Curious if this is expected for such tight sub-timings with all pre-fetchers on and Hypervisor on or am I making a critical timing mistake that is sub-optimal?
Changing tFAW to 20 does nothing notable other than occasionally boost silly AIDA scores.
Changing tWTRL to 14 and tRDWR to 14 is stable but it felt like a regression, though within the margin of error of test variations.
The only thing left is to turn off GDM or attempt to stabilize 7000 2:1 — I had 7200 MT/s running with very loose timings but it would eventually error out VT3 after a few hours. With dual rank's interleaving, 7200MT/s 2:1 was faster in y-cruncher than 6400MT/s in 1:1: https://www.reddit.com/r/overclocking/comments/1kwe6r2/hot_take_ddr5_6400_11_ddr5_7200_21/