A look at the advantages of high-bandwidth memory, including power and cost benefits.
High-bandwidth memory (HBM) is a JEDEC-defined standard, dynamic random access memory (DRAM) technology that uses through-silicon vias (TSVs) to interconnect stacked DRAM die. In its first implementation, it is being integrated with a system-on-chip (SoC) logic die using 2.5D silicon interposer technology.
In June 2015, AMD introduced its Fiji processor, the first HBM 2.5D design, which comprises a graphics processor unit (GPU) connected with four HBM stacks using a silicon-based interconnect. The Fiji processor is powering AMD’s latest generation of Radeon Fury graphics cards for high-performance gaming. This event was the turning point for HBM, opening the floodgates for system integrators to adopt HBM as a memory solution for high-bandwidth, low-power products. In addition to graphics, HBM is being used in high-performance computing (HPC), data center and networking applications.
To read more, click here.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Leave a Reply