Chicago
Track Your Order
0
3 reasons why I run my RAM slower than its rated speed
When I first built my current PC with the Ryzen 9 5900X, I picked up 64GB of DDR4 memory rated to run at 3600MT/s. At the time, this was the sweet spot for Zen 3 processors, and since I wanted enough capacity for gaming, streaming, and multitasking for years to come, I went with four 16GB sticks. I was a little skeptical about using four sticks at first, but fortunately, my kit had no trouble running at its advertised speed. However, that changed when I picked up the Ryzen 7 5800X3D last year.
The same RAM kit that worked perfectly fine with the 5900X for over three years started giving me headaches with the 5800X3D. Sometimes, my PC would fail to boot, and other times, it would randomly freeze or crash under load. Mind you, I hadn't even changed the motherboard for the new CPU. That's when I realized I had no choice but to run my RAM at 3200MT/s instead of 3600MT/s. So, if you've ever wondered why someone wouldn't just run their RAM at its rated speed, let me walk you through my own reasons.
Stability matters more
The 5800X3D's memory controller isn't as forgiving
I expected my RAM kit to perform just as well as it did with my Ryzen 9 5900X, since the 5800X3D uses the same Zen 3 architecture. After all, I was using the same premium Asus ROG Strix X570-E motherboard with the same BIOS settings that had given me no trouble for years. Little did I know that a CPU's memory controller plays a huge role in how far you can push your RAM without running into stability issues. When I started experiencing crashes and boot failures, I immediately went through various Reddit threads to find a solution.
That's when I learned that the 5800X3D has a weaker memory controller compared to the 5900X, and that it simply doesn't tolerate higher speeds like 3600MT/s and beyond very well. Sure, some users had no trouble running their kits at 3600MT/s, but unfortunately, I wasn't one of the lucky ones. From what I've seen, AMD's 3D V-Cache design seems to limit overclocking headroom compared to standard Ryzen chips. This isn't specific to the 5800X3D, but applies to all X3D chips.
So, don't be surprised if you have to settle for slightly lower EXPO speeds with a 7800X3D or 9800X3D, especially if you've filled all four DIMM slots. But then again, these EXPO and XMP speeds aren't guaranteed.
Four sticks of RAM don't always play nice
Filling all four DIMM slots can sometimes limit overclocking headroom
I probably would have gotten away with running my RAM at 3600MT/s if I had gone with two 32GB sticks instead of four 16GB ones when I built my PC. What I didn't realize back then is that filling all four DIMM slots puts more strain on the memory controller, making it harder to maintain higher frequencies. While my 5900X's memory controller allowed me to get away with 3600MT/s for years, the 5800X3D simply wasn't as tolerant, which is why I had to settle for lower speeds.
The combination of a less tolerant memory controller and four populated DIMM slots quickly exposed stability issues that I couldn't iron out, no matter how much I manually tweaked voltages or loosened timings. Dropping to 3200MT/s was really the only way to make my PC fully stable again. This is one of the reasons why I now regret running four sticks of RAM instead of two. At this point, I'm sure I won't repeat that mistake when I upgrade to DDR5, since it's even less forgiving with four sticks at higher EXPO speeds like 6400MT/s and beyond.
Performance hit isn't really a concern
Because the 5800X3D isn't as memory sensitive as the 5900X
When I first dropped my RAM speed from 3600MT/s to 3200MT/s, I expected a noticeable dip in performance, just like most PC enthusiasts would. After all, I had always seen benchmarks showing Ryzen CPUs scaling better with faster memory. But to my surprise, the results were better than I expected. My favorite games, which are mostly CPU-bound, didn't feel any less smooth at high frame rates. The 0.1% and 1% lows still held up well in my experience, and I didn't run into any frame pacing issues.
This is mainly because of the 5800X3D's massive 96MB of L3 cache, which helps reduce how much the CPU relies on RAM for gaming. As a result, memory frequency becomes less of a bottleneck than it would be on a non-X3D chip like the 5900X. Sure, you might notice a small gain in benchmarks at higher RAM speeds, but the real-world difference is negligible. All in all, running my RAM at 3200MT/s for the sake of stability didn't feel like a downgrade, thanks to AMD's 3D V-Cache technology. Now I know I don't need to chase the fastest RAM kits when I upgrade to a newer X3D chip in the future.
I'll sacrifice some speed for stability any day
As much as I love chasing speed when it comes to PC components, I've learned that it isn't always worth the trade-offs. You might be quick to point out that I could've reached the rated speed with two sticks, but I had to work with what I already had. Of course, now that I've experienced these issues, I'll be more thoughtful about RAM configuration when I upgrade. For now, I don't mind sacrificing some frequency to ensure my PC remains stable, especially when my X3D chip makes up for the lower frequency with its massive cache. There's no doubt that pushing for higher speeds is more exciting, but stability is what gives all of us PC users peace of mind in the long run.