Micron started mass production of PCIe Gen6 SSD with read speed of 28 GB/s

Micron has officially started mass production of the first SSD solution for data centers based on the PCIe Gen6 interface. It is the 9650 NVMe SSD model, which reaches a sequential reading speed of up to 28 GB/s and up to 5.5 million IOPS during random reading, which sets a new industry record and marks the beginning of the PCIe Gen6 era in the server data storage segment.

Two years ago, Micron announced the development of a PCIe Gen6 SSD with a target speed of 26 GB/s. After additional optimization and technological improvements, the mass-produced 9650 SSD not only exceeded that goal, but compared to the previous 9550 generation, it doubled the sequential read speed and increased the write speed by about 40 percent, with twice the energy consumption.

READ ABOUT:  We got the launch trailer for the Halo Infinite campaign

PCIe Gen6 SSD 9650 brings a new era of data storage

Model 9650 comes in two variants, adapted to different types of workloads in data centers. The PRO version is intended for scenarios with dominant data reading and supports one full capacity write per day, while the MAX variant is adapted to mixed workloads and allows up to three full writes per day. Both series cover a wide range of capacities, from around 6 TB to over 30 TB, depending on the configuration.

On the hardware side, the 9650 SSD uses PCIe Gen6.2 x4 and NVMe 2.0 interface, and is based on Micron G9 NAND memory. It is the first application of the ninth generation of NAND in a data center SSD, with a six-planar architecture that enables extremely high I/O throughput. The E1.S and E3.S formats are available, with the E1.S also adapted to liquid cooling, which further improves the energy efficiency of the system.

READ ABOUT:  Hitman 3 - Year 2 new content coming early 2022

With twice the bandwidth of PCIe Gen5, PCIe Gen6 removes one of the key bottlenecks between data storage and processing power. The Micron 9650 SSD is especially important for AI workloads, where it allows GPUs to run continuously without stalls caused by slow data access, while also using lower power.

Source link