last fall, HP and SanDisk have announced a partnership to create a new memory-Driven computing computer architecture, where the key role in the performance of computing reserved not central, or GPU, and memory. Now, a year later, Hewlett Packard Enterprise – one of the two companies, which are separated by HP, – announced the successful testing of a new architecture. while this is only a prototype that proves the efficiency of the concepts and functions thanks to the compute nodes from quick-volatile memory, fast on the basis of data channels photons and special software.
The prototype still needs more and memory nodes to reveal their potential, but Hewlett Packard Enterprise is optimistic. Simulation at the design stage suggests a new architecture that operates on several orders of magnitude faster at start code than ordinary PC. In some cases the difference reaches more than 8000 times. The company wants to use this technology for servers and other high-end computing tasks, but this development can positively affect the performance of the Internet of Things devices.
There are currently an obstacle to the mass use of the new architecture is the need for non-volatile memory, and it will not change until 2018 or 2019. So do not wait for the imminent appearance of super computers. A successful test of this technology suggests the possibility of its application in future.