“Our 64-layer V-NAND-based technology, allows us to double the capacity of current SATA storage,” says Samsung’s  Jim Elliott. The SSD uses  16Gb LPDDR4 DRAM based on 1xnm.

A  SATA 3.3-compliant Power Disable (PWDIS) feature allows power management in individual SSD units to maximize energy efficiency. The SSD uses 2.8W of power when reading, and 3.7W when writing.

The SSD, designsted PM883, reads data sequentially at up to 550 megabits per second (MB/s) and sequentially writes at up to 520MB/s. Random reads deliver up to 98,000 IOPS and random writes up to 28,000 IOPS.

The drive has a TBW (total bytes written) rating of 5466TB for the 3840GB drive and 10,932TB for the 7680GB drive.

The 8TB SSD is now being sampled with several server manufacturers, including two who are co-exhibiting in the Samsung booth. The drive will have the first form factor designed specifically for the dense 1U servers widely used in cloud datacenters.

The upcoming Samsung 64-layer 8TB NF1 SSD can deliver I/O at a 0.5 petabytes per second.

Measuring  30.5mm x 110mm x 4.8mm, the drive will improve space utilization and scaling options for artificial intelligence, deep learning and other hyper-scale datacenter server applications. Up to 36 Samsung NF1 SSDs can fit across the front of a 1RU server allowing nearly six times the storage capacity per RU over U.2 SSDs.

Samsung is shipping 16Gb-based 64GB RDIMMs which support interface speeds of 2666 MT/s, and offer a greater than 20% power reduction versus 8Gb-based 64GB LRDIMMs.

Use of the 16Gb monolithic chips allows the maximum DIMM density to increase to 256GB making them suitable for memory-intensive applications such as In-Memory Databases and Virtual Desktop Infrastructure (VDI).