Micron Launches Advanced MRDIMM Memory for High-Performance Computing
Micron has announced the launch of its innovative new multiplexed rank dual inline memory module (MRDIMM), designed for high-performance computing and artificial intelligence data center operations. The MRDIMM provides a solution for systems needing more than 128GB of memory per slot by combining multiple layers of memory modules with a data buffer, resulting in faster data transfers and improved overall performance.
Compared to traditional TSV RDIMMs, MRDIMMs offer increased bandwidth, greater capacity, reduced latency, and enhanced performance per watt. With compatibility for Intel's upcoming Xeon 6 processors, MRDIMMs are set to revolutionize memory capabilities for virtualized environments and data centers.
Boosting Bandwidth and Reducing Latency
MRDIMMs adhere to DDR5 physical and electrical standards, bringing a 39% increase in effective memory bandwidth, over 15% improvement in bus efficiency, and up to 40% reduction in latency compared to RDIMMs. They also feature an advanced thermal design that can lower DRAM temperatures by 20 degrees Celsius, promoting better cooling efficiency and reduced energy consumption.
Ready for High-Performance Servers
The MRDIMM modules are designed for use in 1U and 2U high-performance servers, available in capacities ranging from 32GB to 256GB, and are suitable for both standard and tall form factors. With the promise of seamless integration into current server infrastructures, Micron's MRDIMMs are set to deliver reliable, scalable performance.
Volume shipping of MRDIMMs is expected to begin in the second half of 2024, in line with Intel's launch of its 128 P-core Granite Rapids-AP line. This collaboration is set to establish Intel as the leader in the fastest CPU server market for the first time in a decade, with Granite Rapids being the first platform to support MRDIMMs at 8800MT/s on the host side.
With Micron's new MRDIMM technology, the future of high-capacity, high-speed memory modules looks brighter than ever, promising to enhance performance and efficiency in data centers around the world.
