks like speed is a huge factor for the phenom ii's. but this was an 810 and im getting a 955.

so i guess what im trying to ask is speed or timings? does anybody have any links to benchies? thanks in advance.

810s and 955s should behave similarly to ram settings. The 810 might be slightly pickier due to it's smaller cache, but they'll be close.

In the past it has mattered what you were doing, lots of small data chunks prefer low latencies, fewer larger chunks prefers bandwidth.
I could do a 1333-7-7-7-21 vs 1066 6-6-6-18 test on an 810, don't have any 1600 though.

aye. so i guess ill have to play around with it. my package is out for delivery so i should have it in a little bit...hopefully ups didnt decide to ace ventura it as usual. i've seen some people running the ram i got in the 1700-1800+ range is why im curious http://www.newegg.com/Product/Produ...m_re=g_skill_ddr3_1600-_-20-231-278-_-Product

ill report back when i get it all together

RAM speed only provides greater performance IF it is scaled with the NB properly. If you keep your NB at stock speed, you could run your RAM speed at 2500mhz or 1333mhz with the same timings and get basically the same results. The whole reason AMD only officially supports 1333mhz and anything faster than that is via OC on the mobo is due to the fact that at stock settings 1333mhz is faster than what the stock NB speed can handle. When you begin to overclock and increase your NB speed, then you need to increase the RAM speed to not starve the NB for bandwidth.

Here is an exampel, you have a faucet that dumps water onto a trough, the trough dumps the water into large barrel. The faucet in this example is your RAM, the speed the water comes from the faucet is the RAM speed. The trough is your NB, and the barrel is the cpu. Now when water flows from the faucet, it hits the trough and is carried to the cpu. Think of water as a data flow. Now your trough can only handle so much water at once else it ends up spilling the water over the edges of the trough before it gets to the barrel. This would indicate that your RAM speed is to fast for the NB to keep up with it. So as you increase the water flow from the faucet (RAM speed) you then also need to get a bigger trough (NB) that can carry the more volume of water being dumped onto it.

So if you jack your RAM speed up without scaling the NB to match, you are wasting performance, if you jack your NB speed up to far from what the RAM speed is you are also wasting performance as the NB is being starved for data flow. The key to having a fast AMD system is BALANCE. Faster NB and RAM speed that compliment each other is a good thing. AMD loves lower latencies and does better with it than with higher latences. Since you arent using LN2 or any other extreme cooling methods, it is better to get lower speeds and tighter timings and scale the NB along with it.

Here is a nice little guide that explains things in graphs with some actual testing done. Link