Question Whats the real world performance difference with RAM speeds?

ch33r

Distinguished
BANNED
Jun 13, 2010
316
4
18,685
I'm building a new rig. I've got some parts on order.

Specs: Gigabyte Aurus Master X570. Gigabyte RTX 2080S. Ryzen 3900X. 850W PSU

Question: Will I notice a real world performance difference between DDR4-3600 CL16 and lets say DDR4-3200 CL19? What real world scenarios could I run into where the difference will actually matter and be noticable. Secondly, I would like to use 64GB of RAM. The X570 Master is a daisy chain, and daisy chains work best with 2 sticks. If I use 4x16GB am I going to lose speed? Or be forced into single-channel mode? These are issues Im worry about
 
A small but noticable difference between 3200c19 and 3600c16 in some games and memory hungry workloads. If prices are a good bit apart, go with the cheaper one.

If you use 4 sticks it will run in 2 channel with a command rate of 2. It will be slower than 2 sticks which can run with a command rate of 1, but not much.
 
A small but noticable difference between 3200c19 and 3600c16 in some games and memory hungry workloads. If prices are a good bit apart, go with the cheaper one.

If you use 4 sticks it will run in 2 channel with a command rate of 2. It will be slower than 2 sticks which can run with a command rate of 1, but not much.

What do you mean by "A small but noticeable difference in some games"??

Do you mean like loss of frame rate? Or stuttering? Isnt that GPU dependant? Or is there more to it?
 
Slightly lower framerate. Really only noticable in benchmarks.

If your gpu dependent, nothing really changes.


If you go to 2133mhz or 2666 you will begin to notice the difference.

The reason Im asking is because I see all these RAM modules that are like.... ya know 3200MHz CL19 which is half the price of like a 3600 CL16.... and I'm just like.... does it really matter? Ill spend the money for the extra performance if it really matters, but I'm trying to figure out if its just a gimmick/marketing ploy or if Ill really notice it in real world situations, not just benchmarks. Also, about the whole daisy chain thing? Why would 2 sticks be slower than 4 on a daisy chain, and again, realistically?