I would imagine you'd see much more benefit from the added RAM over the slightly faster latency.
OK, thanks, I bought 24 GB DDR3 1600 CL 9 instead of the faster 12 GB CL7.
Will install and run shortly and send in the PPBM file then.
Would I have done better buying DDR3 2000 ? ( at over twice the price but it's too costly to buy unnecessary ( too ) fast DDR ).
In other words, what is the optimal memory speed if the 950 is stable at OC 4.3 GHz ?
There must be an optimum memory speed in relation to a certain ( i7 ) processor speed.
( BTW : this 24 GB 1600 memory could be oc'd to ~~ 1850 MHz but I see tweakers running oc'd DDR2000 for i7 950 @ 4.3 GHz and there lies the origin of my question ).
What math should I do to learn the optimum memory speed in my case ?
I realize this is no OC tweakers forum but I still guess more people here have considered this question.
Again, thanks for any help.
For memory speeds, you can take any old stick of memory and run it past its rated speed, but the manufacturer doesn't "guarantee" that it will. DDR3 1333 can run at DDR3 2000 if you get some nice sticks (can't really say at what success rate, but there are sticks that can). The same thing goes for latency. I always use Gigabyte motherboards and G Skill RAM. With my Gigabyte X58A-UD3R motherboard and 2 sets of 3x4GB Ripjaw DDR3 1333 CL9 RAM (these guys), I was running at default speeds with 7-7-7-20 latencies. I still haven't gotten around to doing overclocking tests, and not much of an overclocker to begin with.
When it comes to testing optimal settings for processor usage/RAM speeds/RAM latencies, I say go the After Effects testing route. Take a given project layout you see yourself confronting the most, and test that changing one variable at a time, and recording the results. The PPBM test results others have posted is great for assessing where you might be able to improve, without doing the time-consuming test mention before. However, if you are obsessed with the precise optimization of your system settings, nothing beats the one variable retest method, if done correctly. I believe though, that it is quite unnessesary for Premiere Pro with the PPBM results readily available to you. On the other hand, After Effects has no such database because of the diversity in workflows and project variability.
When it comes down to it, 24GB of RAM will always trump any kind of 12GB RAM kit. So yeah, you made a wise decision