A Fusion Computing System and Method Based on Quantum Measurement and Control Boards

Abstract

The present invention provides a fusion computing system and method based on a quantum measurement and control board. The system includes: a CPU, a GPU, and a quantum measurement and control board; the CPU performs data interaction with the GPU and the quantum measurement and control board through a PCIE bus; the quantum measurement and control board is connected to a quantum chip; the CPU is configured to divide a predefined computing task into a classical computing subtask and a quantum computing subtask by using a preset resource scheduling algorithm; compile the quantum computing subtask into a quantum circuit; the quantum measurement and control board is configured to receive the quantum circuit transmitted by the CPU, generate a pulse sequence corresponding to the quantum circuit, and send the pulse sequence to the quantum chip for the quantum chip to execute the quantum circuit to complete the quantum computing subtask; wherein the pulse sequence is used to represent the control signal and read signal of the quantum chip; the GPU is configured to execute the classical computing subtask.

Description

Technical Field

[0001] The present invention relates to the technical field of superconducting quantum computing architectures, and particularly to a fusion computing system and method based on a quantum measurement and control board card. Background Art

[0002] In the past two decades, the field of quantum computers has witnessed rapid development, and in particular, superconducting quantum computers have become one of the platforms that have received much attention. As the number of qubits in superconducting quantum chips continues to increase, higher requirements are put forward for the scalability, synchronization, and low latency of the qubit measurement and control hardware. This is mainly because the internal space of dilution refrigerators is limited, and a more precise control system is needed to cope with the increasing number of qubits.

[0003] Currently, commercially available laboratory equipment, such as arbitrary waveform generators and data acquisition systems, although they can meet the needs of general test and measurement applications, gradually cannot meet the development of superconducting quantum computing as the complexity of quantum systems increases. Therefore, the control signals and read signals generated in front of the quantum processor need to go through successive attenuation and filtering, and the read signals coming out of the quantum processor also need to go through several stages of amplification operations during the return process. Since the coherence time of qubits is very short, a very high requirement is placed on latency in order to run as deep a circuit as possible within a limited time.

[0004] As Figure 1 shown, the traditional superconducting quantum measurement and control system is divided into three parts: a host computer that interacts with users and generates measurement and control tasks, a microwave measurement and control system, and a quantum chip in a low-temperature environment. The current superconducting quantum measurement and control system is used as a separate backend measurement and control device to achieve the operation and control of the quantum chip. The host computer needs to read the information of the current qubits from the measurement and control device through a communication network, and network latency, transmission latency, etc. will be generated during this process. Therefore, as the number of qubits increases, it is difficult for a single superconducting quantum measurement and control system to efficiently measure and control a large number of qubits.

[0005] In summary, in the NISQ era, as the scale and complexity of quantum chips increase, although quantum computing has shown quantum advantages in certain specific tasks, the types of problems it can solve are limited, and it still faces challenges such as qubit scale and errors, and it is still difficult to replace traditional high-performance computing. Therefore, in order to adapt to the growth of future computing power requirements, it is crucial to build a fusion computing architecture system with the ability to handle multiple types of tasks. Summary of the Invention

[0006] In order to meet the measurement and control requirements for a large number of qubits, the present invention provides a fusion computing system and method based on a quantum measurement and control board card.

[0007] On the one hand, the present invention provides a fusion computing system based on a quantum measurement and control board, including: a CPU, a GPU, and a quantum measurement and control board; the CPU performs data interaction with the GPU and the quantum measurement and control board through a PCIE bus; the quantum measurement and control board is connected to a quantum chip;

[0008] The CPU is configured to divide a predefined computing task into a classical computing subtask and a quantum computing subtask by using a preset resource scheduling algorithm; and compile the quantum computing subtask into a quantum circuit;

[0009] The quantum measurement and control board is configured to receive the quantum circuit transmitted by the CPU, generate a pulse sequence corresponding to the quantum circuit, and send the pulse sequence to the quantum chip for the quantum chip to execute the quantum circuit to complete the quantum computing subtask; wherein, the pulse sequence is used to represent the control signal and the read signal of the quantum chip;

[0010] The GPU is configured to execute the classical computing subtask.

[0011] Further, the quantum measurement and control board uses a Xilinx RFSOC chip.

[0012] On the other hand, the present invention provides a fusion computing method for a fusion computing system based on a quantum measurement and control board, including the following steps:

[0013] Step 1: The CPU divides a predefined computing task into a classical computing subtask and a quantum computing subtask by using a preset resource scheduling algorithm;

[0014] Step 2: The CPU compiles the quantum computing subtask into a quantum circuit and transmits the quantum circuit to the quantum measurement and control board;

[0015] Step 3: The quantum measurement and control board receives the quantum circuit, generates a pulse sequence corresponding to the quantum circuit, and sends the pulse sequence to the quantum chip; wherein, the pulse sequence is used to represent the control signal and the read signal of the quantum chip;

[0016] Step 4: The quantum chip executes the quantum circuit according to the pulse sequence to complete the quantum computing subtask, and feeds back the task execution result to the quantum measurement and control board;

[0017] Step 5: The quantum measurement and control board transmits the task execution result to the CPU;

[0018] Step 6: The GPU executes the classical computing subtask;

[0019] Step 7: The CPU makes decisions on the quantum computing subtasks and the classical computing subtasks in real time according to the task execution results, and determines the classical computing subtasks and quantum computing subtasks for the next round;

[0020] Step 8: Repeat Step 2 to Step 7 until the stop condition is met.

[0021] Furthermore, the predefined computing task includes using the Quantum Approximate Optimization Algorithm (QAOA) to solve the Max-Cut problem.

[0022] Advantages of the present invention:

[0023] The fusion computing system and method based on a quantum measurement and control board tightly integrate traditional high-performance computing and quantum computing, providing more powerful computing capabilities for processing complex tasks, as specifically shown below:

[0024] (1) Resource collaborative utilization: The classical computing resources and quantum computing resources achieve complementary advantages when solving optimization problems. The classical computing resources are responsible for problem transformation and parameter optimization, while the quantum computing resources execute quantum circuit evolution and measurement operations, improving the performance of the overall system through the collaborative utilization of hardware resources.

[0025] (2) GPU acceleration optimization: In the classical computing resources, the calculation process of the classical optimization algorithm is accelerated by the GPU. For example, the gradient descent algorithm accelerated by the GPU is used to improve the performance of the QAOA algorithm and speed up the optimization calculation.

[0026] (3) Tightly integrated hardware connection: The hardware components of the entire system are tightly integrated through high-speed connections, reducing communication latency, facilitating cascaded expansion, providing more powerful and efficient fusion computing capabilities, and providing a more powerful solution for processing complex tasks and challenging problems. Description of the Drawings

[0027] Figure 1 It is the architecture diagram of a traditional superconducting quantum measurement and control system;

[0028] Figure 2 It is the architecture diagram of the fusion computing system provided by the embodiment of the present invention;

[0029] Figure 3 It is the schematic flowchart of the fusion computing method based on a quantum measurement and control board provided by the embodiment of the present invention. Detailed Embodiments

[0030] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0031] Embodiment 1

[0032] As Figure 2 shown, an embodiment of the present invention provides a fusion computing system based on a quantum measurement and control board card, including a CPU, a GPU, and a quantum measurement and control board card; the CPU performs data interaction with the GPU and the quantum measurement and control board card through a PCIE bus; the quantum measurement and control board card is connected to a quantum chip.

[0033] Among them, the CPU is used to divide a predefined computing task into a classical computing subtask and a quantum computing subtask by using a preset resource scheduling algorithm; compile the quantum computing subtask into a quantum circuit; the quantum measurement and control board card is used to receive the quantum circuit transmitted by the CPU, generate a pulse sequence corresponding to the quantum circuit, and send the pulse sequence to the quantum chip for the quantum chip to execute the quantum circuit to complete the quantum computing subtask; wherein, the pulse sequence is used to represent the control signal and read signal of the quantum chip; the GPU is used to execute the classical computing subtask.

[0034] Specifically, the quantum measurement and control board card undertakes the manipulation and readout tasks in quantum computing, has advanced quantum measurement and control functions, and is connected to other hardware components at high speed through a PCIE bus. The CPU provides traditional high-performance computing capabilities, can cooperate closely with the quantum measurement and control board card, and quickly exchange data and instructions with other hardware components through the PCIE bus. As an accelerator for high-performance computing, the GPU can play an important role in processing complex tasks. Combined with quantum computing, it can further improve the overall performance and efficiency of the system. As a key component of quantum computing, the quantum chip works in cooperation with the quantum measurement and control board card to provide the computing power of quantum computing for the system and is connected to the quantum measurement and control board card through a microwave circuit.

[0035] In the embodiments of the present invention, the hardware platform of the system includes a CPU, a GPU, and a quantum measurement and control board, fully considering the characteristics of processor chips with different hardware structures in heterogeneous computing. The GPU has a large number of parallel cores and is suitable for accelerating repetitive operations of lightweight tasks, while the quantum measurement and control board has higher hardware acceleration performance and lower device interconnection latency. When processing computing tasks, the heterogeneous platform with the CPU as the control node integrates two types of computing and measurement and control devices, namely the GPU and the quantum measurement and control board. The control node makes decisions according to job requirements and schedules them to different computing devices for parallel execution. In this way, the fusion computing system can make full use of the advantages of various hardware resources to achieve more efficient execution of computing tasks.

[0036] It should be noted that the system places no restrictions on the number of CPUs, GPUs, and quantum measurement and control boards. Figure 2 The system shown is a structure of a single-machine multi-card platform. In this platform, multiple GPUs and quantum measurement and control boards are inserted on the main board. The CPU conducts data interaction with the GPUs and quantum measurement and control boards through PCIE. From the overall structure, this heterogeneous platform is divided into a scheduling node and a computing node. Among them, the CPU serves as the scheduling node, responsible for system logic control, application program partitioning, task scheduling, and resource management. Each computing node is responsible for executing different types of computing tasks. The measurement and control device is replaced by the quantum measurement and control board and is connected to the classical computer through the PCIE interface. Each quantum measurement and control board corresponds to several qubits in the quantum computer QPU in the quantum computing resources. This design reduces the data transmission delay between the host computer and the quantum measurement and control board and also makes the measurement and control system easy to expand.

[0037] The fusion computing system provided by the embodiments of the present invention deeply integrates quantum computing components such as quantum measurement and control boards and quantum chips with classical computing components such as CPUs and GPUs, realizes high-speed connection between quantum hardware and classical hardware through the PCIE bus, and realizes collaborative work between quantum computing and traditional high-performance computing.

[0038] As an implementable manner, the quantum measurement and control board in the embodiments of the present invention uses a Xilinx RFSOC chip, which has the following advantages:

[0039] 1. The quantum measurement and control board using the Xilinx RFSOC chip has highly integrated quantum manipulation and readout capabilities, as well as advanced radio frequency data processing and synchronization technologies, which provide an important basis for realizing multi-channel high-fidelity microwave signals. The high performance and flexible configuration of the RFSOC enable the quantum measurement and control system to meet complex computing and control requirements.

[0040] 2. Support multiple DAC and ADC channels, and have flexible configuration capabilities. It supports up to 4 channels of fixed-frequency Qubit manipulation at most, and 1 channel of readout supports 4-channel fixed-frequency qubit demodulation. This system can perform high-speed data acquisition and precise control operations, meeting the requirements for high-speed and high-precision data processing in quantum computing.

[0041] 3. The clock management circuit inside the board provides flexible clock distribution and support for external reference clocks, ensuring the clock synchronization and stability of the system, which is crucial for the collaborative work between multiple boards.

[0042] 4. Use the PCIE x8 GEN3 interface for communication between the CPU and other quantum measurement and control boards, with a maximum rate of up to 8GB / s. This high-speed data transmission ability helps to achieve fast data exchange and processing, improving the overall performance of the system.

[0043] 5. Generate multi-channel high-fidelity microwaves through the direct output method in the RF DAC Nyquist domain, which is crucial for achieving precise microwave control signals and helps to improve the stability and reliability of the quantum system.

[0044] 6. Support signal conditioning and system expansion with multiple auxiliary components, including a radio frequency front-end unit, a clock unit, and a trigger unit. These auxiliary components provide the functions required for signal processing, clock synchronization, and system-level expansion, supporting the overall performance improvement of the system.

[0045] Embodiment 2

[0046] Based on the above fusion computing system, an embodiment of the present invention provides a fusion computing method for the fusion computing system, including the following steps:

[0047] S201: The CPU uses a preset resource scheduling algorithm to divide the predefined computing tasks into classical computing subtasks and quantum computing subtasks;

[0048] Specifically, the CPU, according to the preset logic of the system, uses a resource scheduling algorithm to finely allocate the predefined computing tasks, including the classical part and the quantum part. These tasks are allocated according to resource characteristics, application characteristics, and busy conditions to support collaborative computing and the efficient operation of tasks.

[0049] S202: The CPU compiles the quantum computing subtasks into quantum circuits and transmits the quantum circuits to the quantum measurement and control board;

[0050] Specifically, when there are multiple quantum measurement and control boards in the system, the compiled quantum circuits can be split into multiple small quantum parts and then transmitted to the quantum measurement and control boards located in different positions.

[0051] S203: The quantum measurement and control board receives the quantum circuit, generates a pulse sequence corresponding to the quantum circuit, and sends the pulse sequence to the quantum chip; wherein, the pulse sequence is used to represent the control signal and read signal of the quantum chip;

[0052] Specifically, the quantum measurement and control board provides a high-speed data transmission channel through the PCIE interface, accepts and executes the measurement and control tasks of the quantum part, and generates a pulse sequence of corresponding control signals and read signals.

[0053] S204: The quantum chip executes the quantum circuit according to the pulse sequence to complete the quantum computing subtask, and feeds back the task execution result to the quantum measurement and control board;

[0054] Specifically, after the control signal and read signal go through successive attenuation, filtering, and amplification operations, they enter the quantum chip to execute the quantum computing task. The read signal after execution is collected by the quantum measurement and control board after a series of amplification operations.

[0055] S205: The quantum measurement and control board transmits the task execution result to the CPU;

[0056] S206: The GPU executes the classical computing subtask;

[0057] Specifically, the GPU executes the classical computing subtask to accelerate the processing of the quantum computing subtask. The parallel computing ability of the GPU can effectively improve the overall performance and efficiency of the system.

[0058] S207: The CPU makes decisions on the quantum computing subtask and the classical computing subtask in real time according to the task execution result, and determines the next-round classical computing subtask and quantum computing subtask; This collaborative computing method of classical and quantum optimizes the allocation of computing resources and the efficiency of task processing.

[0059] S208: Repeat steps S202 to S207 until the stop condition is met.

[0060] The fusion computing method provided by the embodiment of the present invention, through the close cooperation and high-speed connection of each hardware component in the fusion computing system, makes full use of classical computing resources and quantum computing capabilities, realizes the tight integration of traditional high-performance computing and quantum computing, and thus shows an effect beyond traditional computing. This fusion computing method provides a more powerful and efficient solution for processing complex tasks and challenging problems.

[0061] Embodiment 3

[0062] In the embodiment of the present invention, an example is given where the computing task is to use the quantum approximate optimization QAOA algorithm to solve the maximum cut problem, as Figure 3As shown in the figure, the process of performing fusion computing using a fusion computing system is as follows:

[0063] S301: The CPU constructs a Hamiltonian for the maximum cut problem and transforms it into a corresponding quantum circuit structure and evolution method.

[0064] S302: The CPU initializes the parameters of the QAOA algorithm and passes them to the quantum measurement and control board.

[0065] S303: The quantum measurement and control board drives the quantum chip to execute the parameterized quantum circuit through microwave pulses and measures the obtained expected value.

[0066] S304: The expected value is passed back to the CPU. The CPU uses these expected values for further optimization calculations, adjusts the parameters to minimize the objective function, and uses an optimization algorithm (such as gradient descent) for parameter optimization. Part of the optimization calculations can be accelerated by the GPU to improve the computing efficiency.

[0067] S305: The updated parameters are passed to the quantum measurement and control board again for performing the next round of quantum circuit evolution and measurement.

[0068] S306: Iterate in this way until a certain number of iterations is reached or the convergence condition is met.

[0069] S307: Finally, a set of suitable parameters is obtained. Acting on the initial state with the generated quantum circuit and measuring the final state can obtain the solution to the maximum cut problem with a high probability.

[0070] Through the efficient fusion computing system architecture and GPU acceleration optimization, classical computing resources and quantum computing resources can work together to complete the execution of the QAOA algorithm and solve the combinatorial optimization problem, providing a more powerful solution for processing complex tasks.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A fusion computing system based on quantum measurement and control board, characterized in that: include: CPU, GPU and quantum measurement and control board; the CPU exchanges data with the GPU and the quantum measurement and control board through the PCIE bus; The quantum measurement and control board is connected to a quantum chip; The CPU is used to divide the predefined computing task into classical computing subtasks and quantum computing subtasks by using a preset resource scheduling algorithm; Compiling the quantum computing subtasks into quantum circuits; The quantum measurement and control board is used to receive the quantum circuit transmitted by the CPU, generate a pulse sequence corresponding to the quantum circuit, and send the pulse sequence to the quantum chip so that the quantum chip can execute the quantum circuit to complete the quantum computing subtask; wherein the pulse sequence is used to characterize the control signal and read signal of the quantum chip; The GPU is used to execute classical computing subtasks.

2. A fusion computing system based on quantum measurement and control board according to claim 1, characterized in that: The quantum measurement and control board adopts Xilinx RFSOC chip.

3. The fusion computing method of a fusion computing system based on a quantum measurement and control board as claimed in claim 1 or 2, characterized in that: The steps include: Step 1: The CPU uses a preset resource scheduling algorithm to divide the predefined computing task into classical computing subtasks and quantum computing subtasks; Step 2: The CPU compiles the quantum computing subtask into a quantum circuit and transmits the quantum circuit to the quantum measurement and control board; Step 3: The quantum measurement and control board receives the quantum circuit, generates a pulse sequence corresponding to the quantum circuit, and sends the pulse sequence to the quantum chip; wherein the pulse sequence is used to characterize the control signal and read signal of the quantum chip; Step 4: The quantum chip executes the quantum circuit according to the pulse sequence to complete the quantum computing subtask, and feeds back the task execution result to the quantum measurement and control board; Step 5: The quantum measurement and control board transmits the task execution result to the CPU; Step 6: GPU performs classic computing subtasks; Step 7: The CPU makes a decision on the quantum computing subtask and the classical computing subtask in real time according to the task execution result, and determines the next round of classical computing subtasks and quantum computing subtasks; Step 8: Repeat steps 2 to 7 until the stop condition is met.

4. The fusion calculation method according to claim 3, characterized in that: The predefined computing task includes solving the maximum cut problem using a quantum approximate optimization (QAOA) algorithm.

Images