Virtual Instruction Set Computing (VISC)

VISC is the architecture of microprocessors increasing energy efficiency in all segments of a computer ecosystem.

On October 23, 2014 the Soft Machines company announced architecture of VISC microprocessors.

VISC architecture – break in the direction of increase in productivity of microprocessors per 1 watt of power consumption. This development will help to increase energy efficiency in all segments of a computer ecosystem. The VISC architecture was developed as problem solving of increase in frequency of one-nuclear processors and difficulties of programming of multi-core processors.

According to one of founders of company Soft Machines, VISC will lay a way for the third wave of computing architecture. After seven years of work in the privacy mode, developers created the working prototype proving efficiency of the selected approach. The company considers, time - to say achievements came.

Representation of architecture of VISC

The concept of "virtual kernels and virtual hardware flows" is the basis for VISC architecture. New approach implements runtime allocation of resources between computing cores. The microprocessors based on architecture of CISC and RISC use "physical cores" and "program flows". This approach proved to be technology and economically insolvent because of restrictions for use of transistors and energy consumption, increase in frequency. The VISC architecture allows to carry out 3-4 times more instructions for one clock period that leads to performance improvement per watt of consumption by 2-4 times on single-threaded and multithreaded applications. In addition, VISC uses "a lightweight layer of program virtualization" that allows to use this architecture, as in existing, and new program ecosystems.

The VISC architecture is scaled by change of quantity of "virtual kernels" and "virtual flows". This approach provides the uniform architecture capable to meet the needs of applications in the most different areas: from Internet of Things before the mobile market and data centers.