XMP with RAM of same brand, series, frequencey, voltage, and timings but different size?

Discussion in 'General Hardware' started by Reclusive781, Dec 5, 2022.

  1. Reclusive781

    Reclusive781 Ancient Guru

    Messages:
    2,899
    Likes Received:
    1,211
    GPU:
    RX 7800 XT
    Should have made this thread earlier.

    I'm about to pair the 2x8GB kit with the 2x16GB kit. Both are going to be Dark za, same timings, frequency, and voltage. All that's different is the size. Now, I already know which slots they need to go in for it to work.

    Right now I'm worried about how XMP is going to work. I mean, am I going to get the same single profile or am I going to have two different XMP profiles available now?

    **EDIT**

    I thought about getting the 32GB kit but all I was thinking about is how unnecessary and ridiculous 64GB's of RAM is just for gaming, playing around the internet, and the very rare video editing I do. 48GB should last for this entire console generation. Especially when 32GB's becomes the new normal recommended amount.
     
    Last edited: Dec 6, 2022
  2. RealNC

    RealNC Ancient Guru

    Messages:
    5,609
    Likes Received:
    3,868
    GPU:
    4070 Ti Super
    Good question. In any board I ever had, only the XMP profiles from the modules in the primary slots were visible. No idea if that's by spec, or it's up to the board/BIOS.
     
    386SX likes this.
  3. Undying

    Undying Ancient Guru

    Messages:
    26,631
    Likes Received:
    14,006
    GPU:
    Devil RX 6900 XTU
    It depends on the board. If it sees both profiles try using one or the other and if not you can always set timings manually.
     
  4. RealNC

    RealNC Ancient Guru

    Messages:
    5,609
    Likes Received:
    3,868
    GPU:
    4070 Ti Super
    Forgot to mention that some boards (like from MSI) will always show two profiles, even if it only sees one. XMP1 and XMP2. The second one has some MSI tweaks on voltages in case XMP1 doesn't work.
     
    tunejunky likes this.

  5. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    16,080
    Likes Received:
    10,011
    GPU:
    4090@H2O
    Well I hope it works well but I have my doubts... tried to mix RAM of the same brand and size, 2x8 kits and another 2x8, and my IMC crapped out. I hope it works better for you!
     
  6. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,251
    Likes Received:
    7,521
    GPU:
    GTX 1080ti
    AMD has supported this configuration in the past, what you do is put 1 of each size on one channel and the same for the other so that each channel has the same capacity.
     
  7. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,251
    Likes Received:
    7,521
    GPU:
    GTX 1080ti
    This isn't two destinct profiles, what it does is determine if the entire xmp profile is used including secondary and tertiary timings (Profile 2), or ignore those and let the mainboard train them. (Profile 1)
     
  8. RealNC

    RealNC Ancient Guru

    Messages:
    5,609
    Likes Received:
    3,868
    GPU:
    4070 Ti Super
    There was a thread about this here:

    https://forum-en.msi.com/index.php?...p-profile-1-and-2-on-amd-motherboards.361280/

    In this image, you can see both use the same timings, but some of the voltages (starting with VSOC) are different:

    https://forum-en.msi.com/index.php?attachments/xmp-profile-1-jpg.147323/
     
  9. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,251
    Likes Received:
    7,521
    GPU:
    GTX 1080ti
  10. Reclusive781

    Reclusive781 Ancient Guru

    Messages:
    2,899
    Likes Received:
    1,211
    GPU:
    RX 7800 XT
    So far so good. I didn't see any new XMP profiles in the bios. XMP is enabled and running well so far but I haven't done any stability testing yet. Will run memtest86 I think on the 13th of this month. 6 passes should do it.
     
    fantaskarsef likes this.

  11. Reclusive781

    Reclusive781 Ancient Guru

    Messages:
    2,899
    Likes Received:
    1,211
    GPU:
    RX 7800 XT
    I opened up CPU-Z and I see that all the memory modules 8GB and 16GB are single rank which is good. This is telling me that nothing is holding anything back and the only true difference between the modules is the size. 8GB and 16GB.
     
    Last edited: Dec 6, 2022
  12. RealNC

    RealNC Ancient Guru

    Messages:
    5,609
    Likes Received:
    3,868
    GPU:
    4070 Ti Super
    Nitpicking maybe, but memtest86 doesn't test all the RAM. memtest86+ does, so I recommend testing with it first.
     
  13. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,251
    Likes Received:
    7,521
    GPU:
    GTX 1080ti
    memtest86+ can miss heat related errors,

    testmem5 with the 1usmus profile is a good go to for quick memory stability testing, it finishes within half an hour and anything that would fail a large fft should show up in that test.
     
  14. Horus-Anhur

    Horus-Anhur Ancient Guru

    Messages:
    9,247
    Likes Received:
    11,427
    GPU:
    RX 6800 XT
    Also, a good way to quick test the memory controller is running y-cruncher.
     
  15. RealNC

    RealNC Ancient Guru

    Messages:
    5,609
    Likes Received:
    3,868
    GPU:
    4070 Ti Super
    Yes. If you want to be more confident, you should test with several tools, not just one ;)
     

  16. Reclusive781

    Reclusive781 Ancient Guru

    Messages:
    2,899
    Likes Received:
    1,211
    GPU:
    RX 7800 XT
    I ran that test and no errors came back. I will run memtest86+ on the 13th.

    [​IMG]
     
  17. 386SX

    386SX Ancient Guru

    Messages:
    2,166
    Likes Received:
    2,322
    GPU:
    AMD Vega64 RedDevil
    You guys always test 3 or 5 runs and think it is stable. 30+ or it didn't happen, or 20+ at least. I do not want to come over rude, but 3 cycles is worth nothing and proofs "nothing", RAM has just reached operating temperatures then.
     
  18. Horus-Anhur

    Horus-Anhur Ancient Guru

    Messages:
    9,247
    Likes Received:
    11,427
    GPU:
    RX 6800 XT
    1usmusv3 test is ok, but not the best for TM5. And is not enough when used for 15 minutes.
    Use Anta777@Extreme, for a best stress test.
     
  19. Undying

    Undying Ancient Guru

    Messages:
    26,631
    Likes Received:
    14,006
    GPU:
    Devil RX 6900 XTU
    I've had tested the ram for hours and it looked fine then tried running some games and it turned out its not crashed in minutes.

    So yeah, nothing better than testing the games itself.
     
    Passus and cucaulay malkin like this.
  20. Horus-Anhur

    Horus-Anhur Ancient Guru

    Messages:
    9,247
    Likes Received:
    11,427
    GPU:
    RX 6800 XT
    There could be a few reasons for that.
    Most memory tests will hit the memory, but barely stress the memory controller or the CPU. So if then you play a game that happens to push harder for the memory controller or the CPU, at any moment, then it might error out. That's why I also recommend running something like y-cruncher.
    Usually when overclocking memory, it's also necessary to increase a few voltages. For example, it might be ok during the memory test. But a game might require more energy from the whole SoC, and if it doesn't have enough, it might error out.
    Some memory ICs are more sensitive to temperature than others. And when overclocking the memory, we might get closer to the point when temperature becomes an issue. During a memory test, it will only produce a bit of heat from the memory and CPU. But while playing a game, the GPU will emit a lot of heat into the whole case, that can increase the memory's temperature too high.
    We also have to consider that no test is 100% certain.
    Then there is the issue that our planet is constantly being bombarded with radiation. And sometimes some energy particle will hit our hardware and cause a bit flip, that can cause an error. This is very rare, but can happen once in a while.
     
    Undying likes this.

Share This Page