A Research Of Quantum Computer Simulation Based On CPU+GPU Hybrid Architecture Cluster

Quantum computing is a frontier interdiscipline of physics, mathematics, computer science, and engineering. Compared to classical computers, quantum computers have advantages in solving some nonlinear problems. Generally, quantum computers can solve some difficult nonlinear problems in affordable time, like NP problems. However, present technology still cannot reach real quantum computers with high qubit computing ability, either few existing quantum algorithms can reduce time complexity significantly. To improve quantum theory and quantum computing processor hardware implementation, quantum computer simulation is an important alternative option. This thesis is a quantum computer simulation study focus on the model of quantum computer based on quantum circuit, and CPU+GPU hybrid architecture technology, to help build quantum computer simulation platform and research on quantum computing simulation methods. The goal is to provide a theoretical basis for the construction of research on simulating multi-bit quantum computing supercomputers. Including: 1. Design a CPU+GPU hybrid architecture based quantum computer simulation system, to build a large-scale quantum computing simulation platform; 2. Optimize data transmission between nodes and internal nodes between the CPU and multi-GPUs taking advantages of the hybrid architecture cluster storage and simulation algorithm; 3.Use caches to improve GPU data transmission bandwidth and simulation ability based on the GPU multi-level storage architecture; 4.Use QFT algorithm to evaluate the quantum computer simulation system. This thesis will be able to provide theoretical support and systems feasibility validation for large-scale quantum circuit simulation on the design and implementation of simulation systems, the hybrid architecture cluster, the algorithms to optimize system performance, the system assessment and some other aspects.