
When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works .
(Image credit: SK hynix) Share Share by: Copy link Facebook X Whatsapp Reddit Flipboard Share this article Join the conversation Follow us Add us as a preferred source on Google SK hynix on Thursday said that its range-topping 256 GB DDR5 RDIMM based on 32 Gb memory devices has passed the Intel Data Center Certified process, the industry's first memory module at this capacity to do so. Normally, certification of a memory module by Intel would be mundane, but this module is special; it combines capacity, low power consumption, and performance, thus potentially saving data center operators millions of dollars.
AI servers not only consume plenty of premium high-bandwidth memory (HBM) used on AI accelerators like Nvidia B300, but also tons of commodity server DDR5 SDRAM that connect to x86 processors. A single high-capacity DDR5 memory module can consume up to 15W or even 25W (depending on performance, capabilities, and workloads), so a fully equipped 12-channel Xeon 6 memory subsystem can draw between 180W and 300W, comparable to the CPU's power consumption.
A 32 Gb memory chip made on SK hynix's 1b process technology (5 th Generation 10nm-class DRAM process) consumes significantly less power than two 16 Gb memory ICs made on the company's 1a production node (4 th Generation 10nm-class DRAM node), so using a 256 GB DDR5 RDIMM based on the latest DRAMs leads to significant energy savings.
Samsung introduces SOCAMM2 LPDDR5X memory module for AI data centers
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/pc-components/dram/SPONSORED_LINK_URL
- https://www.tomshardware.com/pc-components/dram/intel-certified-256-gb-ddr5-stick-could-cut-xeon-memory-power-by-18-percent-saving-millions-of-dollars-a-32w-per-socket-reduction-could-save-millions-per-hyperscale-data-center#main
- https://www.tomshardware.com
- How to Fine-Tune an LLM on NVIDIA GPUs With Unsloth
- 3 Ways NVIDIA Is Powering the Industrial Revolution
- Over 1,800 North Korean applicants flood Amazon — suspected illicit applicants blocked by the company since April 2024
- Grab this £138 2TB Acer Predator SSD before flash prices soar further — fast Gen 4 speeds make this an essential upgrade for your gaming PC
- OpenAI's Stargate data center gets approval to receive 1.4 gigawatts of power in Michigan — some residents furious as energy company is given go-ahead by regula
Informational only. No financial advice. Do your own research.