RAM speed has slowed after adding four sticks.
RAM speed has slowed after adding four sticks.
Hello, I’m checking to see if there are any topics I might have missed. I’m running my PC for two distinct purposes—gaming and CAD/simulation. I recently upgraded from two 16GB RAM sticks to four, mainly because some simulations were consuming more memory. After the change, my RAM speed dropped from 3600MHz to around 2133MHz. When I enable XMP, it doesn’t improve performance. I attempted to adjust frequency and timings myself, but it didn’t help. I suspect the issue might be related to the dual-channel setup on the motherboard. I’m wondering if it makes sense to experiment with slower frequencies or timings that are a bit lower than my current RAM speed but higher than what’s currently installed. There’s also a question about disabling the extra RAM sticks while gaming. My system specs: Ryzen 5900X (PBO enabled), MSI MAG MPG B550, Tomahawk 4x Trident Z Neo, 16GB RAM (3600-16-16-16-36), RTX 3070 TI.
64gb kit including four 16gb drives? All drives use that 16-16-16 timing? Yes, that’s basically quad rank Samsung B Die – super stressful for the IMCC and really tough on the board. It’s also a bit stressful trying to set the OC to 1.2v or 1.3v if the first doesn’t work; if neither works, just set RAM speed to 3200.
4x16GB of Samsung B die seems to have some issues getting up and running. You might need to adjust a few settings, but be aware you could have to reduce the memory frequency to 3466 for it to function properly. Experiment with the SOC voltage—ideal levels usually sit between 1.1 and 1.2V, and staying below 1.2V is important for CPU health. If you're not keen on that, set it to 1.15V and it should work fine for most processors. Try tweaking the ProcODT; on older B350/B450 boards it can assist with more aggressive memory overclocks, though B550 boards often handle it well on their own. Increasing the DRAM voltage to 1.4 or 1.45V may stabilize higher-speed memory, but only if you need it, as this isn’t always effective outside specific cases. Check if any of these changes help; if not, lowering the speed and retrying could be your next step. Zen 2/3 doesn’t keep improving indefinitely with SOC—most CPUs cap around 1.2V, and pushing beyond risks damaging them quickly. The SOC on these chips is quite sensitive. Just remember, 1.3V will likely harm the CPU soon.
It might help to check things just in case. Also, I need references for the unsafe situations above with non-APU Ryzens; I’m not convinced until I see solid reports showing CPUs failing or breaking down. Even then, I’d have to run my own tests to confirm (like what I can do with 1.55V all day). I assume internal components maintain a stable voltage, so for Intel it’s around 1.45V and for X58 it’s about 1.6V. Still, I’ll need around $120 for a 5600 and more funds to get into an AM4 setup—it’ll take some time before this machine sells. Are you asking about permanent damage or just temporary issues? Also, would you mean the voltage is pushed too high, causing instant failure rather than gradual wear?
AMD and most motherboard makers advise staying below this threshold, and it simply doesn't apply in practice. Many processors perform better at lower voltages, with fewer models exceeding 1.15V effectively. This isn't about taking risks; doing otherwise would likely lead to failure without significant gains. Focus on the XOC team, who consistently suggest a maximum of 1.2V for regular use.
I was checking whether 1.3v would actually allow it to post. If it doesn’t post, I’ll just lower the vsoc setting since nothing seems to help. Running at 1.3v for a couple minutes won’t hurt much, even if I can’t get 1.7v on my e8400 for those 5.1ghz bios—it still runs at 4.32ghz without any obvious issues. I still need references for degraded or heavily overclocked CPUs if you want me to take it seriously. Don’t expect much improvement beyond 1.2v on non-Apu rigs, even with dual or quad configurations; I just don’t want to spread misinformation about overclocking effects or damaging parts. (Like saying a Ryzen needs 1.6v for destruction—genius, but not true.) I’ll still need to test once I have around $120 before I commit. The 5600g is interesting because it’s the same price as a 5500 and has a solid IMC, so I can try boosting it if I don’t end up killing it. But until then, I’m just going to keep testing.
it's about the memory controller on Ryzen. Trying a higher clock speed and increased timings works. Boosting the RAM voltage up to 1.45 is possible—some people say you have enough capacity. They mentioned someone with better results at around 3600 MHz but at 1.36V. Your setup has two sticks, but you're considering four for stability. It seems like a kit might be needed since Ryzen generally doesn't support four sticks. Check if it's listed in the specifications or a forum post. You usually want two sticks plus a kit rather than random ones thrown together. It’s common knowledge that Ryzen isn’t ideal with four sticks, so you might need to purchase a compatible kit.
Oh, you think it's because of the 16-16 timings, right? That could definitely be it...* But I've never heard of BIOS being particularly tough on the imc (though I'm not saying it's not true). It just seems strange since I specifically bought my cheap Trident Z (no Neo, unless I'm talking Matrix, which is a different story). I knew it would work easily and run on anything—just didn't check the QVL. But then again, having four sticks on a Ryzen is definitely risky... *But! maybe the person who set it up just did so and it's not actually a problem, who knows? You'd need the part number to be sure.*
It’s a simpler RAM setup such as Micron Rev. E or DJR that consistently performs better than B die on any platform, even when both are set to their maximum limits. It generally offers the best compatibility outside of X99, since it was originally designed for that chip. Still, it remains one of the most challenging memories to drive, particularly when dealing with multiple ranks. Quad rank B die in that case is essentially a major issue.