Fast and scalable quantum computing simulation on multi-core and many-core platforms

Quantum computing is an emerging and promising computational paradigm that provides substantial speedup for a variety of tasks such as integer factorization, database search, and machine learning. One of the quantum computation features is the possibility of developing quantum algorithms, which could be faster than algorithms developed for classic computers. However, we are still unable to fully realize a physical quantum computer and depend on traditional computers to simulate their behavior and test quantum algorithms. This is a source of complexity since one of the challenges to simulate quantum algorithms is the exponential memory requirement. In this work, we propose a method to distribute the computation load of the simulation process between CPU and GPU to decrease the required memory and computation time. In this approach, we employ a hybrid platform, which simulates the quantum circuit in two phases using parallel array-based and recursive manners. The experimental results demonstrate speedups of 50X over the recursive method implemented on a GPU and 2.9X over the state vector method running on a multi-core CPU. Our approach is more than 10X energy-efficient for simulating 39 qubits compared to the GPU state-of-the-art technique. All in all, this approach is able to simulate more qubits over state-of-the-art GPU approaches and can be used to analyze and simulate large quantum circuits with a low-cost system instead of an expensive supercomputer.