cancel
Showing results for 
Search instead for 
Did you mean: 

Road to XMP stability - DDR5 4x16 (GSkill 6000CL36) on Z690 Formula

Peazmack
Level 7
Looking to share some input as well as try to get some from other, more experienced people in the field of OCing and coping with Z690/DDR XMP issues. So I went ahead of me, and bought 2 kits of GSkill DDR5 2x6 @6000MHz CL36. Of course, like for majority of us I reckon, it does not boot in either of the XMP settings. Now before I give up and switch back to using only 2 sticks instead of 4 which seem to cause more issues, looking for some input on the below. Im using latest bios available as of now for my motherboard (Z690 Formula with 12900K).

Closest to the XMP that I got is 5400 CL36 with VDD and VDDQ left alone at 1.3, System Agent left on auto (1.25 it seems) and Memory Controller on 1.23. That seems to be rock solid, had over 2.5 hours spent in Karhu's RAM Test without any errors the other night. VDD and VDDQ did not really impact the stability here, unless it went over 1.35, then I started getting errors. 1.23 for memory controller was the lowest I could go while being free of errors.

Attempted to go for 5600 CL36 tonight, but hardly any progress. Closest to Stability was VDD and VDDQ at 1.31 (I doubt it made any difference here), System Agent still on Auto and Memory Controller on around 1.29. I could boot no problem, but had constantly first ram error found within first 30 minutes of RAM Test, which I consider a fail. Any advice how to move it forward? Shall I push the Memory Controller above 1.3 threshold and look for stability there? Any other setting I'm missing here?
Additionally, should I settle for 4x16 5400 CL36 quad or switch to 2x16 dual with XMP 6000 CL36? Which would be more beneficial (I know it's rather marginally better) for gaming?
28,738 Views
65 REPLIES 65

rjbarker wrote:
+1 ....I gave up on trying to "combine" kits years ago....many years ago.....why introduce yet another "factor" into a problem....just purchase complete kits that have been "proven"....


Well, I have been using 2 kits with 2 x 8 GB DDR4 for many years, without any problem. Yes, it was known that if you want to push the RAM to the max, 4 modules are limiting you, and not recommended if you want to overclock. But being on various hardware forums for at least 20 years or so, I am not aware of the problems that currently many users are facing to get 2 kits with 2 x16 GB stable even at stock speed. And it seems in quite a few cases their system is not even booting. Also, why do we get 4 DDR5 DIMM sockets when it is virtually impossible to run the system with of all them populated? Are there 4 module DDR5 kits in the market that are binned to work together? I am not aware of them. Interestingly, MSI is offering a high end board with only 2 DIMM sockets. I guess that is the future for DDR5, because the max capacity per module is far higher than for DDR4, there is no need anymore for 4 modules, and managing only 2 DIMM sockets is certainly better/easier for the controller than operating 4 DIMM sockets.

Chainbold wrote:
Well, I have been using 2 kits with 2 x 8 GB DDR4 for many years, without any problem. Yes, it was known that if you want to push the RAM to the max, 4 modules are limiting you, and not recommended if you want to overclock. But being on various hardware forums for at least 20 years or so, I am not aware of the problems that currently many users are facing to get 2 kits with 2 x16 GB stable even at stock speed. And it seems in quite a few cases their system is not even booting. Also, why do we get 4 DDR5 DIMM sockets when it is virtually impossible to run the system with of all them populated? Are there 4 module DDR5 kits in the market that are binned to work together? I am not aware of them. Interestingly, MSI is offering a high end board with only 2 DIMM sockets. I guess that is the future for DDR5, because the max capacity per module is far higher than for DDR4, there is no need anymore for 4 modules, and managing only 2 DIMM sockets is certainly better/easier for the controller than operating 4 DIMM sockets.



4 matched modules, or rather 4 modules binned as one kit to work at a set of timings and frequency, are far less problematic if one has a good methodology in tweaking.

The problem some are facing on Z690 is directly due to combining kits and expecting them to work at XMP without manual intervention, or even at all at the vendor's SPD settings.

As the platform matures overclocking will likely become easier, but if the vendors aren't putting high-speed DDR5 kits in 2DPC configurations there's no reason to "expect" anything more than what's been validated to work by them. Some users have a hard time facing some of these expectations and like to blame everything they can for their own misadventure. Trying to validate by saying " it should work because I've mixed kits before" shows there's still some gaps in ones base knowledge


Stock operation on Alderlake with two slots per channel and all slots populated is 3600 and 4000 for dual and single rank modules respectively.
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

After bios update 1003, I was able to boot into Windows with XMP I/II profiles of 6000Mhz CL36 4x16GB but would get an error while benchmarking Rise of the Tomb Raider, I have everything maxed out and a resolution of 5120x1440, 240Hz max with vsync off.

I started messing around with the manual VDD and VDDQ but left everything else on auto, current tweaks are:

6000MHz
40-40-40-76 (auto)
VDD: 1.325
VDDQ: 1.325

I tested VDD and VDDQ from 1.260 all the way to 1.325 going up 0.005 every time for each voltage simultaneously. Bios memtest would crash almost right away but until I was able to get at least 30 minutes in, I would then run the benchmark and that would crash after 2 scenes; until I got to 1.325V. I know this is not the best stress test, but at least I am better off than I was a week ago not being able to boot passed 5200MHz.

Just to stir in more commotion, one kit is from Newegg and the other kit is from Amazon with slightly different part numbers, but GSkill has confirmed they are the same specs (not bins, obviously). By same specs, I mean frequency, latency, and voltage.

On another note, I was able to pick up a set of Corsair Dominator 2x32GB 5200MHz CL40 and cannot get anything to boot passed other than the default 4000MHz, I did notice that these are not on Asus' list, though the lower latency ones are. Very strange. I will be returning them, such a disappointment, especially for the price. They do look and feel very nice compared to the GSkill. VERY solid heatsinks.

I am planning on running more test but when I have a lot more time. For now, I will enjoy the "win".

Good luck out there!

Edwin7700 wrote:
After bios update 1003, I was able to boot into Windows with XMP I/II profiles of 6000Mhz CL36 4x16GB but would get an error while benchmarking Rise of the Tomb Raider, I have everything maxed out and a resolution of 5120x1440, 240Hz max with vsync off.

I started messing around with the manual VDD and VDDQ but left everything else on auto, current tweaks are:

6000MHz
40-40-40-76 (auto)
VDD: 1.325
VDDQ: 1.325

I tested VDD and VDDQ from 1.260 all the way to 1.325 going up 0.005 every time for each voltage simultaneously. Bios memtest would crash almost right away but until I was able to get at least 30 minutes in, I would then run the benchmark and that would crash after 2 scenes; until I got to 1.325V. I know this is not the best stress test, but at least I am better off than I was a week ago not being able to boot passed 5200MHz.

Just to stir in more commotion, one kit is from Newegg and the other kit is from Amazon with slightly different part numbers, but GSkill has confirmed they are the same specs (not bins, obviously). By same specs, I mean frequency, latency, and voltage.

On another note, I was able to pick up a set of Corsair Dominator 2x32GB 5200MHz CL40 and cannot get anything to boot passed other than the default 4000MHz, I did notice that these are not on Asus' list, though the lower latency ones are. Very strange. I will be returning them, such a disappointment, especially for the price. They do look and feel very nice compared to the GSkill. VERY solid heatsinks.

I am planning on running more test but when I have a lot more time. For now, I will enjoy the "win".

Good luck out there!



Followed the thread as I also run 2x16GB Gskill 6000/36 with Z680 Formula.
I was always able to boot with XMPI or II without issues. But since Bios 1003 I get Game crashes constantly, so almost all games just crash after 5-10 minutes gameplay. Mostly without error description, some with DX issues.
I now have dieled back momory frequency to 5800 and all seem to work again ( So I kept XMPI setting just dialed back frequency by 200MHz)
Verry strange...

SilverSurfer72 wrote:
Followed the thread as I also run 2x16GB Gskill 6000/36 with Z680 Formula.
I was always able to boot with XMPI or II without issues. But since Bios 1003 I get Game crashes constantly, so almost all games just crash after 5-10 minutes gameplay. Mostly without error description, some with DX issues.
I now have dieled back momory frequency to 5800 and all seem to work again ( So I kept XMPI setting just dialed back frequency by 200MHz)
Verry strange...


I am oddly enough happy to read the posts in this thread as I too am now in this boat where games are crashing without any good explanation. Lost Ark closing flat out or Windows throwing random memory related blue screens. Memtest ran clean with XMP on earlier. No corruption found in Windows 11. So I removed all XMP, left it at manual stock 4800 with AIOC running. Now it's stable. I might jump back in and up the speeds later this week when I've got more spare time to tweak. It's just a bummer to see a G.Skill DDR5 16x2 kit rated for 6000/CL36 not actually be able to operate at that stable.

Edit: As others have mentioned, XMP1 clocked down to 5800 seems to be stable.
| Intel Core i7 12700K @ 5.3 GHz | G.SKILL 2x16 GB DDR5 @ 6000 MHz CL36 | ASUS ROG MAXIMUS Z690 HERO | ASUS TUF Gaming GeForce RTX 3070 Ti OC Edition |

Having also joined this boat just today I'm wondering Cutsman what you (or anyone else) is using to test stability?

Is it just the inbuilt Memtest or are you running someone else? Or maybe just using the machine...

I've never really OC'd before - I usually just tell the PC to use XMP and to automatically boost the CPU but I did buy those DDR5 6000 G.Skill modules for a reason.

Any help appreciated.

TonyG

tony_goodhew@hotmail.com wrote:
Having also joined this boat just today I'm wondering Cutsman what you (or anyone else) is using to test stability?

Honestly, it's usually just playing games and whether or not the game crashes or the system blue screens. Those were the problems I was experiencing while running the RAM at 6000. I also ran TM5 with extreme profile for a bit too and saw no errors detected.
| Intel Core i7 12700K @ 5.3 GHz | G.SKILL 2x16 GB DDR5 @ 6000 MHz CL36 | ASUS ROG MAXIMUS Z690 HERO | ASUS TUF Gaming GeForce RTX 3070 Ti OC Edition |

tony_goodhew@hotmail.com wrote:
Having also joined this boat just today I'm wondering Cutsman what you (or anyone else) is using to test stability?

Is it just the inbuilt Memtest or are you running someone else? Or maybe just using the machine...

I've never really OC'd before - I usually just tell the PC to use XMP and to automatically boost the CPU but I did buy those DDR5 6000 G.Skill modules for a reason.

Any help appreciated.

TonyG


Karhu Ramtest or TM5 once in the OS
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Edwin7700 wrote:
After bios update 1003, I was able to boot into Windows with XMP I/II profiles of 6000Mhz CL36 4x16GB but would get an error while benchmarking Rise of the Tomb Raider, I have everything maxed out and a resolution of 5120x1440, 240Hz max with vsync off.

I started messing around with the manual VDD and VDDQ but left everything else on auto, current tweaks are:

6000MHz
40-40-40-76 (auto)
VDD: 1.325
VDDQ: 1.325

I tested VDD and VDDQ from 1.260 all the way to 1.325 going up 0.005 every time for each voltage simultaneously. Bios memtest would crash almost right away but until I was able to get at least 30 minutes in, I would then run the benchmark and that would crash after 2 scenes; until I got to 1.325V. I know this is not the best stress test, but at least I am better off than I was a week ago not being able to boot passed 5200MHz.

Just to stir in more commotion, one kit is from Newegg and the other kit is from Amazon with slightly different part numbers, but GSkill has confirmed they are the same specs (not bins, obviously). By same specs, I mean frequency, latency, and voltage.

On another note, I was able to pick up a set of Corsair Dominator 2x32GB 5200MHz CL40 and cannot get anything to boot passed other than the default 4000MHz, I did notice that these are not on Asus' list, though the lower latency ones are. Very strange. I will be returning them, such a disappointment, especially for the price. They do look and feel very nice compared to the GSkill. VERY solid heatsinks.

I am planning on running more test but when I have a lot more time. For now, I will enjoy the "win".

Good luck out there!


I have the Corsair Micron 2x32GB 5200MHz CL40, Dell SK Hynix 2x32GB 4800MHz kits and both are working fine in XMP 1 profile on Apex board.

Not sure which motherboard you have, I have tested the memory preset profiles in BIOS for 2x32GB Micron, Hynix and the pre-set profiles also work fine.

Corsair 5200 kit:
92200

Dell 4800 Kit:
92224

Joining the party. I’ve navigated through a lot to finally get windows to boot and have a computer that’s somewhat stable but when it starts performing multiple tasks, or even run warzone, windows freezes and it’s because my RGB freaks out. I believe it has something to do with the stability of my trident z5 6000 c40. I’ve dropped as low as 5000MHz and still freeze. Im on the latest bios and will try your settings. I have the formula with a 12900k and a 3090.