If you're thinking of upgrading to AMD's current AM5 platform, you're probably researching not just which CPU to buy, along with the supporting motherboard, but also which DDR5 memory kit you'll need.

To date, we've reviewed and tested almost all AM5 processors, along with countless 600 and 800-series motherboards. However, it's been some time since we looked at memory performance, so today we've got a quick update for you, focusing on G.Skill's new CL26 memory and DDR5-8000 performance.

Why we use DDR5-6000 CL30 Memory for Testing

Since the introduction of AM5, we've been testing with G.Skill's Trident Z5 Neo RGB DDR5-6000 CL30 memory, using a 32GB kit that currently costs $110 – a very reasonable price. AMD initially provided this memory to reviewers for evaluating Zen 4's performance, claiming it enabled optimal results while also stating that DDR5-6000 was the platform's sweet spot, which certainly held true for Zen 4.

With the release of Zen 5, AMD's messaging became a bit unclear. They continued to claim that DDR5-6000 was the sweet spot and recommended that reviewers use the same DDR5-6000 CL30 memory. At the same time, they promoted the new 800-series motherboards by claiming that X870 boards would provide the best performance thanks to support for DDR5-8000.

This was confusing for a few reasons, most notably because the 800 chipsets were essentially rebranded 600-series chipsets, refreshed for Zen 5. So, if an X870 motherboard supports DDR5-8000 memory, then the same should be possible on an X670 board – and this has since turned out to be the case.

However, it gets a bit more complicated because DDR5-8000 support is limited by the motherboard, not the Zen 5 memory controller or chipset. For example, when testing every single X870/X870E motherboard, we found that about half of the 21 boards tested worked with DDR5-8000, while the rest were unstable or simply failed to boot.

This testing was conducted using the same Ryzen 9 9950X CPU and DDR5-8000 memory, so the limiting factor was clearly the motherboard. It's possible that some of these boards could support DDR5-8000 with a BIOS update, but we haven't been able to confirm this yet.

As it stands, DDR5-8000 isn't ideal for all X870E motherboards, despite AMD's claims. Based on our past testing, it's also not necessarily an optimal configuration for performance – but we'll revisit that today.

The reason DDR5-8000 isn't always faster – or significantly faster – than DDR5-6000, despite a 33% increase in theoretical bandwidth, comes down to the frequency at which the integrated memory controller in the I/O die can operate. When running DDR5-6000, the memory clock is 3,000 MHz, and this can be matched by the UCLK (Unified Memory Controller Clock Frequency), which sets the speed for the memory controller. This allows the memory and controller to run in a 1:1 ratio at 3,000 MHz.

However, going beyond a DDR5 memory clock of 3,000 MHz forces the integrated memory controller to run at a 2:1 ratio, as it can't exceed 3,000 MHz while maintaining stability – at least for the majority of silicon. So, when using DDR5-8000, which runs at 4,000 MHz, the memory controller defaults to a 2:1 ratio and operates at just 2,000 MHz – 33% lower than with DDR5-6000.

That said, the additional bandwidth offered by DDR5-8000 can sometimes overcome the penalty of the 2:1 ratio, resulting in better performance – assuming your AM5 motherboard is stable at that frequency.

To find out how this plays out, we're comparing the Ryzen 7 9700X using DDR5-8000 with several DDR5-6000 configurations, including G.Skill's new CL26 memory, and we'll also include a baseline DDR5-5600 configuration. Many of you are probably curious about the 9800X3D, but we didn't include that CPU in this testing – it would have added a lot more work, and we already know it's not particularly sensitive to memory performance. This was most recently demonstrated with the 9950X3D, so DDR5 performance is more relevant to the non-3D chips.

Alright, let's get into the data…

Benchmarks

Cyberpunk 2077: Phantom Liberty

Starting with Cyberpunk 2077 at 1080p, we're seeing up to a 12% performance improvement when moving from the DDR5-6000 CL30 memory used in our reviews to the DDR5-8000 configuration, which delivered the best results. However, it was only 2% faster than the new DDR5-6000 CL26 memory. Unexpectedly, there was very little difference between the DDR5-6000 CL40 and CL30 configurations, and even the DDR5-5600 memory kept up closely.

When increasing the resolution to 4K, lowering the frame rate to just over 100 fps, the performance gap between the slowest and fastest configurations shrinks to just 5%. However, the difference in 1% lows grows, with a larger 13% margin.

Horizon Zero Dawn Remastered

The other single-player title used in our testing is Horizon Zero Dawn Remastered, where we saw a wider spread of results. The DDR5-6000 CL30 review memory averaged 169 fps, making it 5% faster than the official base spec memory. Opting for DDR5-8000 provided an additional 7% performance boost over the review memory, reaching 180 fps, though it was just 2% faster than the new DDR5-6000 CL26 kit.

Switching to 4K resolution, the performance margins narrow significantly. The DDR5-8000 memory was only 5% faster than the DDR5-5600 base configuration. We saw, at most, a 6% gain from the DDR5-6000 CL30 memory to the DDR5-8000 kit, and just a 1% improvement in average frame rate.

Marvel Rivals

For competitive multiplayer shooters, where you're more likely to reduce GPU load by using medium to low quality settings, memory performance tends to have a greater impact – particularly in CPU-limited scenarios. Testing Marvel Rivals at 1080p, we observed a 17% increase in 1% lows when moving from the base DDR5-5600 spec up to DDR5-8000, while average frame rates improved by 9%. That said, DDR5-6000 CL26 memory delivered performance very similar to the DDR5-8000 configuration.

At 4K, performance scaling was somewhat unusual. We only saw a drop in performance when using the DDR5-6000 CL40 memory, which performed comparably to the DDR5-5600 base memory.

Counter-Strike 2

Finally, we have Counter-Strike 2. Here again, the DDR5-8000 and DDR5-6000 CL26 kits delivered nearly identical performance, and the same was true of the CL28 memory. We observed just a 4% improvement when moving from the CL30 review memory to the high-end DDR5-8000 kit. The DDR5-6000 CL40 and DDR5-5600 CL38 memory also produced similar results, falling only slightly behind – by a few percent.

At 4K, the results were nearly entirely GPU-limited, with all configurations achieving over 450 fps.

Memory Options and Pricing

So there you have it – as expected, when CPU-limited, DDR5 memory can improve the performance of AMD's latest generation Zen 5 processors. However, based on what we've seen here, Zen 5 doesn't appear to be as memory-sensitive as previous generations of Ryzen processors.

For example, the new CL26 memory provided at most a 9% uplift over the CL30 memory used in our Zen 5 evaluations. That's a notable improvement, but in most cases, the gains were in the 4 – 5% range. DDR5-8000 was up to 12% faster, but overall, there was very little performance difference between DDR5-6000 CL26 and DDR5-8000 CL38.

Memory performance (at this level) only matters in scenarios where you're playing competitive shooters or games that tend to be more CPU-limited than GPU-limited.

Again, memory performance really only matters in scenarios where you're playing competitive shooters or games that tend to be more CPU-limited than GPU-limited. In GPU-heavy single-player titles – such as Indiana Jones and the Great Circle, Alan Wake 2, and Horizon Zero Dawn – premium memory typically offers very little in the way of extra performance. In those cases, you're better off sticking with sweet-spot memory, which we've consistently found to be DDR5-6000 CL30.

Looking exclusively at 32GB kits, it's clear that DDR5-5600 isn't worth the savings. A decent kit will cost around $80, while a good-quality DDR5-6000 CL30 kit is just $10 more. However, DDR5-6000 CL28 kits jump to around $120, and the new CL26 kits are priced at $180 or more – making them hard to justify when compared to CL30.

You're also looking at spending around $170 for DDR5-8000, though there may be occasional kits that are significantly cheaper, they're still much more expensive than DDR5-6000 CL30.

As has been the case since Zen 4... we recommend pairing your Ryzen AM5 processor with DDR5-6000 CL30 memory.

So, as has been the case since Zen 4, we recommend pairing your AM5 processor with DDR5-6000 CL30 memory. As a side note, we've previously tested many more memory configurations using the Ryzen 7 7700X in eight games with the RTX 4090 and found similar results. The DDR5-6000 CL30 memory used in our reviews delivered optimal performance, and even with manual tuning using timings from Buildzoid, performance only improved by a further 4% on average.

As for the new low-latency DDR5-6000 CL26 memory, we hope to see it become more competitively priced in the near future. There are some performance gains to be had, but at the moment, the premium far outweighs the benefits.

We're also about to begin testing every single AMD B850 motherboard, so it'll be interesting to see how DDR5-8000 support holds up across that lineup. For now, that wraps up this quick update.

The above is the detailed content of Is DDR5-8000 Worth It? The Ryzen AM5 Test. For more information, please follow other related articles on the PHP Chinese website!