Asus XMP I vs II
Asus XMP I vs II
Looking for insights on Asus XMP modes I and II? It seems tricky, especially with the 11700k + B560M-A setup. Previously, a Z490 board worked at 3200 MHz but had issues elsewhere. I hadn’t used XMP since building, but recently tried both modes. Mode I uses Asus’s primary timings while mode II relies on RAM settings. The documentation leans toward mode I, but my experience showed instability and crashes. I’m considering reverting to XMP off or testing gear 2. Maybe mode II is more stable? I’m still exploring further.
The issue with ASUS is that each BIOS update alters the timing settings for DRAM, meaning Mode I might function on some versions but fail on others. It's unclear how much their BIOS differs on Intel platforms; I'm using Ryzen and customizing timings. I often need to capture my BIOS configurations since saved profiles can break after a BIOS upgrade.
I know well enough that just having an XMP profile is not 100% guarantee to run at that speed but now I have another variable doubling the possible combinations to try. I wonder if XMP I (Asus values) is more performance than stability orientated? I need a repeatable ram leaning stability test so might set that option again and run something like aida64 ram test, or maybe prime95. If I can prove instability quickly that'll give me more confidence to try out other combinations looking for both stability and performance. With Rocket Lake I also have the gear 1/2 thing, since 3200 on my model is officially gear 2, but mobo defaults to gear 1 for performance. I'd take stability over performance for general use.
I'm just beginning testing with aida64, focusing only on RAM usage for the stress test. XMP off, lasted one minute without issues. No major concerns here, just a quick review. XMP I, gear 1 showed an error after 30 seconds. That's fine, it looks like it can be reproduced easily. XMP II, gear 1 had an error after 9 seconds. The timing differences between the two XMP modes seem to be the issue, not the subtimings themselves. XMP II, gear 2 is stable so far—after about 16 minutes of continuous use. Switching from XMP I to II revealed those timing discrepancies: RAS# to RAS# Delay L, RAS# to RAS# Delay S, and so on. The cycle times are consistent across the board. I’m not sure what the Auto values were without checking the documentation, but it doesn’t seem worth revisiting unless someone else asks. Maybe the problem lies with the RAM speed or the chipset implementation. It worked smoothly at 3200 gear 1 on a Z490 board and was stable overall. If gear 2 functions, I’ll stick with that and move forward.
It's also possible your memory needs adjusted voltage settings since you probably have a Vdroop from the LLC, making the actual voltages slightly reduced compared to their intended value.
It's a B series board, and it's unclear if it supports OC features like that. It's great that Intel now permits "mem OC" on non-Z boards. Running the CPU stock seems sufficient—no need to adjust voltages for stability. I suspect the issue lies more with the motherboard design than the processor itself. I haven't had a faulty CPU before, but my experiences vary depending on the case. So far in testing, it's stable at level 2. I've completed an AIDA64 mem test for over an hour and a custom Prime95 test for another hour without any errors. I plan to run a P95 blend to verify performance before moving forward. I'd rather enjoy gaming at 99% performance than spend more time troubleshooting stability. Technically, I still enable OC for RAM even in level 2, since I'm using XMP rather than JEDEC timings.
Gear 2 refers to a specific gear setting, and the command rate is the speed at which that gear operates.
Intel's Rocket Lake is similar to the RAM threshold that triggers above 3600 on AMD starting from Zen 2. For details, check reputable Rocket Lake evaluations.