The ground-state properties of atomic and molecular clusters can be obtained using Quantum Monte Carlo (QMC) simulations. We propose a reconfigurable hardware architecture using Field-Programmable Gate Arrays (FPGAs) to implement the kernels of the QMC application. To achieve higher clock rates, we experiment with different pipeline stages for each component of the design and develop a deeply pipelined architecture that provides the best performance in terms of clock rate, while at the same time has a modest use of embedded memory and multiplier resources so we can fit additional functions in a future implementation. Here, we discuss the details of the pipelined architecture and our design decisions while developing a general framework that can be used to obtain the potential energy of atomic or molecular clusters and extended to compute other useful properties.
The ground-state properties of atomic and molecular clusters can be obtained using Quantum Monte Carlo (QMC) simulations. We propose a reconfigurable hardware architecture using Field-Programmable Gate Arrays (FPGAs) to implement the kernels of the QMC application. To achieve higher clock rates, we experiment with different pipeline stages for each component of the design and develop a deeply pipelined architecture that provides the best performance in terms of clock rate, while at the same time has a modest use of embedded memory and multiplier resources so we can fit additional functions in a future implementation. Here, we discuss the details of the pipelined architecture and our design decisions while developing a general framework that can be used to obtain the potential energy of atomic or molecular clusters and extended to compute other useful properties.
Recent technological advances have led to a number of emerging platforms such as multi-cores, reconfigurable computing, and graphics processing units. We present a comparative study of multi-cores, field-programmable gate arrays, and graphics processing units for a Quantum Monte Carlo chemistry application. The speedups of these implementations are measured relative to a multi-core implementation and the accuracy of each implementation compared against the double-precision CPU implementation. The Brook+ AMD implementation shows the best overall speedup and accuracy, with the mixed-precision NVIDIA CUDA implementation outperforming the Brook+ implementation for larger atomic clusters with a slightly lower accuracy than the Brook+ and FPGA implementations. KeywordsQuantum Monte Carlo, Field-Programmable Gate Arrays, Multi-core processors, Graphics Processing Units.
Quantum Monte Carlo (QMC) methods deliver highly accurate electronic structure calculations but are computationally intensive. The quantum Monte Carlo kernel library (QMCkl) provides a modular, portable collection of high-performance kernels implementing the core building blocks of QMC calculations. It offers a C-compatible application programming interface, supports the TREXIO standard for input, and covers essential QMC kernels including atomic and molecular orbitals, cusp corrections, the Jastrow factor, and the necessary derivatives also to perform variational and structural optimization. QMCkl separates algorithmic development from hardware-specific tuning by combining human-readable reference implementations with performance-optimized kernels that produce identical numerical results. The library enables consistent, efficient, and reproducible simulations across different QMC codes and architectures and achieves substantial speedups in the evaluation of the energy and its derivatives. Beyond QMC, QMCkl can accelerate deterministic quantum chemistry workflows and visualization tools, promoting cross-code interoperability and simplifying high-performance scientific software development.
We present the first application of field programmable gate arrays (FPGAs) as new, customizable hardware architectures that can be harnessed for the fast and energy-efficient calculation of quantum dynamics simulations of large chemical/material systems. Instead of tailoring the software to fixed hardware (which is the typical case for writing quantum chemistry code for CPUs/GPUs), FPGAs allow us to directly customize the underlying hardware - even at the level of specific electrical signals in the circuit - to give a truly optimized computational performance for complex quantum dynamics calculations. By offloading the most intensive and repetitive calculations onto an FPGA, we show that the computational performance of our hardware implementation for real-time electron dynamics calculations can even exceed that of optimized commercial mathematical libraries running on high-performance GPUs. In addition to this impressive computational speedup, we show that FPGAs are immensely energy-efficient and consume 4 times less energy than modern GPU or CPU architectures. These energy savings are a practical and important metric for supercomputing centers (several of which exceed over $1 million in power costs alone), as exascale computing capabilities become more widespread and commonplace. Taken together, the implementation techniques and performance metrics of our study demonstrate that FPGAs could play a promising role in upcoming quantum chemistry and materials science applications, particularly for the acceleration and energy-efficient execution of quantum dynamics calculations.
Quantum Annealing (QA) is a classical probabilistic algorithm that provides a heuristic to find the globally optimal solution for a combinatorial optimization problem by using quantum tunneling processes. Quantum annealing simulation can be implemented on Field Programmable Gate Arrays (FPGAs) using Quantum Monte Carlo (QMC) simulation in the transverse Ising model. Since input data of the QMC simulation increases exponentially with the problem size, we have to use the DRAM of an FPGA board to store these data. However, storing data in DRAM causes two problems. One is the limited data access bandwidth, and the other is the limitation of DRAM capacity. We propose a data-transfer-bottleneck-less FPGA-based accelerator for quantum annealing simulation and apply it to implement number partitioning problem, which is one of the combinatorial optimization problems. The critical idea of our architecture is not storing but computing the large data in FPGA kernels and eliminating the burden on data transfer. We implement the proposed architecture on Stratix 10 FPGA and achieve up to 39.6 times speed-up compared to CPU-based quantum annealing simulation. We also achieve up to 2.8 times speed-up and implement 262,144 spins, which is 64 times increase compared to the most recent FPGA implementation.
Abstract Massively parallel architectures offer the potential to significantly accelerate an application relative to their serial counterparts. However, not all applications exhibit an adequate level of data and/or task parallelism to exploit such platforms. Furthermore, the power consumption associated with these forms of computation renders “scaling out” for exascale levels of performance incompatible with modern sustainable energy policies. In this work, we investigate the potential for field‐programmable gate arrays (FPGAs) to feature in future exascale platforms, and their capacity to improve performance per unit power measurements for the purposes of scientific computing. We have focused our efforts on variational Monte Carlo, and report on the benefits of coprocessing with a FPGA relative to a purely multicore system.
We present the first application of field programmable gate arrays (FPGAs) as new, customizable hardware architectures for carrying out fast and energy-efficient quantum dynamics simulations of large chemical/material systems. Instead of tailoring the software to fixed hardware, which is the typical case for writing quantum chemistry code for central processing units (CPUs) and graphics processing units (GPUs), FPGAs allow us to directly customize the underlying hardware (even at the level of specific electrical signals in the circuit) to give a truly optimized computational performance for quantum dynamics calculations. By offloading the most intensive and repetitive calculations onto an FPGA, we show that the computational performance of our real-time electron dynamics calculations can even exceed that of optimized commercial mathematical libraries running on high-performance GPUs. In addition to this impressive computational speedup, we show that FPGAs are immensely energy-efficient and consume 4 times less energy than modern GPU or CPU architectures. These energy savings are a practical and important metric for supercomputing centers (many of which exceed over $1 million in power costs alone), as exascale computing capabilities become more widespread and commonplace. Taken together, the implementation techniques and performance metrics of our study demonstrate that FPGAs could play a promising role in upcoming quantum chemistry and materials science applications, particularly for the acceleration and energy-efficient execution of quantum dynamics calculations.