![]() |
|
Samsung Mass Produces Its First PCIe 6.0 SSD, Hitting 28.4 GB/s To Feed NVIDIA´s Vera Rubin AI Servers - Printable Version +- T0rrentinvites - #1 To Buy, Trade, Sell Or Find Free Tracker Invites! (https://t0rrentinvites.com) +-- Forum: Other Discussions (https://t0rrentinvites.com/forumdisplay.php?fid=30) +--- Forum: General PC Chat And Help (https://t0rrentinvites.com/forumdisplay.php?fid=33) +--- Thread: Samsung Mass Produces Its First PCIe 6.0 SSD, Hitting 28.4 GB/s To Feed NVIDIA´s Vera Rubin AI Servers (/showthread.php?tid=17896) |
Samsung Mass Produces Its First PCIe 6.0 SSD, Hitting 28.4 GB/s To Feed NVIDIA´s Vera Rubin AI Servers - Western - 07-08-2026 ![]() Samsung has announced the mass production of the PM1763 PCIe 6.0 SSD, which is optimized for next-gen AI servers such as NVIDIA's Vera Rubin.
The Rise In AI Requires Next-Gen Storage Solutions Such as Samsung's PM1763 PCIe 6.0 SSD, Now In Mass Production Press Release: Samsung Electronics today announced mass production of PM1763, the company’s PCIe 6.0-based enterprise solid-state drive (SSD) optimized for next-generation AI and HPC server environments. RELATED STORY DeepSeek Is Reportedly Building Its Own Inference Chip to Break Free From Both NVIDIA and Huawei ![]() As the volume of data required for AI training and inference continues to grow rapidly, enterprise SSDs (eSSDs) capable of delivering data quickly and reliably are becoming increasingly essential to AI infrastructure.
Featuring high-speed data transfer and an optimized controller architecture, PM1763 is expected to serve as a key storage solution for high-performance AI platforms. Incorporating Samsung’s 9th-generation V-NAND and a newly developed 4-nanometer (nm) controller, PM1763 significantly improves both performance and power efficiency. ![]() PM1763 is available in 4-terabyte (TB), 8TB, and 16TB capacities, with the 16TB configuration delivering industry-leading performance with sequential read and write speeds of up to 28,400 megabytes-per-second (MB/s) and 21,900MB/s, respectively, more than 2 times the performance of its predecessor, PM1753.
This level of performance allows the transfer of a 40-gigabyte (GB) large language model (LLM) in approximately 1.4 seconds, helping minimize data latency between processors and accelerators while improving overall AI processing efficiency. ![]() PM1763 is optimized for liquid-cooled server environments through direct-to-chip (D2C) cooling technology. This enables sustained peak performance even under intensive workloads and extended operating conditions. Power efficiency is also improved by more than 1.8 times compared to its predecessor, helping reduce overall data center operating costs.
|