Faster RAM is generally poo poo'd because ....as the logic goes .... the increase in performance is smaller than the increase in price. This is false logic.
1. When a new kind of RAM (or any other product for that matter) .... first comes out, production lines still have a few quirks and yields are low..... you have heard of "binning" ? .... sticks that meet say 2400 are binned and sold at a premium and in the beginning these are rare. What doesn't meet 2400 will be tried at 2133 and if still not passed may wind up sold as 1866 or 1600. When rare, as enthusiasts are willing to pay a premium, they are sold at a decent mark up. The problem here is, the reviews are generally done when new RAM comes out and "not worth it" reviews are written .... then these get quoted years later when the price differences are small or non-existent. I still see peeps buying 1600 when 1866 and 2133 are in fact cheaper.
2. The logic then goes that with a performance increase of 2 - 5% on average, why would you pay $240 for 16GB of DDR3-2400 when it costs 50% more then the $160 DDR3-1600 ? That's "inappropriate logic" as it's not your RAM that goes 2 - 5% faster, it's your whole system that goes 2 - 5% faster. In a $2k box, that's an extra 4% money for a 2- 5% increase in performance..... maybe it is worth it, maybe it ain't.
3. However....what happens now is the lines mature, yields improve and the market has more 2400 RAM that they have peeps willing to pay a $80 premium so the price drops...... we also have just about all sticks passing 1866 and 2133. So what we see now is 1866 or 2133 RAM at the same price or sometimes even cheaper thane 1600. Buying 1866 and 2133 at that point is the proverbial "no brainer".
GSkill 1333 CAS 9 = $155
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231486
GSkill 1600 CAS 9 = $165
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231568
Gskill 1866 CAS 9 = $160
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231615
GSkill 2133 CAS 9 = $161 (15% off w/ promo code EMCPDWG33, ends 6/15)
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231571
How do you not buy the 2133 ?
4. The performance increase is application dependent .... some apps benefit significantly, some don't ...... many games benefit not at all but some do.....some do A LOT. Some may or may not depending upon whether your CPU and / or GFX bottlenecks and gain that would otherwise be observed. Metro Last Light shows like a 0.5% increase but F1 shows an 11% increase from 1600 to 2400 (that's with a 290x system)
5. However, the impact of faster RAM is generally greater on minimum fps rather than average fps but so few reviewers look at that these days.
So whether or not to buy it, is strictly a price decision.....
Hard not to buy the 2133 above when it is cheaper than the 1600 and a buck more than 1866. As for jumping to 2400 .... that's a week to week decision as prices change .... I prefer the Mushkin series as I am confident of getting Hynix memory chips .....Corsair used Hynix in their Vengeance Pro line but switched to a less expensive module once the lesser manufacturers yields improved.
Mushkin's prices jumped a bit this week on newegg....right now but they also switched up their lines a bit.
The 1600 CAS 7 is close to 2133 CAS 9 in speed
Mushkin Redline 1600 CAS 8 (5.00ns) = $170
Mushkin Redline 1866 CAS 9 (4.82ns) = $164
Mushkin Redline 1600 CAS 7 (4.38ns) = $185
Mushkin Redline 1866 CAS 8 (4.29ns) = $180
Mushkin Redline 2133 CAS 9 (4.22ns) = $192
Mushkin Redline 2400 CAS 10 (4.17ns) = $190
So with $164 being the cheapest for 4.82 ns is it worth $26 more (1% in a 2k system) for a 16% increase in memory performance ... which may translate anywhere from 0.5 % to 11% in games ?