Compatibility Concerns with Different RAM Versions and Manufacturers

[ardacf]

Hello,

I currently have 2x16 Corsair Vengeance (CMH32GX5M2D6000C36) RAM modules installed, and I recently purchased another kit of the same model. After a fresh installation of Windows 10, I’ve had no issues running AAA games like Wukong and The Last of Us, or rendering animations in software such as Arnold Maya and Blender.

However, I’ve noticed something that raised a question. According to the official website of my motherboard (Asus Z690-E) and the CPU (13900K), as you can see in the image, it recommends using Samsung as the RAM manufacturer and Version 4.43.02 for 2x16 configurations.

The RAM modules I purchased last year match this version, but the new ones are Version 5.43.13. I’m currently running all 4 modules at 5600 MHz, as per the recommendation on the Asus site.

My first question is: Will the difference in version numbers cause any issues?

Additionally, the older 2x16 RAM modules are manufactured by Samsung, while the newly purchased ones are made by SK Hynix. My second question is: Could this difference in manufacturers cause potential problems down the line?

You can access the official website of my motherboard here.

Thank you for your assistance!

Edited October 1 by ardacf

[c-attack]

If it's going to cause a problem, it's usually immediate not 6 months later -- although it will be use dependent.  It's not a good idea combine modules from different kits, even when they are the exact same versions.  Yours are not and you have two different ICs with likely two different sets of preferred timings.  Lower frequency (5600) may allow you to get by, but you need to consider if you really need an extra 32GB of RAM.  Most people do not.  In a marginal case where 32GB is a little low, you would be better off looking for a 2x24GB kit to replace your original 2x16 kit.  It would allow you run enough RAM capacity at a higher throughput and better performance than what you're doing now.  None of these new generation DDR5 boards run better with 4 modules.  You always take a penalty for doing so.  

 

What you likely want to do is run some type of memory test to see if things really are getting along.  Some games are good about crashing to let you know.  Others simply roll over it and the errors manifest as stutters that go unnoticed.  It doesn't need to be an extreme RAM test.  Things like OCCT or y-cruncher can hit errors in 30 min or less and don't require the old "run for 24 hours..." to prove stability.  

Edited October 1 by c-attack

[ardacf]

Thank you for your input! I have used OCCT in the past on my previous computer, and I will run the OCCT test as soon as possible. At this point, could you specify which tests you recommend I perform? I want to make sure I focus on the essential tests and avoid wasting time on unnecessary ones.

Additionally, what specific issues should I look out for that would indicate there is a problem?

Thank you again for your help

[ardacf]

33 minutes ago, ardacf said:

Thank you for your input! I have used OCCT in the past on my previous computer, and I will run the OCCT test as soon as possible. At this point, could you specify which tests you recommend I perform? I want to make sure I focus on the essential tests and avoid wasting time on unnecessary ones.

Additionally, what specific issues should I look out for that would indicate there is a problem?

Thank you again for your help

That's the results...

[c-attack]

Basic memory or CPU+Memory is fine. Unfortunately that is the result I was worried about. 
 

Your motherboard is already using relaxed uniform timings to try and make the two kits get along, so if this is the result at 5600 you’re looking at a very difficult tweaking challenge and not one I would recommend anyone pursue. 
 

I would return the new kit if that’s an option, then look at either a 2x24 or 2x32 kit to replace the original pair if you need the capacity. I normally would recommend the 2x24 kit for anyone on Z790 as it scales better and it substantially less expensive than 2x32. However, I don’t know how Z690 boards do with the newer 24GB modules that did not exist when those boards were released. The QVL lists for them are bleak and it’s clear not much testing was done. This is likely something you could learn from other users on Z690 with 24GB modules. We do know 6000 is more or less the safe spot for 2x32 on Z690 with max possible somewhere between 6200-6600. 

[ardacf]

7 hours ago, c-attack said:

Basic memory or CPU+Memory is fine. Unfortunately that is the result I was worried about. 
 

Your motherboard is already using relaxed uniform timings to try and make the two kits get along, so if this is the result at 5600 you’re looking at a very difficult tweaking challenge and not one I would recommend anyone pursue. 
 

I would return the new kit if that’s an option, then look at either a 2x24 or 2x32 kit to replace the original pair if you need the capacity. I normally would recommend the 2x24 kit for anyone on Z790 as it scales better and it substantially less expensive than 2x32. However, I don’t know how Z690 boards do with the newer 24GB modules that did not exist when those boards were released. The QVL lists for them are bleak and it’s clear not much testing was done. This is likely something you could learn from other users on Z690 with 24GB modules. We do know 6000 is more or less the safe spot for 2x32 on Z690 with max possible somewhere between 6200-6600. 

I also wanted to share that I spoke with Amazon, and they confirmed that they will accept the return, which makes me happy. I plan to do a fresh installation, and as you suggested, I think I need to switch to the F5-6000J3040G32GX2-TZ5RK model in a 2x32 configuration, which seems compatible. I would love to continue with Corsair, but unfortunately, there are no compatible options available in my country right now, just 2x16 kits. I would have switched to 2x48 if I could find compatible ones in the market, but those are also not available.

Thanks for your interest. 
 

 

[c-attack]

That might be for the best on the 2x48.  I don't think the Z690s handle that very well.  

[ardacf]

Asus official website states that it supports 2x48, but you’re saying that I wouldn’t get good performance. Since I’m not sure, could you explain why 2x48 would give less performance compared to 2x36?

[c-attack]

When you increase density, it makes it more difficult (impossible) to run higher frequency.  A 48GB module is denser than a 32GB.  As you go up in size from 8>16>32>48 your maximum possible frequency drops.  Z690 as the first DDR5 board was not fantastic for RAM overclocking.  32GB modules couldn't do more than 5200 or 5600 early on.  While things improved on the BIOS level, you can't change the hardware, trace routes, and physical layout of the board.  A 2x32 kit likely tops out between 6000-6400.  I am worried the 48GB modules won't do more than 5200-5600 and I haven't found anyone running them on Z690.  The QVL suggests the same.  So if you have to choose between a 2x32@6000C30 and a 2x48@5600C40, the 2x32 kit is better in every way except capacity.  So it's back to whether you are using programs eating more than 50-55GB of RAM at a time.  

[ardacf]

Thank you for summarizing the situation so well. Yes, I definitely use programs that require a lot of RAM, such as ZBrush, Maya, and Adobe AE. That's why 32GB is no longer sufficient for me. I believe 64GB will meet my needs quite well. In this context, even though the conclusion I reached is somewhat disappointing, I had to purchase the F5-6000J3040G32GX2-TZ5RK model, as Corsair’s other models are unavailable in the market. I just hope I won’t regret this decision and that luck will be on my side ♥