If you are considering an upgrade to AMD’s latest AM5 platform, your research will likely cover not just which CPU to choose, but also the compatible motherboard and the DDR5 memory kit that suits your needs.
As of now, we have evaluated and tested nearly all AM5 processors, in addition to numerous B650 and X670/X570 motherboards. It has been a while since we focused on memory performance, so today we are offering a brief update on G.Skill’s new CL26 memory and its performance at DDR5-8000 speeds.
Why We Utilize DDR5-6000 CL30 Memory for Testing
Since the launch of AM5, our testing has involved G.Skill’s Trident Z5 Neo RGB DDR5-6000 CL30 memory, in a 32GB kit priced at around $110 – a very competitive price point. AMD initially supplied this memory to reviewers for assessing Zen 4’s capabilities, asserting that it delivered optimal performance while stating that DDR5-6000 was the platform’s sweet spot, which indeed proved accurate for Zen 4.
With Zen 5’s arrival, AMD’s messaging became somewhat muddled. They maintained that DDR5-6000 was the sweet spot and urged reviewers to continue using the same DDR5-6000 CL30 memory. Concurrently, they advertised the new 800-series motherboards, claiming that X870 boards would deliver superior performance due to support for DDR5-8000.
This caused confusion for several reasons, most importantly because the 800 chipsets were effectively rebranded versions of the 600-series chipsets, revised for Zen 5. Thus, if an X870 motherboard supports DDR5-8000 memory, it should also be achievable on an X670 board – and indeed, this has become evident.
However, complicating matters, DDR5-8000 support is determined by the motherboard rather than the memory controller or chipset of Zen 5. For instance, in testing all tested X870/X870E motherboards, we discovered that about half of the 21 boards evaluated supported DDR5-8000, while the others were unstable or failed to boot altogether.
This assessment employed the same Ryzen 9 9950X CPU and DDR5-8000 memory, so it was clear that the motherboard was the limiting factor. It’s possible some of these boards could support DDR5-8000 with a BIOS update, but we’ve yet to confirm this.
As it stands, DDR5-8000 is not ideal for all X870E motherboards, despite AMD’s assertions. From previous tests, it has also proven not to represent an optimal performance configuration – and this will be re-evaluated today.
The issue with DDR5-8000 not always providing better performance – or a significant performance advantage – over DDR5-6000, despite a corresponding 33% increase in theoretical bandwidth, stems from the operational frequency of the integrated memory controller located in the I/O die. Operating with DDR5-6000, the memory clock is 3,000 MHz, which can be matched by the UCLK (Unified Memory Controller Clock Frequency), which determines the memory controller speed. This allows both memory and controller to operate in a 1:1 ratio at 3,000 MHz.
However, exceeding a DDR5 memory clock of 3,000 MHz necessitates the integrated memory controller to function at a 2:1 ratio, as it cannot surpass 3,000 MHz while maintaining stability – at least, for the vast majority of silicon. Therefore, in the case of DDR5-8000, which operates at 4,000 MHz, the memory controller defaults to a 2:1 ratio and runs at just 2,000 MHz – resulting in a 33% lower frequency compared to DDR5-6000.
That said, the added bandwidth provided by DDR5-8000 can occasionally counterbalance the 2:1 ratio disadvantage, resulting in improved performance – provided your AM5 motherboard is stable at this frequency.
To see how this plays out, we are comparing the Ryzen 7 9700X using DDR5-8000 against several DDR5-6000 configurations, including G.Skill’s new CL26 memory, and we will also analyze a baseline DDR5-5600 configuration. Many readers might be interested in the 9800X3D, but we decided not to include that CPU in this test – adding that would complicate things considerably, and we already know it doesn’t show significant sensitivity to memory performance. The most recent example of this was with the 9950X3D, indicating that DDR5 performance is more pertinent to non-3D chips.
Now, let’s delve into the data…
Benchmarks
Cyberpunk 2077: Phantom Liberty
Beginning with Cyberpunk 2077 at 1080p, we observe a performance increase of up to 12% when transitioning from the DDR5-6000 CL30 memory utilized in our reviews to the DDR5-8000 configuration, which yielded the best results. However, it was only 2% quicker than the newly evaluated DDR5-6000 CL26 memory. Surprisingly, there was little distinction between the DDR5-6000 CL40 and CL30 configurations, with the DDR5-5600 memory performing closely as well.
Upon increasing the resolution to 4K, the performance difference between the slowest and quickest configurations narrows down to merely 5%. Conversely, the delta in 1% lows expanded, exhibiting a larger margin of 13%.
Horizon Zero Dawn Remastered
The second single-player title employed in our assessments is Horizon Zero Dawn Remastered, where we discerned a broader range of results. The DDR5-6000 CL30 review memory averaged 169 fps, which is 5% quicker than the official base spec memory. Transitioning to DDR5-8000 provided an added 7% performance increase over the review memory, reaching 180 fps, although this was just 2% faster than the newly released DDR5-6000 CL26 kit.
Transitioning to a 4K resolution, the performance margins are considerably closer. The DDR5-8000 memory achieved a mere 5% advantage over the DDR5-5600 base configuration. Notably, we observed at most a 6% increase transitioning from the DDR5-6000 CL30 memory to the DDR5-8000 kit, with essentially a 1% enhancement in average frame rate.
Marvel Rivals
For competitive multiplayer shooters, where reducing GPU load may result in utilizing medium to low-quality settings, memory performance often plays a more significant role—especially in CPU-constrained situations. Testing Marvel Rivals at 1080p revealed a 17% improvement in 1% lows when moving from the base DDR5-5600 specification up to DDR5-8000, alongside a 9% increase in average frame rates. However, DDR5-6000 CL26 memory demonstrated performance very close to that of the DDR5-8000 configuration.
In 4K, performance scaling became somewhat unpredictable. A drop in performance was seen solely when using the DDR5-6000 CL40 memory, which performed comparably to the DDR5-5600 base memory.
Counter-Strike 2
Lastly, we analyze Counter-Strike 2. Here, too, the DDR5-8000 and DDR5-6000 CL26 kits exhibited nearly identical performance, as did the CL28 memory. We noted just a 4% improvement when transitioning from the CL30 review memory to the premium DDR5-8000 kit. Similarly, the DDR5-6000 CL40 and DDR5-5600 CL38 memory delivered comparable results, lagging slightly by a few percentage points.
At 4K, the results were predominantly restricted by the GPU, with all configurations exceeding 450 fps.
Memory Options and Pricing
To summarize, when CPU-limited, DDR5 memory can enhance the performance of AMD’s most recent generation of Zen 5 processors, as anticipated. However, based on our findings, Zen 5 does not seem to be as sensitive to memory as prior Ryzen generations.
For instance, the new CL26 memory only delivered a maximum uplift of 9% compared to the CL30 memory used in our Zen 5 evaluations. This marked improvement occurred, but typically, the gains hovered around 4% to 5%. While DDR5-8000 provided up to a 12% faster experience, notably, there was minimal difference in performance between DDR5-6000 CL26 and DDR5-8000 CL38.
Memory performance (at this level) is only significant in scenarios involving competitive shooters or games that tend to be more CPU-limited than GPU-limited.
To restate, memory performance is truly only crucial in situations where you engage in competitive shooters or games with a heavier CPU load rather than GPU. In GPU-intensive single-player titles—like Indiana Jones and the Great Circle, Alan Wake 2, and Horizon Zero Dawn—premium memory typically results in minimal performance enhancement. In these instances, favoring sweet-spot memory, which we have consistently identified as DDR5-6000 CL30, is advisable.
When evaluating exclusively 32GB kits, it becomes evident that DDR5-5600 isn’t worth the lower cost. A good-quality kit retails for approximately $80, whereas a reliable DDR5-6000 CL30 kit is only $10 more. On the other hand, DDR5-6000 CL28 kits escalate to around $120, with the latest CL26 kits priced at $180 or more – making these harder to justify when compared with CL30.
Furthermore, you would need to allocate around $170 for new DDR5-8000. Although there may be occasional lesser-priced kits, they remain significantly more expensive than DDR5-6000 CL30 options.
As has been suggested since Zen 4… we endorse pairing your Ryzen AM5 processor with DDR5-6000 CL30 memory.
Therefore, as has been reiterated since Zen 4, it is advisable to couple your AM5 processor with DDR5-6000 CL30 memory. On a related note, we have previously tested a broader array of memory configurations using the Ryzen 7 7700X across eight games with the RTX 4090, yielding similar outcomes. The DDR5-6000 CL30 memory used in our evaluations consistently exhibited optimal performance, and manual fine-tuning utilizing timings from Buildzoid resulted in only a modest 4% increase on average.
As for the emerging low-latency DDR5-6000 CL26 memory, it will be interesting to observe if its pricing becomes more competitive shortly. Some performance gains exist; however, at present, the premium significantly outweighs the advantages.
Finally, we are about to commence testing on every available AMD B850 motherboard, so it will be fascinating to verify how DDR5-8000 support varies across this array. For now, this concludes our brief update.
