DDR4 RAM - what does "tested latency" mean (and does it matter?) - comparing two nearly identical sets of RAM

kukov

Distinguished
May 18, 2015
28
0
18,530
Hi folks,

I recently upgraded my setup to a new i7-5820K with the Asus X99-A and a set of Corsair Vengeance LPX 16GB (4x4) RAM.

I decided it would be worthwhile to spring for the extra set of 16 to fill out my motherboard, and got another set of, what I thought was, identical RAM. Took the new sticks home and before opening the package I noticed a difference in the SKU.

My original set is: CMK16GX4M4A2666C16R

While my new set is: CMK16GX4M4A2666C15R

At first I couldn't figure out what the difference was - thought it might have been a typo. But after investigating on Corsair's site it seems the two models are identical except for one detail, their "Tested Latency" (I have no idea what this is).

The 16R model lists 16-18-18-35
While the 15R model lists 15-17-17-35

So my question is - is this a significant difference?

I don't really want to go back and exchange this RAM, but, of course, I will if installing these new sticks will make a noticeable difference. If it's only something like 5% for the new sticks only then I suppose it doesn't matter much.

Can anyone offer advice? Should I return the new sticks or does it not matter?
 
Solution
Lower latency is better, the C15 new sticks will be slightly better then the old ones, this difference however will not be noticeable.

kukov

Distinguished
May 18, 2015
28
0
18,530


Thanks - glad to hear it won't make a difference.

But I thought the latency was determined by the clock speed that the RAM is set to? So in this case if the RAM can go 2666 (or more if overclocked) then isn't that what's determining latency? How do those four other numbers come into play?
 

Vosgy

Honorable
Nov 24, 2014
715
1
11,360
Ok so if you change the Hz rate on the RAM the timings will change around as the response time is a limitation of the RAM itself. C10 @ 1600MHz OC to @ 2400MHz may change this to C15 as 15 cycles at 2400MHz is approximately C10 @ 1600Mhz going down in Hz does the same thing in reverse. The timings provided give a representation of the Best response time they managed to get out of the RAM. So if you go up in Hz the timings may become c20 @ 2800Mhz which although a higher clock rate as many more cycles are required for a response ends up with slower response.
 

kukov

Distinguished
May 18, 2015
28
0
18,530


Thanks for the response! But I'm still confused. Does the "C" you refer to refer to the model number of the RAM? If so, uh, what? If I'm understanding what you're saying, C16 would be better than C15? But that's not the case with the stats above.

I suppose I don't really need to understand what's going on - as long as I know that there's basically no noticeable difference between the RAM I'll be fine. I just want to make sure having different models will not hurt my OC potential.
 

Vosgy

Honorable
Nov 24, 2014
715
1
11,360


So C15 or CAS15 is the Latency Timings of the RAM, typically would mean 15,15,15,30 timings.
C16 or CAS16 would mean something like 16,16,16,32 timings.
These can vary a bit to things like 15,17,17,35 which is the timings on the C15 RAM you linked.
Or like the 16,18,18,35 on the C16 RAM you linked.

Each one of these represents a different type of call the CPU can make to the RAM. They represent the Number of Cycles the RAM goes though before it responds to that call type. Therefor if the two RAM's have the same amount of cycles per seconds also know as Hz the one with lower Numbers in the CAS Timings is the better one (assuming all other numbers are the same ie. voltage), as this one responds to requests from the CPU faster then the other one.

So at the same Hz, Voltage etc. C14 > C15 > C16 etc.
 

kukov

Distinguished
May 18, 2015
28
0
18,530



Thanks! So just to confirm, the difference between C15 and C16 will basically be unnoticable, even if I plan to OC?