Micron has introduced its new DDR5 memory modules featuring 24GB and 48GB capacity, reports News.Mynavi.jp (via @momomo_us). The modules are compatible with both AMD EXPO and Intel XMP 3.0 profiles for quickly setting them up and are designed for desktop PCs running AMD's Ryzen 7000-series as well as Intel's 12th and 13th Generations Core processors.
Micron's new family of DDR5 memory modules feature data transfer rates of 5200 MT/s and 5600 MT/s as well as CL46 latency at 1.1V. The DDR5-5600 DIMMs come in traditional 8GB, 16GB, and 32GB versions, but the interesting capacities are the new 24GB and 48GB models. Those are probably based on 24Gb memory chips, whereas the former likely use 16Gb DRAM ICs. Meanwhile, DDR5-5200 modules are available only in 8GB, 16GB, and 32GB sizes.
Typically, 24GB and 48GB capacities are considered to be optimal for new-generation server platforms as they allow systems to precisely balance memory capacity and the number of cores, which ultimately means lower costs. Meanwhile, support for AMD EXPO and Intel XMP 3.0 profiles designed primarily for enthusiasts in mind indicate that these modules are indeed aimed at desktops.
Using a couple of 24GB or two 48GB modules instead of a pair of 32GB and 64GB DIMMs allows to build PCs with 48GB or 96GB of dual-channel memory, which are cheaper than machines with 64GB or 128GB of RAM. Meanwhile, capacities like 48GB and 96GB are more optimal for modern CPUs with 16 or 24 cores — you get potentially 2GB or 3GB per core if you're doing VMs, for example.
Unfortunately, Micron has not disclosed recommended prices for its DDR5-5200 and DDR5-5600 modules. Keeping in mind that demand for PCs is projected to be low in Q1 and DDR5 SDRAM prices are set to decline by 18–23% in Q1 2023, expect the new memory modules to be relatively inexpensive despite their high capacity and increased performance.
Stay on the Cutting Edge
Join the experts who read Tom's Hardware for the inside track on enthusiast PC tech news — and have for over 25 years. We'll send breaking news and in-depth reviews of CPUs, GPUs, AI, maker hardware and more straight to your inbox.
Anton Shilov is a Freelance News Writer at Tom’s Hardware US. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.
This is probably what Linus made in his Micron video 😊Reply
On the one hand, weird memory sizes. On the positive, a lot of 12GB and 24GB GPUs, so you can keep that VRAM to RAM ratio a nice round number.Reply
Might get 48GB of ram for no reason on my next upgrade.
This is odd, BUt I would consider 2x24GB as I rarely need all 64GB on my system, I throw most of it away to primo cache to speed things up.Reply
How is a ratio that has no importance whatsoever a 'positive' thing?Eximo said:On the positive, a lot of 12GB and 24GB GPUs, so you can keep that VRAM to RAM ratio a nice round number.
The real benefit of this half-step size is to give people cost-optimized options between 32GB, 64GB and 128GB using only one DIMM per channel for best compatibility. The "optimal amount of RAM per core" argument is specious at best since that would be entirely application-dependent, as is any tie with GPU VRAM size.
I suspect the real benefit of this half-step size is just programmatically disabling erring chips on already produced larger modules and selling this crap for a good price.InvalidError said:The real benefit of this half-step size is
Too bad these are so slow. Fine for server platforms, but no ethusiast would touch these. I'd much rather have 32GB fast ram instead of 48GB slow ram.Reply
No, it's not. The native capacity of the dies is actually different. There's no way they have so many dies with defects.Alex/AT said:I suspect the real benefit of this half-step size is just programmatically disabling erring chips on already produced larger modules and selling this crap for a good price.
SK Hynix is also launching their own 3/4-sized parts: