FEATURED Corsair Vengeance Pro RGB 2x8GB DDR4-3600 CL18 - CMW16GX4M2D3600C18

[Woomack]

Today I have the latest Corsair Vengeance Pro RGB memory kit. Tested memory is rated to run at DDR4-3600 CL18-22-22 1.35V. These timings are quite relaxed so many enthusiasts and overclockers may skip it going for CL16 kits. This is also one of the reasons why this memory is interesting. Another reason is its price. Depends on store can find it for about $107. There are also a bit different product numbers with similar specs for about $135 but I'm not sure if used IC is the same as Corsair is using everything in these kits.





The tested kit is based on Micron E-die which appeared on the market not so long time ago and made some noise with some memory clock world records. Depends on luck this IC overclocks between 4200 and 5000+. In most cases closer to 4400-4600 at quite low CL18.
I forgot to take a screenshot from Thaiphoon Burner and I had this memory for a weekend only so you have to trust me there is a Micron E-die. CPU-Z is showing only Micron.


All tests were performed on the Ryzen 3600 and ASRock X570 Extreme4 motherboard with L1.64 BIOS.
XMP profile works without issues but it's 3600 CL18-22-22 so it was expected.


XMP@ 3600 18-22-22 1.35V

Spoiler

I will skip all the RGB lighting as it's the same in so many memory kits that it's getting boring already. Corsair has a nice software to manage RGB but for some reason this memory doesn't work with ASRock software so it's not possible to manage all LEDs on the mobo and RAM at the same time.

 

[Woomack]

3200 CL16-17-17-35 1.35V

Spoiler

3466 CL16-17-17-36 1.35V (a bit more relaxed subs than the 3200)

Spoiler

3600 CL16-19-19-39 1.35V

Spoiler

3800 CL16-19-19-39 1.35V (a bit more relaxed subs than the 3600)

Spoiler

4666 CL18-24-24-48 1.50V

Spoiler

DDR4-4666 is the maximum stable clock. Memory doesn't scale well with higher voltage. Up to 1.7V couldn't run at 4800. After testing 12 different Micron E modules I can only say that they simply hit a wall at some frequency and can't make much more. For this kit it's 4666 while some others have anything between 4200 and 5200. One of my other kits (not Corsair) is scaling up to 1.6V and it let it to run at 4800 CL18 stable. I guess that with some luck Corsair Vengeance Pro can make the same.

In some days I should get G.Skill Royal DDR4-3600 CL18 which should be on similar Micron IC so I will be able to compare it to the Corsair kit.

One thing to add. Tested Corsair memory works the same on ASRock X570 Extreme4 and Gigabyte X570 I Pro WiFi.

 

[maxfly]

Nice stuff for only a little over $100! Thanks Woomak.

 

[Woomack]

Nice stuff as long as you get a kit based on Micron E. There are lower binned Samsungs and SpecTek too and to this specs also fits Hynix C. There is a G.Skill Royal DDR4-3600 at the same timings (another thread) which I thought that is on the same Microns (was listed on ASRock QVL as Micron E) but is using a lower binned Samsung which is clearly worse. I know that Corsair is using SpecTek in their new 3600-4000 kits from Vengeance LPX (already listed on some mobo QVL) and everything else in Dominators Platinum RGB. I have in tests new Nanya which runs about the same so simply good luck with IC as it can be everything

 

[fortyone]

I have in tests new Nanya which runs about the same so simply good luck with IC as it can be everything

Hi, have you tested the Nanya dies ? I have the same RAM with 2x8GB 3600MHz C18 Nanya A-die and I'm wishing to tighten the timings a bit more. If you have tested the Nanya dies, whats the best timing and voltage that you are able to get while keeping the frequency at 3600MHz ? Do I need to change sub-timings or just changing the main timings are enough ? Thanks

 

[Woomack]

I had no Nanya based memory kit for a while and the last one I remember couldn't even pass DDR4-3200. My Corsair kits in the last 6-8 months were all based on Hynix or Micron IC.
I think that ADATA uses Nanya recently but there are problems with review samples where I live as ADATA says there is too high competition in EU so they don't care to send samples. Corsair uses everything and it's a matter of luck.

I guess you have to check that on your own. Typically, tight timings are something like SPD/auto setting. Just use these timings and check how high you can go. I can't help you much with sub timings as I'm not sure how this IC is overclocking.

 

[Bluek]

Hi what are the best 4000mhz settings for this ram or other timings

 

[Woomack]

I don't remember exactly, but considering new chipsets, it's best to keep it at ~3733 Rank 1 or 4500+ Rank 2. 4000 will be something like CL16-20-20-38/16-21-21-39 ~1.40V. This is assuming you have Micron E chips ... and this RAM was with everything that could work under the specified XMP (even Nanya, as it was mentioned in this thread).

 

[Bluek]

Nie pamiętam dokładnie, ale biorąc pod uwagę nowe chipsety, najlepiej utrzymać go na poziomie ~3733 Ranga 1 lub 4500+ Ranga 2. 4000 będzie coś w rodzaju CL16-20-20-38/16-21-21-39 ~1,40 V. Zakłada się, że masz chipy Micron E ... i ta pamięć RAM zawierała wszystko, co mogło działać pod określonym XMP (nawet Nanya, jak wspomniano w tym wątku).

I have i9 13900k

Post magically merged: Jul 10, 2023

these settings don't work

 

[Bluek]

What is better? 4266 cl 18 24 24 44 or 3866 cl 17 19 19 40 or 4000 cl 18 23 23 44

 

[EarthDog]

I doubt you'd notice a difference between them... 4266 CL18 is the best, while the 3866 and 4000 values are really close.

 

[Bluek]

4266 cl18 or go further maybe 4400 cl19.


4266 cl18 vs 3600 cl16 what's better?

 

[EarthDog]

Again, there is little difference between those two values. Your choice.

 

[Bluek]

Again, there is little difference between those two values. Your choice.

4266 cl18 vs 3600 cl16.
there will be little difference
If the difference is small, which one is better?

 

[EarthDog]

4266 cl18 vs 3600 cl16.

There will be a notable difference here. The speed is a lot higher and the CL rating not much higher. I'd run 4266 CL18 over 3600 CL16.

Exactly how much performance difference? Maybe a percent or two. Not worth wasting much time on, honestly.