The goal of high-performance computing (HPC) is to pursue
quick and efficient computing performance. The computing performance of the early vectored computers,
considered to be the ‘first-generation HPCs’, paralleled computers and computer clusters can mainly be enhanced
by improving the CPU frequency. The technical route for this has seen a progression from the traditional single core
to multi-cored processers. So, currently, the running-CPU-cores of HPCs has reached an order of magnitude
of 100,000 . Therefore, the whole system structure has become more and more complicated, resulting in increases
in power consumption and running costs. Meanwhile, the demand for HPC has risen with the continuous
development of application research. The traditional technology route is therefore approaching its
developmental limits and incurring both high levels of resource wastage and operating costs .