Ryzen and RAM CAS Latency

[iiNNeX]

Does CAS Latency matter with Ryzen 7 CPUs?

 

For example a 16GB GSkill Trident Z RGB kit at 3200mhz costs £165 or so for CL16, but for CL14 it would be £205. For a few quid more you can actually get a 3600mhz kit of the same ram (but I know 3200 is the sweet spot at the moment with a high overclock on the CPU).

 

I don't mind paying the extra if it will actually give me a performance boost.

 

Thanks!

[WereCat]

https://community.amd.com/community/gaming/blog/2017/07/14/memory-oc-showdown-frequency-vs-memory-timings

[iiNNeX]

5 minutes ago, WereCat said:

https://community.amd.com/community/gaming/blog/2017/07/14/memory-oc-showdown-frequency-vs-memory-timings

Thank you bud, that's a good read.

 

Just going through the details now, but I've also noticed that all 3200mhz ram used is CL14 on that test, with some being CL12. I guess things would only be worse with CL16 right?

[DrMikeNZ]

I personally use a 3200CL14 kit running at 3333CL14. In my workloads it made zero difference compared to 2133CL15, but in several games (not all) you will notice a small improvement, I personally wouldn't consider the benefit worth the cost. Depending on your luck with silicon lottery though, you may not even get the rated speed out of the RAM due to the IMC on the CPU or poor RAM overclocking still present on some motherboards.

[Tiwaz]

3 minutes ago, iiNNeX said:

Thank you bud, that's a good read.

 

Just going through the details now, but I've also noticed that all 3200mhz ram used is CL14 on that test, with some being CL12. I guess things would only be worse with CL16 right?

yep, but for gaming CL doesnt really matter, i mean with DDR4 gaming performance didnt get worse, even tho DDR3 had much lower timings, like CL9

[iiNNeX]

3 minutes ago, DrMikeNZ said:

I personally use a 3200CL14 kit running at 3333CL14. In my workloads it made zero difference compared to 2133CL15, but in several games (not all) you will notice a small improvement, I personally wouldn't consider the benefit worth the cost. Depending on your luck with silicon lottery though, you may not even get the rated speed out of the RAM due to the IMC on the CPU or poor RAM overclocking still present on some motherboards.

I see, but the difference in games from 3200CL16 to 3200CL14 should be pretty low if any at all?

 

I will be pairing it up with the Asus Crosshair 6 or the Asrock Taichi and plan to buy 2x R7 1700s and return the weaker one (thanks Amazon Prime lol).

2 minutes ago, Tiwaz said:

yep, but for gaming CL doesnt really matter, i mean with DDR4 gaming performance didnt get worse, even tho DDR3 had much lower timings, like CL9

That's what I thought but I wanted to be sure from somebody who knows more about the Ryzen platform, as I've used nothing but Intel for the past 5 years.

 

Thanks guys

[Sakkura]

17 minutes ago, Tiwaz said:

yep, but for gaming CL doesnt really matter, i mean with DDR4 gaming performance didnt get worse, even tho DDR3 had much lower timings, like CL9

You can't compare the timings directly, you have to compensate for the difference in clocks. That's because timings are quoted in the number of clock cycles, not in nanoseconds (what actually matters to performance).

 

DDR3-1600 CL9, for example, has a CAS latency of 11.25 nanoseconds. DDR4-3200 CL16 has a CAS latency of 10 nanoseconds, so it's faster despite CL16 sounding a lot worse than CL9.

[WereCat]

40 minutes ago, iiNNeX said:

Thank you bud, that's a good read.

 

Just going through the details now, but I've also noticed that all 3200mhz ram used is CL14 on that test, with some being CL12. I guess things would only be worse with CL16 right?

yea, if they are on the same clock

[iiNNeX]

Check this dude out:

 

"I am running a Asus Crosshair VI Hero mobo, a 1700 and 32gb's of GSkill FlareX 3200mhzC14 ram. I have the cpu oc'd to 3933.6mhz, and the ram running at 3496mhz. This thing is blazing! I am getting 1760's on Cinebench cpu. I might be able to push things a bit higher as this was done pretty easily compared to how hard it was just to get it to a couple of points lower previous to reading this article. turning off those settings (BSG & GDM) really helped with stability. Also setting the Proc ODT to 40ohms was a big boost as well. I will try to push it to 55ohms tomorrow and try to hit 4000mhz on my 1700."

 

Damn, hope I get the same luck