Hybrid Memory Cube (HMC) technology is poised to take a bow in the high-performance computing space as a potential answer to the problem of one of the main bottlenecks in computing - the poor relative-performance of memory. After that, the hope is that the technology will "trickle down" into mainstream computing.
While multi-core microprocessors have, temporarily at least, helped to maintain improvements in computer performance by enabling multiple instructions to be handled simultaneously, memory technology has struggled to keep pace.
Also referred to as 3D memory, HMC technology promises higher performance combined with lower power consumption. Initially for supercomputing and high-performance servers, it is expected that it will eventually appear in graphics cards, networking devices and other more mainstream devices.
Supported by Intel and Microsoft, among others, as members of the Hybrid Memory Cube Consortium, HMC technology is a next-generation memory architecture that stacks memory chips in cubic arrays, connecting them in parallel to a logic layer and a package substrate. This ought to mean shorter paths, greater capacity and speed, and lower energy consumption.
According to figures from memory chip maker Micron, one of the companies involved in the development of HMCs, the technology offers 15-times the performance while using 70 per cent less energy per bit compared to standard DDR3 memory. It also takes up less space.
Micron and Samsung are the two main developers of the technology, setting up the HMC Consortium in October 2011 to get industry momentum behind their work. They are now promising new HMC devices for this year after getting field programmable gate array manufacturers Altera and Xilinx on board, as well as IBM Microelectronics for a technology called Through Silicon Vias (TSVs) that will form part of the architectural specification.
In August 2012, the consortium released the initial draft of the HMC interface specification ahead of a final version that is due imminently. When that appears, it will be a race against time to see whether Micron, Samsung or Hynix are first to market with a working product.
With ARM, HP, memory-chip maker Hynix and IBM also "developer members" of the consortium alongside Intel, Microsoft, Micron and Samsung - along with "adopter members" numbering more than 80 strong - the technology now certainly has momentum in the run-up to its launch this year.
The next question, though, will be how long it takes before the technology is offered as an option when businesses and consumers configure their next PCs.
Sometimes, the power of the mainframe is the most cost effective answer. Computing's Peter Gothard puts Computing's readers' questions on the future of the mainframe to IBM's Z13 expert Steven Dickens.
This Dummies white paper will help you better understand business process management (BPM)