High-Performance Ptychographic Reconstruction with Federated Facilities

  title={High-Performance Ptychographic Reconstruction with Federated Facilities},
  author={Tekin Bicer and Xiaodong Yu and Daniel J. Ching and Ryan Chard and Mathew J. Cherukara and Bogdan Nicolae and Rajkumar Kettimuthu and Ian T. Foster},
Beamlines at synchrotron light source facilities are powerful scientific instruments used to image samples and observe phenomena at high spatial and temporal resolutions. Typically, these facilities are equipped only with modest compute resources for the analysis of generated experimental datasets. However, high data rate experiments can easily generate data in volumes that take days (or even weeks) to process on those local resources. To address this challenge, we present a system that unifies… 
Linking Scientific Instruments and HPC: Patterns, Technologies, Experiences
Common patterns associated with online analysis methods are reviewed and methods for instantiating those patterns are described, which present experiences with the application of these methods to the processing of data from different scientific instruments, each of which engages HPC resources for data inversion, machine learning model training, or other purposes.


Rapid Tomographic Image Reconstruction via Large-Scale Parallelization
Methods that leverage highly parallel computers to improve the performance of iterative tomographic image reconstruction applications and implement the introduced methods in high-performance MapReduce-like computing middleware, which is further optimized for reconstruction operations.
Interactive Supercomputing for Experimental Data-Driven Workflows
A delegated batch reservation model is presented that allows scientists at the experimental facility to couple generation of data to the allocation of compute, data and network resources at the supercomputing centre, and is easily incorporated to the experimental facilities existing IT resource management and scheduling pipelines.
High-Performance Multi-Mode Ptychography Reconstruction on Distributed GPUs
This work reports an accelerated version of the multi-mode difference map algorithm for ptychography reconstruction using multiple distributed GPUs that leverages available scientific computing packages in Python, including mpi4py and PyCUDA, with the core computation functions implemented in CUDA C.
Real-Time Data Analysis and Autonomous Steering of Synchrotron Light Source Experiments
An autonomous stream processing system that allows data streamed from beamline computers to be processed in real time on a remote supercomputer, with a control feed-back loop used to make decisions during experimentation is proposed.
Optimization of tomographic reconstruction workflows on geographically distributed resources.
These performance models allow evaluation and estimation of the execution time of any given iterative tomographic reconstruction workflow that runs on geographically distributed resources and a workflow management system is built, which can automate the execution of workflows and minimize the user interaction with the underlying infrastructure.
Parallel ptychographic reconstruction.
A parallel method for real-time ptychographic phase retrieval that 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.
Scalable and accurate multi-GPU-based image reconstruction of large-scale ptychography data
An optimized intranode multi-GPU implementation that can efficiently solve large-scale ptychographic reconstruction problems and proposes a novel hybrid parallelization model to address the performance bottlenecks in the CG solver are provided.
SHARP: a distributed, GPU-based ptychographic solver
A set of algorithmic and computational methodologies used at the Advanced Light Source, and DOE light sources packaged as a CUDA based software environment named SHARP is introduced, aimed at providing state-of-the-art high-throughput ptychography reconstructions for the coming era of diffraction limited light sources.
Deep Learning Accelerated Light Source Experiments
This work describes here how a deep convolutional neural network can be integrated into the real-time streaming tomography pipeline to enable better-quality images in the early stages of data acquisition.
Petascale XCT: 3D Image Reconstruction with Hierarchical Communications on Multi-GPU Nodes
The design involves three novel optimizations: optimization of (back)projection operators by extending the 2D memory-centric approach to 3D to perform hierarchical communications by exploiting "fat-node" architecture with many GPUs and utilization of mixed-precision types while preserving convergence rate and quality.