Instability in DDR5 RAM when using AM5 is a known issue.
Instability in DDR5 RAM when using AM5 is a known issue.
It's interesting since I thought we had understood this right away after Zen4's release in 2022, and I purchased it on the same day. I noticed what AMD was showcasing in their press materials for reviewers—a 2x16GB 6000CL30 kit from G.Skill. Assuming they included it with their CPUs for testing would be the best choice. Even though we spent a lot of money on the same motherboard they provided, we didn't discuss that much. Ultimately, it's proven to be quite reliable over the past two years. I did some tweaks using Buildzoid, but they didn't significantly impact real-world performance beyond benchmarks I don't prioritize, and I never bothered re-entering BIOS settings again (a real hassle with all those timing details).
I haven't built a system since intel 6th Gen so I was a bit out of the loop, I was running 4 sticks of fast DDR4 on that system with no issue, I didn't think twice about getting 4 sticks of ram for my new system until after I attempted to use Expo and that when I found out that its not ideal. It is possible to get it working if you have time or can find good settings online though and theres an element of luck I feel, AMD only guarantee 5200MHz on 2 sticks and 3600 on 4 for the 7000 series all the same.
If you were me and I was you, not so long ago since I needed more memory because my current setup was hitting RAM limits on a 2x16GB board, I’d seriously consider the 6400MHz CL32 kit... not to actually crank it up to 6400MHz, but because it likely uses high-quality M-Die Hynix RAM. It should handle CL30 at 6200MHz consistently throughout the day with tight timings on the others, using the standard 1.35V supply—something I expect from my new kit and most in that price range. If you can maintain 6200MHz and get your clock speed to around 2200MHz (which some do and others don’t, especially when pushing past 6000MHz), you’ll notice a noticeable performance gain over similar builds. The IF at 2200MHz gives a solid boost, and keeping it at 6200MHz with good timings is the rest. Tuning your secondary timings—even basic adjustments from the Buildzoid’s safe M-Die settings—can further improve results. This build seems aimed at both gaming and heavy workloads, not just pure gaming. I’m moving it toward a production/workload role while still keeping it capable for games. The jump from 2x16GB to 2x48GB was a major step in that direction.
You're focusing heavily on gaming, with some interest in video editing and 3D rendering. Regarding your RAM, a 6400MHz CL32 chip can handle CL30 speeds when you reduce the game's settings. As for identifying M-Die Hynix RAM, check the manufacturer's part number or packaging details—it usually starts with "M" followed by a specific code.
I would say you have like a 90% chance of doing 6200 CL30 with a voltage bump (I'm running 1.4v VDD & stock is 1.35v). But keep in mind my kit is 2x48GB so it's 2 sticks that are dual-sided dimms. Anything above 24GB sticks I believe has to be dual-sided (meaning there's RAM chips on both sides of the PC & not just one side), but I could be wrong. What I was told for telling M-Die from A-Die is look at the sticker on the physical sticks, if the number at the bottom right above the bar code (on GSkill RAM for sure but I believe this applies to all kits) has "###M" at the end (### meaning 3 numbers, mine have 820M) at the end of it it's Hynix M-Die. If it's "###A", it's Hynix A-Die. If it's something else, it's probably some weird stuff. Corsair, and a few other brands share that code affinity so it's the same on theirs. Unfortunately I believe it's almost impossible to tell if you're getting M-Die or A-Die (or something else) without looking at the kit in person. Both are good Hynix stuff though which is used on pretty much everything at 5600MHz or above, M-Die is more common, A-Die is less common.