Corsair Vengeance RGB DDR5-6000 CL36 2x 16 GB

[ir_cow]

Corsair Vengeance RGB brings a competitive DDR5 kit, with an XMP 6000 MT/s profile and modest timings of 36-36-36-76. Designed to fit right into the Corsair ecosystem, it is time to take a closer look and see what all the fuss is about!

Show full review

 

[P4-630]

I see for gaming fps wise I'm doing OK with my Corsair Dominator Platinum's 6000MHz 36-36-36-76 (SK Hynix) @ 1.35V

 

[ir_cow]

I see for gaming fps wise I'm doing OK with my Corsair Dominator Platinum's 6000MHz 36-36-36-76 (SK Hynix) @ 1.35V

Do you have RGB control issues with the Gigabyte Master? I haven't tried that one myself, but the GB Tachyon requires a BIOS intervention.

 

[P4-630]

Do you have RGB control issues with the Gigabyte Master? I haven't tried that one myself, but the GB Tachyon requires a BIOS intervention.

Well I couldn't control it with the Gigabyte RGB app, installed icue once to try, that worked, but uninstalled it short after that and now just seeing the rainbow effect which I also see on my MSI GPU without the MSI software installed... Oh and there are 2 places on the motherboard changing colors (not much noticeable imo) , It's fine by me as is now.

 

[ir_cow]

Yeah its not going to show up in GB Fusion, but that's nice its detected by iCUE. Its so random which motherboards have this SPD Write disabled or enabled.

 

[cellar door]

No one should bother with these... the next gen ICs are ready and already available in no name kits, and they reach close to 8000mhz.

DDR5 ICs der nächsten Generation im Test: SK hynix 16 Gbit A-Die - 2x 16 GB mit Overclocking und Gaming | igor´sLAB

Heute habe ich einen Blick in die nähere Zukunft der RAM-Technologie für euch und wie so oft in der Branche der Computer-Hardware, kommt dieser aus China, wo bereits heute die neuen DDR5 Speicher-ICs…

www.igorslab.de

 

[ir_cow]

No one should bother with these... the next gen ICs are ready and already available in no name kits, and they reach close to 8000mhz.

DDR5 ICs der nächsten Generation im Test: SK hynix 16 Gbit A-Die - 2x 16 GB mit Overclocking und Gaming | igor´sLAB

Heute habe ich einen Blick in die nähere Zukunft der RAM-Technologie für euch und wie so oft in der Branche der Computer-Hardware, kommt dieser aus China, wo bereits heute die neuen DDR5 Speicher-ICs…

www.igorslab.de

Don't buy these because it can't do 8000+ isn't a very good argument. First, it doesn't change the fact that no 4-DIMM slot Z690 MB can run 6933+ stable (or even boot). What about all the folks who just want 6000 MT/s memory?

If you read the igor article you would have seen this in the memory overclocking section.

Unfortunately, higher clock rates or higher voltages only lead to instability with my kit, even with the PMIC unlock of the MSI BIOS. The timings are not revolutionary compared to M-Die and are even a bit more conservative in places, but the clock rate can be increased significantly again and at a relatively low voltage. Incidentally, a 120 mm fan takes care of the cooling, which takes over the job of the non-existent heat spreader.

The take away is A-Die has higher freq with less voltage (This is good), but the timings aren't better. In fact, probably worse at the same frequency. Though this can't be directly compared as M-Die can't do 8000 without LN2. But for this sake, we can say 6000 M-Die vs 6000 A-Die. The article suggests M-Die would have better lower overall timings.

Last thing to note is the DDR5-8000 overclock was with a SINGLE DIMM on a Z690 Kingpin. DDR5-7455 was the highest two DIMMS could go in that article AND it was also on a 2-DIMM motherboard.

 

[kapone32]

Don't buy these because it can't do 8000+ isn't a very good argument. First, it doesn't change the fact that no 4-DIMM slot Z690 MB can run 6933+ stable (or even boot). What about all the folks who just want 6000 MT/s memory?

If you read the igor article you would have seen this in the memory overclocking section.

The take away is A-Die has higher freq with less voltage (This is good), but the timings aren't better. In fact, probably worse at the same frequency. Though this can't be directly compared as M-Die can't do 8000 without LN2. But for this sake, we can say 6000 M-Die vs 6000 A-Die. The article suggests M-Die would have better lower overall timings.

Last thing to note is the DDR5-8000 overclock was with a SINGLE DIMM on a Z690 Kingpin. DDR5-7455 was the highest two DIMMS could go in that article AND it was also on a 2-DIMM motherboard.

Looks like DDR4 issues all over again. At least the transfer rates are high but maybe we should wait until latency is solved (as it always is) to get a proper kit of DDR5.

 

[ir_cow]

Looks like DDR4 issues all over again. At least the transfer rates are high but maybe we should wait until latency is solved (as it always is) to get a proper kit of DDR5.

Yep. A-Die might be the new speed king, but just like DDR4, you don't want to break the Gear Ratio. That's going to be around DDR5-8000 for Alder Lake without EXTREME voltages on the IMC. In the coming years, my guess is 8000 MT/s with CL30 will be the best DDR5 will ever see. Give it another year

 

[thegnome]

iCUE ruining these sticks.. Wish some pretty decent kits like these actually got some more premium features than badly working RGB and one single XMP profile..

 

[tomcug]

No one should bother with these... the next gen ICs are ready and already available in no name kits, and they reach close to 8000mhz.

While that is somewhat true, DDR5 clock scaling is so poor on Alder Lake CPUs that it doesn't really matter.

 

[cellar door]

Don't buy these because it can't do 8000+ isn't a very good argument. First, it doesn't change the fact that no 4-DIMM slot Z690 MB can run 6933+ stable (or even boot). What about all the folks who just want 6000 MT/s memory?

If you read the igor article you would have seen this in the memory overclocking section.

The take away is A-Die has higher freq with less voltage (This is good), but the timings aren't better. In fact, probably worse at the same frequency. Though this can't be directly compared as M-Die can't do 8000 without LN2. But for this sake, we can say 6000 M-Die vs 6000 A-Die. The article suggests M-Die would have better lower overall timings.

Last thing to note is the DDR5-8000 overclock was with a SINGLE DIMM on a Z690 Kingpin. DDR5-7455 was the highest two DIMMS could go in that article AND it was also on a 2-DIMM motherboard.

Of course I have read igor's article - why else would have I linked it here? Think about that...

If you re-read my comment, I mention "and they reach close to 8000mhz" - I didn't say 8000 exactly. 7455 is a hell of a lot better than the 6000 this corsair kit able to achieve at its best. The frequency more than makes up for slightly worse timings, the test results in igor's test speak for themselves. Igor even mentions that the speeds he reached are 'modest' when compared to what other testers got with these kits.


And if anyone need more reasons to skip the corsair kit, just look at its exorbitant price. Or the fact that corsair switches ic's in their kits ALL the time, what this review has might be completely different even a month from now. But hey feel free to buy this corsair kit, it's your money.

Anyways that just my opinion.

 

[tomcug]

the test results in igor's test speak for themselves

Question is - do they? Except for useless AIDA/Geekbench memory benchmarks and PyPrime, in other synthetics overclocked Hynix A-die is 2% faster than overclocked Samsung B-die and in gaming tests (1080p) it's quicker by 0-1%. That indeed is an extraordinary advantage!

 

[GrimReaper85]

Don't buy these because it can't do 8000+ isn't a very good argument. First, it doesn't change the fact that no 4-DIMM slot Z690 MB can run 6933+ stable (or even boot). What about all the folks who just want 6000 MT/s memory?

Also AMD AM5 platform should be go-to in 2 days anyway.

I already read something about it:

DDR5-6000 Memory To Be The Sweet Spot For AMD Ryzen 7000 "Zen 4" CPUs, Will Offer 1:1 Infinity Fabric Ratio With EXPO

DDR5-6000 is looking to be the sweet spot for AMD Ryzen 7000 "Zen 4" AM5 Desktop CPUs, offering a 1:1 Infinity Fabric ratio with EXPO support.

wccftech.com

Higher frequency DIMMs are supported but as soon as you go above the DDR5-6000 limit, you will drop down to a 1:2 IFC. We are told that DDR5-6400 running at 1:2 will produce poor results and is not recommended if you are looking for better gaming performance.

There's a couple of strange statements here to me. 1:2 IFC all the way up to 3000 MHz IF, that is pure wrong.
Because here it says it is 1733 by default and it goes up to 2000, rare ones more than 2000:

AMD confirms Ryzen 7000 memory 'sweet spot' is DDR5-6000 - VideoCardz.com

AMD Ryzen 7000 memory sweet spot it DDR5-6000 Robert Hallock, Technical Marketing Manager for AMD, confirmed on official Discord channel what would be the best memory configuration for Ryzen 7000 series. For Ryzen 7000 series, the best choice for DDR5 memory would be 6000 MT/s. By sweet spot...

videocardz.com

I also take it from what was written that AUTO FCLK has a fixed value of 1733. Or maybe you can even set it manually.

Then only above 7000 it is worth setting FCLK:UCLK:MCLK to 2:1:1. At DDR5-7200 FCLK would be 1800 MHz.

But what I didn't account before reading your post, AMD will probably also have some DDR speed limit, like you say it is 6933 for Z690.
What if X670 only goes up to 6400 or so. Then really no point in buying Hynix RAM. They are more expensive than now easily obtainable Samsung 6000 36-36-36 (260€).

What about UCLK (memory controller) limit? But that shouldn't be a problem, at least on X570 you could set UCLK = MCLK/2.
There will probably be some good AM5 OC guides out there anyway.

Then another strange thing I read here, P4-630 says his 6000MHz 36-36-36-76 are SK Hynix. Does Hynix really make 36-36-36? Because G.Skill Hynix are 30-38-38 and 30-40-40 and 32-38-38.

 

[ir_cow]

Then another strange thing I read here, P4-630 says his 6000MHz 36-36-36-76 are SK Hynix. Does Hynix really make 36-36-36? Because G.Skill Hynix are 30-38-38 and 30-40-40 and 32-38-38.

@GrimReaper85 Yes. I have a SK-Hynix 6000 CL36-36-36 in my hands. They do exist.

 

[GrimReaper85]

G.Skill probably uses Samsung for them or there is no rule? Is it possible to check which ICs it uses?

 

[P4-630]

Then another strange thing I read here, P4-630 says his 6000MHz 36-36-36-76 are SK Hynix. Does Hynix really make 36-36-36? Because G.Skill Hynix are 30-38-38 and 30-40-40 and 32-38-38.

My set is CL 36-38-38-76 originally (SK Hynix), I selected to run a Samsung memory profile in the BIOS which is 36-36-36-76.

 

[Wolverine2349]

With a Core i9 13900K on an Asus Z790-F Strix and then after switching motherboard to Asus Z790 hero wondering if the Strix was not up to the task, these are the results I have gotten with both mobos confirming that DDR5 is difficult to run at XMP with many kits

I had 2 kits of 32GB G.Skill Trident Z5 6400 32-39-39-102 and 7200 34-45-45-115. Both would BSOD or freeze randomly during an OCCT Variable Large Data Set run at XMP 1 and XMP 2 settings. Even tweaking other things they still failed. I had tried reducing the speed of the 7200 to 6400 and 32-39-39-39 timings and would still randomly fail. Then I lowered it to 6200 and it seemed more stable as it finally got through a full 1 hour run of OCCT Large Data Set variable without fail. However a 2nd run, froze up completely at around 30 minutes.

Frustrating after trying so many things and no where to turn, I was close to going back to a DDR4 board to pair with my 13900K.

But one last thing I decided to try Corsair Vengeance DDR5 6000MHz CL 36-36-36-76 kit of 32GB. Apparently the G.Skill Trident Z5 are Hynix A and M Die kits and the Corsair is a Samsung kit. Was wondering maybe if my mobo or Hynix based kits are just not as stable at XMP settings or in general and are only better at tweaking and achieving higher frequencies if you have the patience.

So far one full pass at DOCP/XMP1 settings on the Corsair 6000 kit.

On a 2nd run of OCCT Large Data Set Variable test and almost 40 minutes in and no freeze/BSOD nor errors. Hopefully this is the answer.

Perhaps maybe the lower heat spreader underneath the Noctua NH-D15 Fan is better for my setup as well as the Trident Z5 are taller and less room to maneuver the NH-D15 fan.

Or are Asus mobos just better with Samsung than Hynix.

Its strange how now Hynix is supposedly the better RAM with DDR5 where as with DDR4 it was Samsung and Hynix was mediocre or crap. Perhaps Samsung is better for stability at XMP out of the box for both??

Assuming this Corsair Vengeance 6000 continues to pass and I run into no issues.

I am really hoping this is finally it as I certainly did not want to have to settle for less than 6000MHz DDR5 speeds even though even 4800MHz default outperforms good DDR4 mostly except for certain latency sensitive only type tasks. I definitely was hoping for 6400MHz and certainly wanted at least 6000 for DDR5 to be worth it.

SPD Hub Temp averages 53C to 57C under load. It peaks at 63C though whenever it does if it even does it is so fast I cannot even see it when I sit in front of it to babysit it during the test. Are those temps ok. Those were ironically the same temps I was getting with the G.Skill Trident kits. I wonder if Samsung is less sensitive to temps within spec for stability??

 

[ir_cow]

@Wolverine2349 I think you might have something else going on in your system. I've had nothing but good results with 13900K (13th gen). Pretty much all my M-Die kits can run DDR5-7000-7200 with 1.5V using a basic $250 Z790 MB. 13900K in a Z690 and I can also reach 6800-7000 on 4-DIMM MBs and 7600 on Z690 Tachyon. Switch over to A-Die and I'm capped at DDR5-8000 until I get a better motherboard.

I've had issues before with mounting pressure messing with memory, though usually it just don't boot anything. Also if you are running 4-DIMMs (4x 16GB), even with 13th Gen, DDR5-6000+ is a hard to achieve. DDR5-7000 boots, but not stable until DDR5-6600 after messing with voltages.

 

[Wolverine2349]

@Wolverine2349 I think you might have something else going on in your system. I've had nothing but good results with 13900K (13th gen). Pretty much all my M-Die kits can run DDR5-7000-7200 with 1.5V using a basic $250 Z790 MB. 13900K in a Z690 and I can also reach 6800-7000 on 4-DIMM MBs and 7600 on Z690 Tachyon. Switch over to A-Die and I'm capped at DDR5-8000 until I get a better motherboard.

I've had issues before with mounting pressure messing with memory, though usually it just don't boot anything. Also if you are running 4-DIMMs (4x 16GB), even with 13th Gen, DDR5-6000+ is a hard to achieve. DDR5-7000 boots, but not stable until DDR5-6600 after messing with voltages.

I have never run 4 DIMMs. I use only 2 and have them in Slots A2 and B2 like every mobo manufacturer recommends.

Could temps be an issue. All 3 RAM kits I have tried hover around 55 to 57C under full load per SPD Hub reading in HWInfo64. Is that too hot or is that an ok temp??

And sadly Corsair Vengeance 6000 I ran into stability issues once again though took longer to fail.

And I know it is not CPU as RAM runs and passes fine multiple OCCT large Data set with no issues if I have it on Auto which puts it at SPD settings gimped at CL 40 4800MHz.

 

[ir_cow]

@Wolverine2349 Sorry to hear you are having problems. Since you aren't running 4-DIMMs and 57c is okay, you may have to do some more troubleshooting. Usually I start to get WHEA errors at 65C+. I will do more investigation about temps though. It brings up a few questions.

 

[Wolverine2349]

@Wolverine2349 Sorry to hear you are having problems. Since you aren't running 4-DIMMs and 57c is okay, you may have to do some more troubleshooting. Usually I start to get WHEA errors at 65C+. I will do more investigation about temps though. It brings up a few questions.

Thats based on the reading in HWInfo64 SPD hub correct??

Technically for all the RAM there were times it peaked at 63C, but I never saw that reading realtime and it was so fast or an error under Maximum Temp. It always hovered around 57C under current temp for load running OCCT Large Data Set Variable Test.

 

[ir_cow]

Thats based on the reading in HWInfo64 SPD hub correct??

Yeah, I use HWInfo64 SPD Hub temp sensor as well.

 

[SoppingClam]

My corsair 2x16gb 5600mhz vengeance DDR5 RGB has those same timings.
Upped the speed to 6000mhz and voltage from 1.25 to 1.35 and seems they have no problem with it.
Using the CMH32GX5M2B5600C36W
Repackaged and xmp profile changed to 6000mhz?

 

[SoppingClam]

My corsair 2x16gb 5600mhz vengeance DDR5 RGB has those same timings.
Upped the speed to 6000mhz and voltage from 1.25 to 1.35 and seems they have no problem with it.
Using the CMH32GX5M2B5600C36W
Repackaged and xmp profile changed to 6000mhz?

Had no issues running the 2x16gb no problems at 6000mhz CL 34-36-36-76 in instead of standard xmp 5600mhz CL 36-36-36-76