Problem with classical RAM on MSI PRO Z790-P and T-Force Delta 6000MHZ system
Problem with classical RAM on MSI PRO Z790-P and T-Force Delta 6000MHZ system
They’re identical RAM modules purchased from the same supplier. You’re absolutely correct—they should match in speed, maybe one is faulty. I was expecting this to be consistent, especially since they share a similar manufacturer and have both performed well for nearly a year. It’s puzzling. Slot #1 is labeled MMIBA2, which should be the leftmost unit; I’ll focus my testing around that stick once I have more time. Thanks for the advice—I’ll revisit the BIOS settings multiple times to explore manual RAM configuration. Today I’m running EXPO again, hoping it will yield new results since both models reference Intel’s EXPO, though nothing has changed yet. I might go ahead and order two 32GB sticks or one 32GB unit if budget allows, or check if the PCI slot is the problem.
For reference, your chipset runs an Intel i7 and requires XMP settings. The comparable option for AMD is EXPO. The clock speeds might be similar, but your system's BIOS likely blocks EXPO when paired with an Intel processor.
It seems you're dealing with various ICs in your setup, which can lead to compatibility issues. There are many of them due to differing requirements and specifications, causing some mixed configurations to behave unexpectedly. For testing Super PI runs, I need clarification on the procedure since it won't perform as expected on a single stick compared to dual-channel operation. Make sure all clockspeeds and RAM timings remain consistent across tests to clearly identify any variations.
@Sawa Takahashi Your approach was understandable, but the system performance was quite poor. It seemed the BIOS allowed me to test it briefly before I had to restart and go back to a profile-free setup. XMP didn’t help much either. In MSI Bios Click 5 there isn’t an option to fine-tune settings for the specific IC, nor can you manually adjust RAM usage. I still have it set to 6000MHZ-32 with "memory try it." Should I lower it to 5600? @Beerzerker You’re also correct—when ordering parts, they mentioned two identical RAM sticks. If there was a mix-up or an issue caused the IC to fail, that could explain the problems. I thought benchmarking would be slower with just one stick, but I haven’t tried it yet since I have tight deadlines for a 2000-word paper this Sunday. I’ll wait until then to experiment. Thanks for your advice. In any case, I’ll order two 32GB DDR5 sticks. That way, if the slot is PCI and I need virtual machines for classes, I can use one stick; if a RAM stick fails, I’ll have a fresh 64GB ready. If neither fits, I might need to reset CMOS, update BIOS, and hope for the best.
A new batch of sticks should resolve the issue, provided you avoid another batch that was hastily produced. I’ve encountered this before, so it’s not a major concern.
The seller should fulfill the exchange request for RAM. You could reach out to them, explain the problem, and ask for a replacement. If they are trustworthy, they will handle it on your behalf. I tend to be cautious when purchasing RAM and prefer buying directly from Crucial. I haven’t faced any problems with the 15 models used in compatibility tests before.
Here’s a revised version of your message:
Thanks for your assistance in diagnosing the issue! I passed my exam and installed 64(32x2) RAM today, then swapped it out. Everything worked smoothly with the full 64GB DDR5 6000MHz XMP configuration. It felt a bit expensive at $200, but it was a great gift for myself—$100 if I sell the remaining RAM (warranty included). I also liked the new B&W colors and how they matched the system. In the future, I plan to replace the fan with a custom scythe design, but that depends on needing to swap out thermal paste. Eventually, one of the RAM sticks might have failed or the motherboard’s locking mechanism didn’t engage properly during DIMMA2 attempts. I’m considering sending both RAM sticks back for testing, though I’m unsure what to do next. Probably can’t run with the current Vengeance dual-channel setup. Pi, plus other memory tests, are all working well, and I can now run many Linux virtual machines and continue my Python testing. Thanks again—I’ll pass the responsibility to Beerzerker since he’s been here longer. P.S. I forgot I kept using the MSI version of CPU-Z all along!