Going 1866->2133 question

[taq8ojh]

I have Corsair Vengeance CMZ16GX3M2A1866C9 memory in my PC. Except for the fact my board refuses to work with the XMP profile in a stable way, I like it. Not bad bang for the buck.
Recently I had the idea to see how high could it possibly overclock, but since I never really cared about this (at least not since DDR2 times), I am probably hitting a wall due to lack of knowledge in this particular field.

By default this memory runs at 9-10-9-27, 1.5V.
I just set everything to auto, the multiplier to 21.33, and kept raising voltage until I could successfully boot into Windows and possibly pass some time in Prime95.
I was fairly surprised I basically couldn't get under 1.65V in order to have at least an impression of stability. Do you think it's normal for this kind of speed jump?
Also, is there any general guideline for setting the four basic timings? I believe there's some headroom, but I have absolutely no idea how to go about this. Back in the days of my AthlonXP days, I believe I read something like tRAS must be higher than the sum of the other timings, which - if it ever was true - probably doesn't apply today.

What do you suggest I try?

 

[coffeejunkee]

I suggest you try nothing. Seriously, even if you get 2133 you'll probably need such loose timings that it won't be better than 1866 + you might damage the imc going over 1.575V.

Besides some exceptions (Samsung, Crucial lp ballistics) most memory has been binned to dead. If your kit could do 2133 Corsair would have sold it as 2133. I have GSkill 1600 1.35V ram, you'd expect it to easily do 1866 at 1.5V but it doesn't. Even 1.575V and very loose timings won't let me boot into Windows.

For timings, just look at cpu-z spd profiles or steal from other kits. But ras certainly shouldn't be higher than the sum of other timings, see standard 9-9-9-24: 3x9 = 27.