Reza Unveils Second-Gen DDR5 MRDIMM Chips, Boosting Speeds to 12800 MT/s
Reza has taken a major step forward in memory technology with the launch of its latest DDR5 MRDIMM interface chips. Designed for high-performance computing (HPC), AI, and data centers, the new RRG50120, RRG51020, and RRG53220 chips enable second-gen MRDIMMs to achieve speeds of up to 12800 MT/s—a staggering 4000 MT/s improvement over the first generation.
How DDR5 MRDIMMs Push Boundaries
DDR5 MRDIMMs (Multiplexer Rank DIMMs) utilize Multiplexer Chips to deliver double the data throughput compared to standard DDR5 DIMMs. This makes them ideal for bandwidth-intensive workloads like AI training and HPC. While first-gen MRDIMMs peaked at 8800 MT/s, Reza’s latest chips provide the following advancements:
Speed: Achieves up to 12800 MT/s, delivering 35% more bandwidth.
Power Efficiency: The MRCD RRTG50120 reduces power consumption by 45%.
Enhanced Protection: The PMIC RRG53220 offers better electrical protection and efficiency.
Advanced Data Buffering: The MDB RRG51020 optimizes multi-channel data reuse.
These technologies are set to enter production in the first half of 2025, catering to the increasing demand for energy-efficient and high-performance systems.
A Future-Ready Solution for AI and Data Centers
Reza’s MRDIMM innovations promise to revolutionize the industry by delivering higher memory bandwidth and capacity while maintaining a low power footprint. According to Davin Lee, Reza’s Executive, the company is collaborating with industry leaders to set the standard for next-gen computing technologies.
For those looking for even higher memory capacities, Reza is also developing Tall Form Factor (TFF) MRDIMMs, which stack more memory chips in the same module width for enhanced scalability.
The Bigger Picture
While MRDIMMs aim to lead in server and data center applications, CUDIMMs (Compute DIMMs) have already surpassed 12000 MT/s on mainstream motherboards, catering to different markets. Together, these technologies signal the dawn of an era marked by unparalleled performance and efficiency.
Are you excited about the potential of MRDIMMs for HPC and AI? Share your thoughts below on how this could reshape the tech landscape!
