IBM proposed the first index to measure the speed of quantum computing CLOPS
In IBM's view, there are three key attributes to measure the performance of quantum computers in the near future: scale, quality and speed. Scale and mass are measured by quantum bit number and quantum volume (qv) respectively, but there is no index about speed. Therefore, IBM proposed the first index to measure the speed of quantum computing - circuit layer operations per second (CLOPS).
IBM measures quantum computing performance through three indicators: quantum bit number, quantum volume and CLOPS.
Figure 1 three key indicators to measure the performance of quantum computing: scale, quality and speed
The number of qubits is the most basic index, which determines the amount of information that can be encoded for calculation in quantum computer, and thus determines the scale of the problem that can be solved. Because fault-tolerant computing requires a large number of qubits, scale is a key index for the development of quantum computers. The number of qubits can also be used as a resource to improve the other two indicators - quality and speed. For example, auxiliary qubits are often used to reduce the depth of circuits and improve their fidelity. Additional qubits can also be used in multi-channel programming of QPU to improve its circuit processing speed.
Quantum volume (qv) represents the realization degree of quantum circuit in quantum computing system. IBM defines QV layer as a permutation between one layer of qubits and a pair of random Su (4) double qubit unitary gates, as shown in Figure 1. QV is defined by the width or number of QV layers of the largest random square circuit that the quantum processor can successfully run (the width is equal to the number of layers).
QV is very sensitive to coherence, gate fidelity and measurement fidelity. QV is also affected by connectivity and compiler, which can effectively minimize the impact of decoherence. QV is an overall indicator, which can not be improved only by improving one aspect of the system, but by improving all parts of the system in a collaborative manner.
The circuit layer operations per second (clips) is related to how many QV circuits a QPU can execute per unit time. It is defined as the number of QV layers executed per second using a set of parameterized QV circuits, where each QV circuit has d = log З QV layer. Circuit execution time includes updating circuit parameters, submitting jobs to QPU, executing on QPU, and sending results back for processing. Then calculate clips, that is, the total number of QV layers executed divided by the total execution time.
Link:
