12GB DDR3-1600 versus 16GB DDR3-1333

[PingSpike]

I have an ivybridge i5-3470 setup and after looking around at the various modules I have I can either put 12GB of DDR3-1600 in (2x2GB + 2x4GB) or 16GB of DDR3-1333 (4x4GB). I'm going to be stuck running 2t either way I suspect. Previously I just ran 2x4GB and didn't really have any complaints, but I've been playing with virtual machines a lot more lately and I think I could make use of more ram for once. I've decided against buying a 2x8GB kit because the stuff is expensive now and I don't think I am really DYING for more ram or anything.

I guess my question is there really any performance difference between DDR3-1600 and DDR3-1333? In the past it seemed to be pretty minimal in games. I'm not sure what applications it makes the most difference in.

 

[zir_blazer]

It depends on your memory usage. If you DO are going to use all those 16 GB of RAM, they will be much better even if they run at 1333 MHz. Unless you're using the IGP for gaming, the performance loss is small.
I do suggest you to go for a 2 * 8 GB kit instead, may want to read this.

 

[PingSpike]

Unless I'm mistaken that thread didn't go into 1t versus 2t that I could see and it seemed to be about AMD platforms. I don't plan on using the IGP for anything graphically intense, I have a 6870 for that.

I agree the 2x8 kit would be best. But at $100+ for even a cheap kit of no name ddr3-1333 that is a lot of scratch to drop versus the 4x4GB kit that I already have. I'm leaning towards the 12GB setup at the moment but I suppose I can just benchmark them whenever I get around to it.

I had found a website where a user compared 1t versus 2t but he put his results up in images, which have since been deleted. Disappointing!

 

[Cerb]

If you have the 4x4GB kit, just try both. If the 12GB mixed passes memtest, you're good to go; or, you can use the matched 16GB set.

If you could use more RAM, though, and you have a 16GB kit, hwy not just go for it?

RAM today takes speed differences of 30% or more, combined with applications that have poor cache behavior, to get past a couple percent difference in performance. Even if the differences are clearly measurable, they will be negligible from a practical perspective.

 

[PingSpike]

Yeah, I'll probably just do a little testing to see what works. I'm starting to come around to the more is better camp.

 

[Makaveli]

I would go with the 16GB kit.

You could always raise the timings and run it at 1600 if need be.