Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Graphics Processing Units (GPUs)”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9
All sourcesNASA NTRSDOE OSTI
DOE OSTI2015

Latent uncertainties of the precalculated track Monte Carlo method

Purpose: While significant progress has been made in speeding up Monte Carlo (MC) dose calculation methods, they remain too time-consuming for the purpose of inverse planning. To achieve clinically usable calculation speeds, a precalculated Monte Carlo (PMC) algorithm for proton and electron transport was developed to run on graphics processing units (GPUs). The algorithm utilizes pregenerated particle track data from conventional MC codes for different materials such as water, bone, and lung to produce dose distributions in voxelized phantoms. While PMC methods have been described in the past, an explicit quantification of the latent uncertainty arising from the limited number of unique tracks in the pregenerated track bank is missing from the paper. With a proper uncertainty analysis, an optimal number of tracks in the pregenerated track bank can be selected for a desired dose calculation uncertainty. Methods: Particle tracks were pregenerated for electrons and protons using EGSnrc and GEANT4 and saved in a database. The PMC algorithm for track selection, rotation, and transport was implemented on the Compute Unified Device Architecture (CUDA) 4.0 programming framework. PMC dose distributions were calculated in a variety of media and compared to benchmark dose distributions simulated from the corresponding general-purpose MC codes in the same conditions. A latent uncertainty metric was defined and analysis was performed by varying the pregenerated track bank size and the number of simulated primary particle histories and comparing dose values to a “ground truth” benchmark dose distribution calculated to 0.04% average uncertainty in voxels with dose greater than 20% of D{sub max}. Efficiency metrics were calculated against benchmark MC codes on a single CPU core with no variance reduction. Results: Dose distributions generated using PMC and benchmark MC codes were compared and found to be within 2% of each other in voxels with dose values greater than 20% of the maximum dose. In proton calculations, a small (≤1 mm) distance-to-agreement error was observed at the Bragg peak. Latent uncertainty was characterized for electrons and found to follow a Poisson distribution with the number of unique tracks per energy. A track bank of 12 energies and 60000 unique tracks per pregenerated energy in water had a size of 2.4 GB and achieved a latent uncertainty of approximately 1% at an optimal efficiency gain over DOSXYZnrc. Larger track banks produced a lower latent uncertainty at the cost of increased memory consumption. Using an NVIDIA GTX 590, efficiency analysis showed a 807 × efficiency increase over DOSXYZnrc for 16 MeV electrons in water and 508 × for 16 MeV electrons in bone. Conclusions: The PMC method can calculate dose distributions for electrons and protons to a statistical uncertainty of 1% with a large efficiency gain over conventional MC codes. Before performing clinical dose calculations, models to calculate dose contributions from uncharged particles must be implemented. Following the successful implementation of these models, the PMC method will be evaluated as a candidate for inverse planning of modulated electron radiation therapy and scanned proton beams.

07 ISOTOPE AND RADIATION SOURCES↗
DOE OSTI2015

Multi-GPU implementation of a VMAT treatment plan optimization algorithm

Purpose: Volumetric modulated arc therapy (VMAT) optimization is a computationally challenging problem due to its large data size, high degrees of freedom, and many hardware constraints. High-performance graphics processing units (GPUs) have been used to speed up the computations. However, GPU’s relatively small memory size cannot handle cases with a large dose-deposition coefficient (DDC) matrix in cases of, e.g., those with a large target size, multiple targets, multiple arcs, and/or small beamlet size. The main purpose of this paper is to report an implementation of a column-generation-based VMAT algorithm, previously developed in the authors’ group, on a multi-GPU platform to solve the memory limitation problem. While the column-generation-based VMAT algorithm has been previously developed, the GPU implementation details have not been reported. Hence, another purpose is to present detailed techniques employed for GPU implementation. The authors also would like to utilize this particular problem as an example problem to study the feasibility of using a multi-GPU platform to solve large-scale problems in medical physics. Methods: The column-generation approach generates VMAT apertures sequentially by solving a pricing problem (PP) and a master problem (MP) iteratively. In the authors’ method, the sparse DDC matrix is first stored on a CPU in coordinate list format (COO). On the GPU side, this matrix is split into four submatrices according to beam angles, which are stored on four GPUs in compressed sparse row format. Computation of beamlet price, the first step in PP, is accomplished using multi-GPUs. A fast inter-GPU data transfer scheme is accomplished using peer-to-peer access. The remaining steps of PP and MP problems are implemented on CPU or a single GPU due to their modest problem scale and computational loads. Barzilai and Borwein algorithm with a subspace step scheme is adopted here to solve the MP problem. A head and neck (H and N) cancer case is then used to validate the authors’ method. The authors also compare their multi-GPU implementation with three different single GPU implementation strategies, i.e., truncating DDC matrix (S1), repeatedly transferring DDC matrix between CPU and GPU (S2), and porting computations involving DDC matrix to CPU (S3), in terms of both plan quality and computational efficiency. Two more H and N patient cases and three prostate cases are used to demonstrate the advantages of the authors’ method. Results: The authors’ multi-GPU implementation can finish the optimization process within ∼1 min for the H and N patient case. S1 leads to an inferior plan quality although its total time was 10 s shorter than the multi-GPU implementation due to the reduced matrix size. S2 and S3 yield the same plan quality as the multi-GPU implementation but take ∼4 and ∼6 min, respectively. High computational efficiency was consistently achieved for the other five patient cases tested, with VMAT plans of clinically acceptable quality obtained within 23–46 s. Conversely, to obtain clinically comparable or acceptable plans for all six of these VMAT cases that the authors have tested in this paper, the optimization time needed in a commercial TPS system on CPU was found to be in an order of several minutes. Conclusions: The results demonstrate that the multi-GPU implementation of the authors’ column-generation-based VMAT optimization can handle the large-scale VMAT optimization problem efficiently without sacrificing plan quality. The authors’ study may serve as an example to shed some light on other large-scale medical physics problems that require multi-GPU techniques.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

SU-D-206-02: Evaluation of Partial Storage of the System Matrix for Cone Beam Computed Tomography Using a GPU Platform

Purpose: Iterative reconstruction algorithms in computed tomography (CT) require a fast method for computing the intersections between the photons’ trajectories and the object, also called ray-tracing or system matrix computation. This work evaluates different ways to store the system matrix, aiming to reconstruct dense image grids in reasonable time. Methods: We propose an optimized implementation of the Siddon’s algorithm using graphics processing units (GPUs) with a novel data storage scheme. The algorithm computes a part of the system matrix on demand, typically, for one projection angle. The proposed method was enhanced with accelerating options: storage of larger subsets of the system matrix, systematic reuse of data via geometric symmetries, an arithmetic-rich parallel code and code configuration via machine learning. It was tested on geometries mimicking a cone beam CT acquisition of a human head. To realistically assess the execution time, the ray-tracing routines were integrated into a regularized Poisson-based reconstruction algorithm. The proposed scheme was also compared to a different approach, where the system matrix is fully pre-computed and loaded at reconstruction time. Results: Fast ray-tracing of realistic acquisition geometries, which often lack spatial symmetry properties, was enabled via the proposed method. Ray-tracing interleaved with projection and backprojection operations required significant additional time. In most cases, ray-tracing was shown to use about 66 % of the total reconstruction time. In absolute terms, tracing times varied from 3.6 s to 7.5 min, depending on the problem size. The presence of geometrical symmetries allowed for non-negligible ray-tracing and reconstruction time reduction. Arithmetic-rich parallel code and machine learning permitted a modest reconstruction time reduction, in the order of 1 %. Conclusion: Partial system matrix storage permitted the reconstruction of higher 3D image grid sizes and larger projection datasets at the cost of additional time, when compared to the fully pre-computed approach. This work was supported in part by the Fonds de recherche du Quebec - Nature et technologies (FRQ-NT). The authors acknowledge partial support by the CREATE Medical Physics Research Training Network grant of the Natural Sciences and Engineering Research Council of Canada (Grant No. 432290).

60 APPLIED LIFE SCIENCES↗
DOE OSTI2025

The critical importance of software for HEP

Particle physics has an ambitious and broad global experimental programme for the coming decades. Large investments in building new facilities are already underway or under consideration. Scaling the present processing power and data storage needs by the foreseen increase in data rates in the next decade for HL-LHC is not sustainable within the current budgets. As a result, a more efficient usage of computing resources is required in order to realise the physics potential of future experiments. Software and computing are an integral part of experimental design, trigger and data acquisition, simulation, reconstruction, and analysis, as well as related theoretical predictions. A significant investment in computing and software is therefore critical. Advances in software and computing, including artificial intelligence (AI) and machine learning (ML), will be key for solving these challenges. Making better use of new processing hardware such as graphical processing units (GPUs) or ARM chips is a growing trend. This forms part of a computing solution that makes efficient use of facilities and contributes to the reduction of the environmental footprint of HEP computing. The HEP community already provided a roadmap for software and computing for the last EPPSU, and this paper updates that, with a focus on the most resource critical parts of our data processing chain.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Full event interpretation with machine-learning-based particle-flow reconstruction in the CMS detector

The particle-flow (PF) algorithm constructs a global description of each particle collision by producing a comprehensive list of final-state particles, and is central to event reconstruction in the CMS experiment at the CERN LHC. The existing PF implementation relies on physics-motivated heuristics and assumptions that can be replaced by machine-learning (ML) models trained directly on simulated data and naturally suited to modern graphics processing units (GPUs). A state-of-the-art ML-based PF (MLPF) reconstruction algorithm, implemented within the CMS software framework, is presented. The MLPF algorithm performs a learnable full-event reconstruction on GPUs, generalizes across detector conditions and collision energies, and replaces multiple modular reconstruction steps with a single unified model. Physics performance comparable to standard PF reconstruction is achieved in both simulation and data, with improved jet energy resolution and inference time. In simulated top quark-antiquark events under LHC Run-3 (2023-2024) conditions, the jet energy resolution improves by 10-20% for jets with transverse momentum between 30-100 GeV. Inference time is evaluated using simulated multijet events, with a median of $20\,\hbox {ms}$ per event on an Nvidia L4 GPU, compared to approximately $110\,\hbox {ms}$ for the standard CMS PF reconstruction.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗
DOE OSTI2016

Understanding GPU Power. A Survey of Profiling, Modeling, and Simulation Methods

Modern graphics processing units (GPUs) have complex architectures that admit exceptional performance and energy efficiency for high throughput applications.Though GPUs consume large amounts of power, their use for high throughput applications facilitate state-of-the-art energy efficiency and performance. Consequently, continued development relies on understanding their power consumption. Our work is a survey of GPU power modeling and profiling methods with increased detail on noteworthy efforts. Moreover, as direct measurement of GPU power is necessary for model evaluation and parameter initiation, internal and external power sensors are discussed. Hardware counters, which are low-level tallies of hardware events, share strong correlation to power use and performance. Statistical correlation between power and performance counters has yielded worthwhile GPU power models, yet the complexity inherent to GPU architectures presents new hurdles for power modeling. Developments and challenges of counter-based GPU power modeling is discussed. Often building on the counter-based models, research efforts for GPU power simulation, which make power predictions from input code and hardware knowledge, provide opportunities for optimization in programming or architectural design. Noteworthy strides in power simulations for GPUs are included along with their performance or functional simulator counterparts when appropriate. Lastly, possible directions for future research are discussed.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2017

Efficient Simulation of Nested Hollow Sphere Intersections: for Dynamically Nested Compartmental Models in Cell Biology

In the particle-based simulation of cell-biological systems in continuous space, a key performance bottleneck is the computation of all possible intersections between particles. These typically rely for collision detection on solid sphere approaches. The behavior of cell biological systems is influenced by dynamic hierarchical nesting, such as the forming of, the transport within, and the merging of vesicles. Existing collision detection algorithms are found not to be designed for these types of spatial cell-biological models, because nearly all existing high performance parallel algorithms are focusing on solid sphere interactions. The known algorithms for solid sphere intersections return more intersections than actually occur with nested hollow spheres. Here we define a new problem of computing the intersections among arbitrarily nested hollow spheres of possibly different sizes, thicknesses, positions, and nesting levels. We describe a new algorithm designed to solve this nested hollow sphere intersection problem and implement it for parallel execution on graphical processing units (GPUs). We present first results about the runtime performance and scaling to hundreds of thousands of spheres, and compare the performance with that from a leading solid object intersection package also running on GPUs.

Koester, Till↗
DOE OSTI2023

Mixed-Precision S/DGEMM Using the TF32 and TF64 Frameworks on Low-Precision AI Tensor Cores

Using NVIDIA graphics processing units (GPUs) equipped with Tensor Cores has enabled the significant acceleration of general matrix multiplication (GEMM) for applications in machine learning (ML) and artificial intelligence (AI) and in high-performance computing (HPC) generally. The use of such power-efficient, specialized accelerators can provide a performance increase between 8 × and 20 ×, albeit with a loss in precision. However, a high level of precision is required in many large scientific and HPC applications, and computing in single or double precision is still necessary for many of these applications to maintain accuracy. Fortunately, mixed-precision methods can be employed to maintain a higher level of numerical precision while also taking advantage of the performance increases from computing with lower-precision AI cores. With this in mind, we extend the state of the art by using NVIDIA’s new TF32 framework. This new framework not only burdens some constraints of the previous frameworks, such as costly 32 16-bit castings but also provides an equivalent precision and performance by using a much simpler approach. We also propose a new framework called TF64 that attempts double-precision arithmetic with low-precision Tensor Cores. Although this framework does not exist yet, we validated the correctness of this idea and achieved an equivalent of 64-bit precision on 32-bit hardware.

Valero Lara, Pedro↗
DOE OSTI2023

Experiences Detecting Defective Hardware in Exascale Supercomputers

In May 2022, the newest supercomputer to top the TOP 500 list was Frontier at Oak Ridge National Laboratory, demonstrating the capability of computing more than 1.1 quintillion (1018) floating-point calculations every second. Driving this ground-breaking rate of computing is Frontier’s more than 37,000 graphics processing units (GPUs) and 9,408 central processing units (CPUs). In total, Frontier contains more than 60 million parts. At this scale, the smallest margin of error may generate hundreds of hardware errors across the system. These errors are capable of directly hindering world-class science performed on Frontier if not found. In this work, we describe and evaluate two strategies for finding hardware-level faults in Frontier’s 9,408 compute nodes. There are two strategies developed: the first uses the Slurm scheduler to scavenge available compute time to run the node screen, the second builds upon the lessons learned in the first strategy and enforces a weekly screen of each node. Using June 2023 as a case study, we find that the first scheduling strategy consumed more than ten times the resources as the second scheduling strategy, but successfully detected five hardware defects in Frontier. We summarize the lessons learned while developing and running a node screen on the world’s first exascale supercomputer.

Hagerty, Nick↗
DOE OSTI2025

A Study of Performance Portability of Low-bit Fused Matrix-Vector Multiplication Kernels in SYCL

Understanding the causes of performance gaps between a portable programming model and a vendor-specific programming model is important for improving performance portability. This paper studies performance portability of low-bit fused general matrix-vector multiplication kernels in SYCL on vendors’ graphics processing units (GPUs). This work introduces the use case, explains the kernel implementations in detail, evaluates the performance of the CUDA, HIP, and SYCL kernels on datacenter, desktop, and laptop GPUs, and investigates the causes of performance gaps. The results show that loop unrolling, kernel dispatch overhead, and sum reduction contribute to the gaps.

Jin, Zheming [ORNL] (ORCID:000000027197780X)↗
DOE OSTI2022

HyKKT

HyKKT (pronounced as "hiked") is a package for solving systems of linear equations of Karush-Kuhn-Tucker (KKT) form, which typically arise in optimization problems, such as optimal power flow analysis. HyKKT uses Cholesky instead of LDL^T factorization and solves the general KKT system to a desired numerical precision via block reduction and conjugate gradient on the Schur complement. Such implementation is more suitable for implementation on graphic processing units (GPUs).

Regev, Shaked↗
DOE OSTI2025

Metal additive manufacturing simulation across length, time, and computing scales

Metal additive manufacturing (AM) offers a unique opportunity for production of advanced materials and complex geometries. However, variability in microstructure and properties challenges conventional approaches to design, process optimization, qualification, and materials selection. Modeling and simulation can improve understanding of AM processing and materials, but also poses major challenges for existing computational methods. Simultaneously, modern scientific computing hardware has become increasingly complex, most notably with the adoption of hybrid architectures such as Graphical Processing Units (GPUs). If appropriately utilized, emerging computational capabilities provide an opportunity to reveal new insight into AM processing and the resulting material structure and properties. In this review we describe the computational AM landscape, identify critical gaps, and highlight opportunities to impact the development and application of AM. First, the requirements and challenges of representative AM problem statements will be defined. Here, these problems range from scientific studies to industrial applications and are designed to capture the breadth of challenges facing the AM community. Next, the current state of AM modeling and simulation is evaluated, broken down by enabling hardware and software, process simulation, microstructure simulation, and property simulation. Each section describes the diversity of simulation approaches and associated trade-offs in physical fidelity and computational expense. Each area is then assessed based on their suitability and readiness for current and developing computational architectures. Lastly, the greatest opportunities for future research and application are highlighted, including gaps in modeling capabilities, opportunities for near-term application, and key scientific challenges.

additive manufacturing↗
DOE OSTI2016

Designing a graphics processing unit accelerated petaflop capable lattice Boltzmann solver: Read aligned data layouts and asynchronous communication

The lattice Boltzmann method is a well-established numerical approach for complex fluid flow simulations. Recently, general-purpose graphics processing units (GPUs) have become available as high-performance computing resources at large scale. We report on designing and implementing a lattice Boltzmann solver for multi-GPU systems that achieves 1.79 PFLOPS performance on 16,384 GPUs. To achieve this performance, we introduce a GPU compatible version of the so-called bundle data layout and eliminate the halo sites in order to improve data access alignment. Furthermore, we make use of the possibility to overlap data transfer between the host central processing unit and the device GPU with computing on the GPU. As a benchmark case, we simulate flow in porous media and measure both strong and weak scaling performance with the emphasis being on large-scale simulations using realistic input data.

Computer Science↗
DOE OSTI2017

Acceleration of the IMplicit–EXplicit nonhydrostatic unified model of the atmosphere on manycore processors

We present the acceleration of an IMplicit-EXplicit (IMEX) nonhydrostatic atmospheric model on manycore processors such as graphic processing units (GPUs) and Intel's Many Integrated Core (MIC) architecture. IMEX time integration methods sidestep the constraint imposed by the Courant-Friedrichs-Lewy condition on explicit methods through corrective implicit solves within each time step. In this work, we implement and evaluate the performance of IMEX on manycore processors relative to explicit methods. Using 3D-IMEX at Courant number C = 15, we obtained a speedup of about 4x relative to an explicit time stepping method run with the maximum allowable C = 1. Moreover, the unconditional stability of IMEX with respect to the fast waves means the speedup can increase significantly with the Courant number as long as the accuracy of the resulting solution is acceptable. We show a speedup of 100x at C = 150 using 1D-IMEX to demonstrate this point. Several improvements on the IMEX procedure were necessary in order to outperform our results with explicit methods: (a) reducing the number of degrees of freedom of the IMEX formulation by forming the Schur complement, (b) formulating a horizontally explicit vertically implicit 1D-IMEX scheme that has a lower workload and better scalability than 3D-IMEX, (c) using high-order polynomial preconditioners to reduce the condition number of the resulting system, and (d) using a direct solver for the 1D-IMEX method by performing and storing LU factorizations once to obtain a constant cost for any Courant number. Without all of these improvements, explicit time integration methods turned out to be difficult to beat. We discuss in detail the IMEX infrastructure required for formulating and implementing efficient methods on manycore processors. Several parametric studies are conducted to demonstrate the gain from each of the abovementioned improvements. Lastly, we validate our results with standard benchmark problems in numerical weather prediction and evaluate the performance and scalability of the IMEX method using up to 4192 GPUs and 16 Knights Landing processors.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

A two-level GPU-accelerated incomplete LU preconditioner for general sparse linear systems

This paper presents a parallel preconditioning approach based on incomplete LU (ILU) factorizations in the framework of Domain Decomposition (DD) for general sparse linear systems. We focus on distributed memory parallel architectures, specifically, those that are equipped with graphic processing units (GPUs). In addition to block-Jacobi, we present general purpose two-level ILU Schur complement-based approaches, where different strategies are presented to solve the coarse-level reduced system. These strategies are combined with modified ILU methods in the construction of the coarse-level operator, in order to effectively remove smooth errors by targeting an algebraically smooth vector. We leverage available GPU-based sparse matrix kernels to accelerate the setup and the solve phases of the proposed ILU preconditioner. We evaluate the efficiency of the proposed methods as a smoother for algebraic multigrid (AMG) and as a preconditioner for Krylov subspace methods on challenging anisotropic diffusion problems and a collection of general sparse matrices.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

A GPU-based compressible combustion solver for applications exhibiting disparate space and time scales

High-speed chemically active flows pose significant computational challenges due to their disparate space and time scales, with stiff chemistry often dominating simulation time. While modern scientific computing programs achieve exascale performance by leveraging graphics processing units (GPUs), existing GPU-based compressible combustion solvers face critical limitations in memory management, load balancing, and handling the highly localized nature of chemical reactions. To this end, we present a high-performance compressible reacting flow solver built on the AMReX framework and optimized for multi-GPU settings. Here, our approach addresses three GPU performance bottlenecks: memory access patterns through column-major storage optimization, computational workload variability via a bulk-sparse integration strategy for chemical kinetics, and multi-GPU load distribution for adaptive mesh refinement applications. The solver adapts existing matrix-based chemical kinetics formulations to multi-grid contexts. Using representative combustion applications, including 2D and 3D detonations and a 3D jet-in-crossflow configuration, we demonstrate 1.4–5× performance improvements over initial implementations on an in-house cluster of NVIDIA H100 GPUs, and near-ideal weak scaling on the Frontier supercomputer (Oak Ridge Leadership Computing Facility) with up to 1024 AMD Instinct MI250X GPUs. Roofline analysis reveals substantial improvements in arithmetic intensity for both convection (∼ 10 ×) and chemistry (∼ 4 ×) routines, confirming efficient utilization of GPU memory bandwidth and computational resources.

42 ENGINEERING↗
DOE OSTI2018

Real-time data analysis for medical diagnosis using FPGA-accelerated neural networks

Real-time analysis of patient data during medical procedures can provide vital diagnostic feedback that significantly improves chances of success. With sensors becoming increasingly fast, frameworks such as Deep Neural Networks are required to perform calculations within the strict timing constraints for real-time operation. However, traditional computing platforms responsible for running these algorithms incur a large overhead due to communication protocols, memory accesses, and static (often generic) architectures. In this work, we implement a low-latency Multi-Layer Perceptron (MLP) processor using Field Programmable Gate Arrays (FPGAs). Unlike CPUs and Graphics Processing Units (GPUs), our FPGA-based design can directly interface sensors, storage devices, display devices and even actuators, thus reducing the delays of data movement between ports and compute pipelines. Moreover, the compute pipelines themselves are tailored specifically to the application, improving resource utilization and reducing idle cycles. We demonstrate the effectiveness of our approach using mass-spectrometry data sets for real-time cancer detection. We demonstrate that correct parameter sizing, based on the application, can reduce latency by 20% on average. Furthermore, we show that in an application with tightly coupled data-path and latency constraints, having a large amount of computing resources can actually reduce performance. Using mass-spectrometry benchmarks, we show that our proposed FPGA design outperforms both CPU and GPU implementations, with an average speedup of 144x and 21x, respectively. In our work, we demonstrate the importance of application-specific optimizations in order to minimize latency and maximize resource utilization for MLP inference. By directly interfacing and processing sensor data with ultra-low latency, FPGAs can perform real-time analysis during procedures and provide diagnostic feedback that can be critical to achieving higher percentages of successful patient outcomes.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2014

Parallel ptychographic reconstruction

Ptychography is an imaging method whereby a coherent beam is scanned across an object, and an image is obtained by iterative phasing of the set of diffraction patterns. It is able to be used to image extended objects at a resolution limited by scattering strength of the object and detector geometry, rather than at an optics-imposed limit. As technical advances allow larger fields to be imaged, computational challenges arise for reconstructing the correspondingly larger data volumes, yet at the same time there is also a need to deliver reconstructed images immediately so that one can evaluate the next steps to take in an experiment. Here we present a parallel method for real-time ptychographic phase retrieval. It uses a hybrid parallel strategy to divide the computation between multiple graphics processing units (GPUs) and then employs novel techniques to merge sub-datasets into a single complex phase and amplitude image. Results are shown on a simulated specimen and a real dataset from an X-ray experiment conducted at a synchrotron light source.

97 MATHEMATICS AND COMPUTING↗