DDR5 — it’s all PC gamers can take about now that AMD Ryzen 7000 is about to launch. Although Intel has supported DDR5 since the launch of its 12th-gen Alder Lake processors, Ryzen 7000 is the catalyst that will kill last-gen DDR4 off for good. When you next upgrade your PC, you’ll need DDR5, but paying up for a faster kit of memory may not translate into real-world performance gains.
One of the best DDR5 kits will still offer a great gaming experience, but the delicate balance of speed and latency puts high-end DDR5 in a precarious position. On one hand, faster DDR5 can offer practical differences in some games, but on the other, even faster kits can result in lower performance. And in some games, RAM speed doesn’t matter at all.
It sounds like a lot, and it is. But stick with me, and I’ll help you understand what to look for when buying your first DDR5 kit so you can get past misleading marketing and have the best gaming experience possible.
RAM is confusing. It has specific nomenclature that isn’t applicable to other components in your system, plus loads of specificity. The term “RAM” isn’t even a correct phrase to use, as it’s colloquially used to describe DRAM despite the fact that it’s an umbrella term that can also apply to other types of RAM.
Gamers mostly care about the main spec marketed on RAM kits: frequency. Frequency is actually the data rate of the memory (expressed as megatransfers per second), but you can still think of it as frequency like you would with a CPU or GPU. This is one of the few specs that RAM shares with your other components, and you’d be forgiven for thinking that a higher frequency means faster speed. That’s not always the case, though.
Going a level deeper, you have the CAS latency of the RAM (you’ll see it as “CL38” or something similar on DDR5 boxes). This is the time it takes for the RAM to hand off data when called upon by the CPU, and lower latencies are preferred. Using the CAS latency and data rate, you can calculate the real-world latency by dividing the CAS latency by the data rate, then multiplying by 2,000. DDR4-3200 memory with a CAS latency of 14, for example, has a latency of 8.75 nanoseconds.
Memory speed matters for gaming. By how much is a different question.
It’s important to take both the data rate and CAS latency into account when choosing your RAM kit, as a faster, more expensive kit on paper can actually result in identical performance to a cheaper kit. I’ll dig into that more in the next section, so hold tight.
In the early days of a new memory standard like DDR5, you’re usually looking at high latencies so companies can market high speeds (regardless of how that balance actually shakes out for performance). That’s not the case with DDR5, though. It has already matured enough for the bandwidth improvements to overtake the higher latencies compared to DDR4. That means the speed of your DDR5 actually matters for gaming. By how much, though, is a different question.
With the technobabble out of the way, let’s get to the fun stuff: benchmarking. I tried out three DDR5 kits at different data rates and latencies to see if there was a difference in gaming performance, and I found a few differences in some titles.
I ran all of my tests with an Intel Core i9-12900K, 12GB Nvidia RTX 3080, and an MSI Z690 Carbon Wi-Fi motherboard. The relationship between the CPU and RAM is critical, so keep in mind that DDR5 performance with AMD’s Ryzen 7000 processors might be different from what I found here. It should be largely similar to previous AMD generations, though, with a similar architecture at the helm.
I tested four speeds: DDR5-4800 CL38, DDR5-5200 CL38, DDR5-6000 CL40, and DDR5-6200 CL42.
Most, if not all, DDR5 kits will run at 4800MHz out of the box, so you’ll need to turn on either Intel Extreme Memory Profile (XMP) or AMD’s new EXPO standard for the advertised speed. Toggling on the DDR5-5200 kit didn’t provide an immediate uplift. I actually saw slightly slower results in Gear Tactics, but F1 2022 saw a solid 3% improvement.
That trend continued into DDR5-6000, with Gears Tactics jumping by 3% and F1 2022 by an additional 8%. That quickly changed with the PNY XLR8 Gaming DDR5-6200 kit, though. I actually saw lower performance in Gears Tactics and an identical result in F1 2022. Then there was Red Dead Redemption 2, which I’ll circle back to it in a moment.
I should note that this PNY kit was the only one with three XMP profiles on board, including a DDR5-5600 option that matched my DDR5-6000 results.
If you put on your math cap for a moment, these results make sense. The DDR5-6000 kit actually has the lowest real-world latency out of the three, so it’s giving the best result out of these kits. Lower latency isn’t inherently better across memory generations, so always take it incontext of bandwidth. Solid DDR4-3200 CL14 kits still have lower latency overall, but the decreased bandwidth means they aren’t as performant as DDR5 kits with higher latency.
For the upcoming generation, the balance of CAS latency and speed is important so you don’t buy a kit you don’t need. In the previous generation, we saw this dynamic between DDR4-3200 CL16 and DDR4-3600 CL16. The 3600MHz kit is faster, sure, but the latency is identical. That means spending up for a DDR4-3600 kit would effectively be wasted money.
That’s what we’re seeing here with DDR5-6000 and DDR5-6200. With nearly identical latency, you’ll see the same or slightly lower performance with the faster kit, especially depending on the memory dies inside.
Just like raw data rate, you shouldn’t put all the value on latency. Red Dead Redemption 2 is a prime example that the games you play are the largest influence on performance. I saw no difference across the four DDR5 speeds tested, showing that this game isn’t particularly sensitive to memory speed (it’s GPU bound in many cases, as you’ll typically find with modern games).
Although it’s interesting to look at DDR5 speeds and how they impact gaming performance, that raw data doesn’t tell you which kit to buy. After all, you can’t just buy a bunch of RAM kits to find the best speed for the games you play.
The only place to start is price. There isn’t any real difference between the base DDR5-4800 kits you’ll find and DDR5-5200. A 2x16GB kit of Corsair Vengeance DDR5-4800 is $150 and the DDR5-5200 is only $2 more. Big deal. There are some DDR5-5600 kits between $160 and $200, but most kits go toward RGB lighting instead of speed as prices climb.
If you want to jump to DDR5-6000, you’ll spend $220 at minimum for the same capacity. Beyond that, anything is fair game, with some kits jumping up to $300 or above. Clearly, an extra $70 isn’t going to buy you extra performance in every game, but as DDR5 prices continue to drop, we’ll likely see pricing tighten up among different speeds.
The DDR5 memory you buy all comes down to the games you play.
So, which DDR5 kit should you buy? At current prices, a solid DDR5-5200 kit is all most people need, offering a solid bump in CPU-limited games without costing much more than DDR5-4800. AMD says the sweet spot for upcoming Ryzen 7000 CPUs is DDR5-6000, though, so I’ll revisit this topic once we have new CPUs in hand. For now, you don’t need to climb that high.
As is always the case with PC hardware, it all comes down to the games you play. As I wrote about in my last entry on upgrading your gaming CPU, you can learn a lot about what components you should upgrade by analyzing the games you play. A big open-world game like Red Dead Redemption 2 may not care about faster memory, but a CPU-limited title like F1 2022 favors it a lot. At the end of the day, the games you want to play should be your touchstone for making informed PC upgrades.
This article is part of ReSpec – an ongoing biweekly column that includes discussions, advice, and in-depth reporting on the tech behind PC gaming.
- I built a couch gaming PC that puts the PS5 to shame — and you can too
- Turns out DirectStorage isn’t killing frame rates in Forspoken after all
- Don’t wait on next-gen gaming laptops — here’s what you should buy instead
- GeForce Now Ultimate is a winner, even halfway across the country
- RAM requirements are out of control in PC games — is it time to upgrade?