DDR5 Memory Scaling on Alder Lake Conclusion

The launch of Intel's Alder Lake has enabled options for users to consider. One of them is the Intel Z690 chipset, which supports either DDR4 memory or the latest DDR5 memory, albeit not on the same motherboard. This means the most premium Z690 models support DDR5 memory because the best hardware needs the best components to be the best. The more 'value-orientated' Z690 models typically have variants for either DDR5 or DDR4 support, with DDR4 still widely available to purchase.

Ultimately when it comes down to performance, as per our Core i9-12900K review, DDR5 memory has a superior advantage to DDR4 in mostly heavy multithreaded scenarios. With the uplift in memory overall memory bandwidth coupled with the potential for even faster memory, SK Hynix announced last year that it was planning to produce up to DDR5-8400 memory. The sky is the limit. We've reached the pinnacle of DDR4 in terms of performance across multiple AMD and Intel platforms, but now it's the time for DDR5 to make its mark, despite the fact it's only supported on Intel's Alder Lake architecture at the time of writing.


The Intel Core i9-12900K processor (Alder Lake)

The biggest question we wanted to address within this article are, 'how does DDR5 scale with frequency'.

Increasing Memory Frequency, The Performance Isn't Linear

Deciphering the bigger picture from our variety of CPU/motherboard tests and benchmarks from our suite shows that increasing memory frequency, for the most part, from DDR5-4800 to DDR5-6400, doesn't play as much of a critical role as first thought. Going from DDR5-4800 to DDR5-6400 in terms of raw MT/s is a 33.3% jump. The effect in the increase of frequency doesn't in any way relate to the real-world impact and performance increase, if any at all.

This is down to the design of Intel's Alder Lake - the bottleneck is usually somewhere else in the system. Even the infancy of the operating system and new scheduler might play a role more than the memory. Our results paint quite a simplistic picture for the most part, with three main points to take away:

  • 1. CPU intensive benchmarks with smaller memory workloads provided zero uplift
  • 2. Benchmarks where memory workloads were higher show benefit from higher frequencies/tighter latencies
  • 3. Games we tested benefitted from increased memory speed and tighter timings, but we're more likely to see a bigger improvement from an increase in CPU frequency or GPU frequency.

Taking our results from our Shadow of the Tomb Raider benchmark testing at 1440p, we barely saw an increase of 3.7% in average framerate from DDR5-4800 CL36 to DDR5-6400 CL36, with a 4% increase in the 95th percentile performance. There was a small uplift, even with the somewhat noisy results.

The TLDR of it comes down to one main point, there are a lot of titles out there, some more CPU intensive, some more reliant on graphical power. Memory frequency does play a part in uplifting that performance slightly. It might be a small part of the overall grand scheme of things, but it's still increasing performance whatever way you look at it.

Looking at the results of our WinRAR 5.90 testing, this is where we saw the most significant variation in performance from top to bottom. Going from DDR5-4800 CL36 to DDR5-6400 CL36, we saw an uplift in performance of just over 14%. What's very interesting from our testing is when we went low latency at DDR5-4800 with CL32 latencies. Tightening up the primary latencies at the same frequency netted us an additional 6.4% jump in performance, which shows there that increasing frequency isn't the only way to improve overall performance.

DDR5 Memory Pricing and Availability

One of the most frustrating aspects of building a new system or upgrading to the latest generation is availability. The current global chip shortage has made things very difficult, not just for consumers but for manufacturers too. This has resulted in low stock worldwide of computer components from processors, memory ICs, and graphics cards; yes, the mining craze has also played a massive part in gobbling up all of those precious high-performance GPUs. For memory, the issue isn't DDR5 chips themselves, but the power management controllers each module needs. There isn't enough to meet market demand.


The G.Skill Trident Z5 DDR5 memory in silver and black

Looking at the pricing and availability of the G.Skill Trident Z5 DDR5-6000 32GB (2x16) memory kit, it's available for around $430 at the time of writing. Trying to find any value in DDR5 at the moment is difficult to justify, and it's even harder to get a solid baseline on pricing, given demand outweighs the current supply. This inherently pushes pricing up to uncomfortable levels.

Comparing a kit of G.Skill Trident Z Neo DDR4-4000 CL18 32 GB (2x16) kit which currently costs around $169, the G.Skill Trident Z5 DDR6-6000 CL36 kit with the same capacity costs roughly 154% more. This large hike in price doesn't come close to the increase in performance, and as we've mentioned previously about linear performance, it's just not a subjective area to make fair comparisons. Pricing for DDR5 and trying to find value in it at the time of writing is sadly nonexistent.

Also yes, we've seen DDR5 memory kits on eBay going for over $1000. That is somewhat insane.

Final Thoughts

Since the launch of Intel's 12th generation Alder Lake processors, the availability of the processors has been relatively decent for a new launch. The biggest issue to unlocking many of the memory performance benefits is that DDR5 stock hasn't been available, unless you're willing to pay almost double the DDR4 equivalent cost. This means that users looking to use Alder Lake have had to either wait for DDR5 stock or opt for DDR4 and a relative Z690 motherboard.

While there is still value to be had from DDR4 on Alder Lake, as it's much more cost-effective, we typically always recommend holding out for the newer memory. It means the hardware has better resale value, and provides a platform for future improvement. Buying into DDR4 now means you're investing in a platform that has reached the ceiling. But in this market, perhaps just being able to buy what is available matters more. Opting for DDR4 memory, Alder Lake, and a compatible Z690 motherboard will still yield benefits over previous generations, but it's clear from our testing so far that going for DDR5 does perform better. Although the scalability of DDR5 is not as large in our testing as a user might first think, there's still much more room left in terms of raw MT/s left for manufacturers to eke out. The question is whether any of our usual software actually sees that as a bottleneck these days,

We've got some additional vendor memory kits in for testing, which we'll put into a review early next year. Stay tuned for that. 

Gaming Performance: High Resolution
Comments Locked

82 Comments

View All Comments

  • Targon - Thursday, December 23, 2021 - link

    Looking at these numbers, and how DDR5-5800 and DDR5000 both seem to have a performance penalty, but the other numbers aren't all that different implies something in the Alder Lake design or BIOS or SOMETHING isn't making very good use of the memory.

    Some may just chalk it up to the RAM, but I suspect it has more to do with Alder Lake itself supporting both DDR4 and DDR5 memory. At that point, I suspect the memory controller on the CPU and how it links to the CPU cores is at fault. If the memory controller were actually making better use of the changes in DDR5 compared to 4(dual 32 bit per clock eliminating certain wait states as one example), then we SHOULD see a definite improvement with better memory and not this trivial difference.

    It will be interesting to see if AMD Zen4 Ryzen shows better scaling between memory speeds, because if there is, that will show very clearly how poorly Intel implemented DDR5 memory support. Just getting it working isn't the same as taking advantage of the benefits.
  • mikk - Thursday, December 23, 2021 - link

    They didn't test the RAM, it's a GPU test. It's called GPU limit.
  • haukionkannel - Friday, December 24, 2021 - link

    AMD use most likely more cache, so the difference will be smaller.
  • felixbrault - Thursday, December 23, 2021 - link

    Why is Anandtech still using Windows 10 for testing?!?
  • Ryan Smith - Thursday, December 23, 2021 - link

    Windows 11 has been very, er, "quirky" to put it politely. We're keeping an eye on it and running it internally, but thus far we've found it to be rather inconsistent on performance benchmarks.

    In the interim, even when it behaves itself and doesn't halve our results for no good reason, we just end up with results similar to Windows 10. So there's no net benefit to using it right now.
  • Oxford Guy - Friday, December 24, 2021 - link

    Probably because Windows 11 is worse than 10.
  • Samus - Thursday, December 23, 2021 - link

    So basically the same story as ever, good timings mean almost as much as frequency, and paying ultra premiums for frequency still don't net reasonable returns on the investment.

    As always, just stick with reliable, quality memory at JEDEC speeds and invest the savings elsewhere.
  • Oxford Guy - Friday, December 24, 2021 - link

    ‘quality memory at JEDEC speeds’

    Lol, no. Certainly not with a CPU like Zen 1. There was a huge difference between JEDEC and 3200-speed DDR-4, without a huge cost increase. Zen 3 is optimal at 3600. I would think you’re aware of the fabric speed being tied to the RAM speed, hence there being a price-performance sweet spot. That spot hasn’t been JEDEC for a long time.

    Your point may apply to early adopter RAM at most. Once the DDR-5 market is more mature... In this situation it looks like getting Alder with DDR-4 capable of low latencies is most optimal. Too bad there’s no data on that here.
  • TheinsanegamerN - Monday, January 3, 2022 - link

    JEDEC DDR3 was 1066, and later updated to 1333 when everyone was running 1866 or 2133.
    JEDEC ddr4 is 2133 mhz, up to 2666, when 3200-3600 is the performance sweet spot.
  • haukionkannel - Friday, December 24, 2021 - link

    Well it seems that we are not memory speed bottle necked at the current state...
    And timings seems to affect more than pure band wide.

Log in

Don't have an account? Sign up now