DDR4-3600C17 vs. DDR4-3200C14

Toggle sidebar Toggle sidebar
M

modeonoff

Distinguished
Jul 16, 2017
1,399
17
19,295
Hello, DDR4-3600 seems to be faster than DDR4-3200 due to higher number MHz (memory clock?). However, I read that memory with lower C value is faster. So, in this case, is DDR4-3200C14 faster or 3600C17 faster? (I recall seeing a thread in this forum that somebody did some calculations and said that 3200C14 is actually faster than 3600C17. Don't recall the details. I could be wrong.)

I asked G.skill to recommend me memory for i9-7900X. The chat connection was broken before I finished asking all the questions so I ended up talking to two guys. One guy recommended 3600C17 and the other recommended 3200C14. The former mentioned that 3600C17 provides "noticably" better performance if I OC and there is no problem with stability. However, the other guy recommended 3200C14 and said that even my motherboard does not list 3200 but 3600 as one of the compatibility memory type, I can use 3200C14 stable by just enabling OC XMP in the Bios. Who is correct?

The first technician mentioned that both memory kits can be used also in Intel 8700K and the Threadripper. However, the 2nd technician said otherwise. Who is correct?

For the DDR4-3200C14 and DDR4-3600C17 Trident Z RGB RAM Kits, how likely can I re-use them in PCIe 4.0 and 5.0 systems when they come out in 1.5-2 years?
 
Solution
n0ns3ns3
simply put, without going into explanations 3200 at CL14 much better than 3600 at CL17.
you can look at it as at two delivery guys
one is driving at 3200 but knows the shortest routes.
the other is driving at 3600 but uses only long routes.
you can get a simple comparison between them by dividing MHz by CL.
The higher the result, the higher the performance.
But that's very general. there are secondary and tertiary timings that also affect performance. especially on AMD CPUs.
Sort by date Sort by votes
3200Mhz CL14 Transfer Time .313ns Cycle Time .625ns First word 8.75ns
3600Mhz CL17 Transfer Time .278ns Cycle Time .556ns First word 9.44ns

RAM latency is a lot more complicated. The first thing you need to know that the CL rating only tells half the story: effective latency (in ns) is calculated by (2000 x CL) / rating.
So your DDR4 3200/CL 14 would have a real latency of 8.75 ns, which is pretty good compared to any of the previous common specifications already.
Latency directly affects frame times for any game with a core data set larger than the CPU's cache system can hold - any object that needs to be checked for render calls and any object affected by player actions, physics or AI needs to be loaded from RAM with every single frame drawn, and every single of these actions will have the CPU waiting on the RAM for the duration.
As it currently is, with CPU performance stagnating for a long time now and limiting the complexity of physics and AI systems, this won't be an issue for gaming if your FPS target is 60.
Once you move to higher FPS targets though and frame time targets get inversely shorter, latency starts to show an effect. Too high RAM latency can cause missed frames (-> constant stutter) and heavily bottleneck a high-end GPU for FPS.
And if you're trying for 288 or even 330 FPS in competitive games on a 144/165Hz monitor to optimize for input latency and complete elimination of screen tearing without relying on adaptive synchronization, you need low latency RAM to achieve it. Or an ungodly fast GPU that doesn't exist yet. This is also where high single-thread performance CPUs come in.
For productivity, latency is important for applications that cause a large amount of random access to large data sets - database applications can't get low enough latency and scale extremely well.


As for ram usage in motherboard with PCIe 4.0 or 5.0 as long as the board still uses ddr4 ram you can use it.
 
Upvote 0 Downvote
Ram support is determined mainly by the motherboard.

You want documented ram compatibility. If you should ever have a problem, you want supported ram.
Otherwise, you risk a finger pointing battle between the ram and motherboard support sites, claiming "not my problem".
One place to check is your motherboards web site.
Look for the ram QVL list. It lists all of the ram kits that have been tested with that particular motherboard.
Sometimes the QVL list is not updated after the motherboard is released.
For more current info, go to a ram vendor's web site and access their ram selection configurator.
Enter your motherboard, and you will get a list of compatible ram kits.

At equal ram speeds, lower latency is better.
But higher ram speeds is accompanied by higher latency, negating some of the benefit.

Most of the time speed does not matter much.
Here is an older study.
http://www.silentpcreview.com/article1478-page1.html
Think low single digit performance impact.

Some apps may be more impacted than others.
In general, more ram trumps faster ram.

pcie support for graphics cards is forward and backward compatible.
To date, there is minimal impact between pcie 2.0/3.0 in graphics card performance.
By the time pcie4/5 become mainstream, your current pars will be near obsolete

 
Upvote 0 Downvote
Thanks. The system is mainly for doing computationally intensive AI machine learning simulation research and number crunching, not for gaming. High memory bandwidth between CPU and PCIe 3.0 GPUs is very important. In this case, is 3200C14 better than 3600C17? I will use 32GB or 64GB.
 
Upvote 0 Downvote
That's really hard to answer. 😉

...but I suppose if it was a reputable ram vendor, that would give them some credibility.

....and then there's the fact that lots of memory that's not on a QVL will work on a motherboard....although not exactly recommended....I have done it lots of times.
 
Upvote 0 Downvote
Intel is pretty easy on their ram, ive never looked at the QVL on a intel board to see if the ram would work or not and have yet to find a kit that didnt work.

Not ever owning an AMD chip, i have heard the new ryzen cpus are very picky and would check the QVL from the MB manufacture to see if they are on the list.

As for ram in your case i would probably go with the 3600Mhz CL17 ram
 
Upvote 0 Downvote
"3200Mhz CL14 Transfer Time .313ns Cycle Time .625ns First word 8.75ns
3600Mhz CL17 Transfer Time .278ns Cycle Time .556ns First word 9.44ns "

I am very sorry but what do these values mean? Is smaller the Cycle Time value the higher the performance (i.e. 3600C17 in this case)?

What is mean by First word? 3600C17 has shorter cycle time but longer First word. 3200C14 is the opposite. I am getting more confused.
 
Upvote 0 Downvote
For gaming you want to find the fast ram with the lowest CL for the fastest First word, since gaming is has more random calculations.

For what your doing, AI machine learning simulation research and number crunching, you just need the fastest cycle/transfer times possible
 
Upvote 0 Downvote
Using your ram for example, just so you know how to calculate out for future reference.

Transfer time = 1 / data rate -> 1/3600Mhz = .278ns
Command rate = Data rate / 2 -> 3600mhz/2 = 1800Mhz
Cycle time = 1 / Command rate or 2 × Transfer time -> 1 / 1800Mhz = .556ns or 2 x .278ns = .556ns
First word = CL x 2000 / Data rate -> 17 x 2000 / 3600Mhz = 9.44ns
 
Upvote 0 Downvote
simply put, without going into explanations 3200 at CL14 much better than 3600 at CL17.
you can look at it as at two delivery guys
one is driving at 3200 but knows the shortest routes.
the other is driving at 3600 but uses only long routes.
you can get a simple comparison between them by dividing MHz by CL.
The higher the result, the higher the performance.
But that's very general. there are secondary and tertiary timings that also affect performance. especially on AMD CPUs.
 
Upvote 0 Downvote
Solution
Confusion again. Isn't 3600 correspond to the lowest value in cycle time? It has 0.556ns but 3200 has 0.625ns.

3200Mhz CL14 Transfer Time .313ns Cycle Time .625ns First word 8.75ns
3600Mhz CL17 Transfer Time .278ns Cycle Time .556ns First word 9.44ns "
 
Upvote 0 Downvote
On the same frequency, Lower timings = better performance.
I already explained above how to compare kits with different speed and timings. divide frequency by CL and compare results. higher result = better performance.
Now you can use a calculator to figure out which is better.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom