Intel further cements its datacentre strategy at the "Data-Centric Innovation Day" event with the launch of the 2nd generation Xeon Scalable processors aimed at enterprise, cloud and communication service providers.

Previously dubbed "Cascade Lake," Xeon Scalable processors support Optane DC persistent memory, bringing support for up to 36TB of system-level memory capacity in combination with traditional DRAM. Turbo Boost 2.0 technology ramps up to 4.4GHz, alongside memory subsystem enhancements with support for DDR4-2933 MT/s and 16 Gb DIMM densities.

Samsung announces the development of an industry first-- 3rd generation 10-nanometer class (aka 1z-nm) 8Gb DDR4 DRAM, the smallest process node for memory produced without need for extreme ultraviolet lithography (EUVL).

Set for mass production within H2 2019, the 1z-nm process allows Samsung to make 20% more 8Gb DDR4 chips per wafer compared to the 2nd generation 10nm process (aka 1y-nm). An increase in the dies per wafer means the 8Gb DDR4 memory produced is around 20% smaller than the previous generation, bringing about a decrease in production costs and, ultimately, cheaper DRAM prices on the consumer side.

Samsung takes to the Nvidia GPU Technology Conference (GTC) to present a High Bandwidth Memory (HBM2E) product-- Flashbolt a high-performance DRAM claiming data transfers reaching up to 3.2Gbps per pin.

Initially aimed for use in next-gen supercomputers, graphics systems and AI, HBM2E offers a 33% increase in speed over the previous generation HBM2, with a density of 16Gb per die (or double the capacity of the previous generation). Such improvements, Samsung says, allow a single HBM2E package to provide 410GBps data bandwidth and 16GB of memory.

Nvidia looks to grow further into datacentre space as it shells out $6.9 billion for Mellanox Technologies, the Israel-based maker of end-to-end ethernet and InfiniBand smart interconnect solutions and services.

“The emergence of AI and data science, as well as billions of simultaneous computer users, is fueling skyrocketing demand on the world’s datacentres,” Nvidia says. “Addressing this demand will require holistic architectures that connect vast numbers of fast computing nodes over intelligent networking fabrics to form a giant datacentre-scale compute engine."