Search NASA⌕ Search

SEARCH · Search NASA

Results for “data transfer”

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 919 records · Page 51

Evaluating Unified Memory Performance in HIP

Heterogeneous unified memory management between a CPU and a GPU is a major challenge in GPU computing. Recently, unified memory (UM) has been supported by software and hardware components on AMD computing platforms. The support could simplify the complexities of memory management. In this paper, we attempt to have a better understanding of UM by evaluating the performance of UM programs on an AMD MI100 GPU. More specifically, we evaluate data migration using UM against other data transfer techniques for the overall performance of an application, assess the impacts of three commonly used optimization techniques on the kernel execution time of a vector add sample, and compare the performance and productivity of selected benchmarks with and without UM. The performance overhead associated with UM is not trivial, but it can improve programming productivity by reducing lines of code for scientific applications. We aim to present early results and feedback on the UM performance to the vendor.

Jin, Zheming↗

Asynchronous in situ connected-components analysis for complex fluid flows

Fluid flow in porous media is at the heart of problems such as groundwater contamination and carbon sequestration, and presents an important challenge for scientific computing. For this class of problem, large three-dimensional simulations are performed to advance scientific inquiry. On massively parallel computing systems, the volume of data generated by such approaches can become a productivity bottleneck if the raw data generated from the simulation is analyzed in a post-processing step. We present a physics-based framework for in situ data reduction that is theoretically grounded in multiscale averaging theory. We show how task parallelism can be exploited to concurrently perform a variety of analysis tasks with data-dependent costs, including the generation of iso-surfaces, morphological analyses, and connected components analysis. A task management framework is constructed to launch asynchronous analysis threads, manage dependencies between different tasks, promote data locality and hide the impact of data transfers. The framework is applied to analyze GPU-based simulations of two-fluid flow in porous media, generating a set of averaged measures that represents the overall system behavior. We demonstrate how the approach can be applied to perform physically-consistent averaging over fluid subregions using connected components analysis. Simulations performed on Oak Ridge National Lab s Titan supercomputer are profiled to demonstrate the performance of the associated multi-threaded in situ analysis for typical production simulation of two-fluid flow.

Mcclure, James↗

Asynchronous In Situ Connected-Components Analysis for Complex Fluid flows, In: ISAV '16 Proceedings of the 2nd Workshop on In Situ Infrastructures for Enabling Extreme-scale Analysis and Visualization

The simulation of multiscale physics is an important challenge for scientific computing. For this class of problem, large three-dimensional simulations are performed to advance scientific inquiry. On massively parallel computing systems, the volume of data generated by such approaches can become a productivity bottleneck if the raw data generated from the simulation is analyzed in a post-processing step. To address this, we present a physics-based framework for in situ data reduction that is theoretically grounded in multiscale averaging theory. We show how task parallelism can be exploited to concurrently perform a variety of analysis tasks with data-dependent costs, including the generation of iso-surfaces, morphological analyses, and connected components analysis. All analyses are performed in parallel using distributed memory and use the same domain decomposition as the simulation. A task management framework is constructed to leverage available parallelism within a node for analysis. The capabilities of the framework are to launch asynchronous analysis threads, manage dependencies between different tasks, promote data locality and minimize the impact of data transfers. The framework is applied to analyze GPU-based simulations of two-fluid-phase flow in porous media, generating a set of averaged measures that represents the overall system behavior. We demonstrate how the approach can be applied to perform physically-consistent analysis over fluid sub-regions determined from connected components analysis. Simulations performed on Oak Ridge National Lab's Titan supercomputer are profiled to demonstrate the performance of the associated multi-threaded in situ analysis approach for typical production simulation of two-fluid-phase flow.

Computer Science↗

Remote Hardware-in-the-Loop Approach for Microgrid Controller Evaluation

Utilities have been installing microgrids because of the increased resilience and reliability advantages they may provide to the distribution system. A microgrid controller is a critical component in microgrids. It is of great benefit to derisk the installation of microgrid controllers before field deployment. Hardware-in-the-loop (HIL) testing is used by controller developers and utilities to evaluate the controllers under stressful conditions. In this work, a microgrid control function developed by the Synchrophasor Grid Monitoring and Automation (SyGMA) laboratory at the University of California, San Diego is tested in a remote HIL (RHIL) setup. The digital real-time simulation of the detailed microgrid system was operated at the National Renewable Energy Laboratory's Energy Systems Integration Facility. Under such RHIL setup, successful controller operation is contingent on understanding and characterizing the communications channel and in particular network latencies. The novelty of this paper is the proposed use of a RHIL setup that leverages existing power system communications protocols to evaluate the controller in conjunction with the simulation capabilities of a remote facility. The work presented here will provide the complete setup of the HIL evaluation platform, the details of the communications protocols used by the setup for data transfer between the two organizations, test cases developed to evaluate the controller, and the results from the experiments.

controller hardware-in-the-loop↗

Improving Signal-to-Noise Ratio (SNR) for Readout Signals Using Adaptive Filters on Reconfigurable Controls Hardware

This study investigates the optimization of Signal-to-Noise Ratio (SNR) in superconducting quantum computing readout signals through adaptive filtering. Quantum computing technology has the potential to revolutionize various fields by delivering exponential speedup in solving certain computational problems. However, the technology's practical implementation is hindered by the difficulty of extracting clean, reliable signals during the readout phase, with various sources of noise presenting a significant barrier to clean signals. This noise, often present in readout profiles due to imperfect isolation, degrades the system's overall SNR, thus impeding the ability to extract the quantum state accurately. The research leverages the power of adaptive filtering to improve the SNR of quantum computing readout signals. Specifically, an adaptive filter is implemented in a PYNQ overlay on an FPGA, and eventually will be connected to a quantum computing system. The system models the noise with a Least Mean Squares (LMS) adaptive filter, and then subtracts the estimated noise from the received signal to improve the SNR. A Direct Memory Access (DMA) channel is used to handle the signal processing, delivering efficient, high-speed data transfer between the PYNQ system and the hardware. The study explores the benefits of this adaptive filtering technique, potentially providing a significant contribution to practical and fast quantum computing.

Johnson, Hans↗

Efficient Implementation of Many-body Quantum Chemical Methods on the Intel Xeon Phi Coprocessor

This paper presents the implementation and performance of the highly accurate CCSD(T) quantum chemistry method on the Intel Xeon Phi coprocessor within the context of the NWChem computational chemistry package. The widespread use of highly correlated methods in electronic structure calculations is contingent upon the interplay between advances in theory and the possibility of utilizing the ever-growing computer power of emerging heterogeneous architectures. We discuss the design decisions of our implementation as well as the optimizations applied to the compute kernels and data transfers between host and coprocessor. We show the feasibility of adopting the Intel Many Integrated Core Architecture and the Intel Xeon Phi coprocessor for developing efficient computational chemistry modeling tools. Remarkable scalability is demonstrated by benchmarks. Our solution scales up to a total of 62560 cores with the concurrent utilization of Intel Xeon processors and Intel Xeon Phi coprocessors.

Environmental Molecular Sciences Laboratory↗

funcX: Federated Function as a Service for Science

Here, funcX is a distributed function as a service (FaaS) platform that enables flexible, scalable, and high performance remote function execution. Unlike centralized FaaS systems, funcX decouples the cloud-hosted management functionality from the edge-hosted execution functionality. funcX's endpoint software can be deployed, by users or administrators, on arbitrary laptops, clouds, clusters, and supercomputers, in effect turning them into function serving systems. funcX's cloud-hosted service provides a single location for registering, sharing, and managing both functions and endpoints. It allows for transparent, secure, and reliable function execution across the federated ecosystem of endpoints-enabling users to route functions to endpoints based on specific needs. funcX uses containers (e.g., Docker, Singularity, and Shifter) to provide common execution environments across endpoints. funcX implements various container management strategies to execute functions with high performance and efficiency on diverse funcX endpoints. funcX also integrates with an in-memory data store and Globus for managing data that may span endpoints. We motivate the need for funcX, present our prototype design and implementation, and demonstrate, via experiments on two supercomputers, that funcX can scale to more than 130000 concurrent workers. We show that funcX's container warming-aware routing algorithm can reduce the completion time for 3,000 functions by up to 61% compared to a randomized algorithm and the in-memory data store can speed up data transfers by up to 3x compared to a shared file system.

97 MATHEMATICS AND COMPUTING↗

COUPLING OF RELAP5-SCDAP MOD4.0 AND NEUTRONIC CODES

High-fidelity and accurate nuclear system codes play a key role in the design and analysis of complex nuclear power plants, which consist of multiple subsystems, such as the reactor core (and its fuel, burnable poisons, control elements, etc.), the reactor internal structures, the vessel, and the energy conversion subsystem and beyond to grid demand. Most commonly the interplay between these various subsystems is modeled using coupled codes, each of which represents one of the subsystems. And the most common direct coupling is that of thermal-hydraulics and neutronics codes. The subject of this paper is the coupling of codes that model not only thermal-hydraulics and neutronics, but also structural components damage. Furthermore, the neutronic component is not limited to the sole core solver. The coupled code system encompasses thermal-hydraulics, material performance of the fuel, neutronic solver, and neutronic data preparation. Thus, this paper presents a framework for coupling RELAP5/SCDAPSIM/MOD4.0 with a suite of neutron kinetics codes that includes NESTLE, DRAGON and a version of the ENDF library. The version of the RELAP5/SCDAPSIM/MOD4.0 code used in this work is one developed by Innovate System Software (ISS) as part of the international SCDAP Development and Training Program (SDTP) for best-estimate analysis to model reactor transients including severe accident phenomena. This RELAP5/SCDAPSIM/MOD4.0 code version is also capable of predicting nuclear fuel performance. It uses nodal power distributions to calculate mechanical and thermal parameters such as heat-up, oxidation and meltdown of fuel rods and control rods, the ballooning and rupture of fuel rod cladding, the release of fission products from fuel rods, and the disintegration of fuel rods into porous debris and molten material. On the neutronics side, this work uses the NESTLE and DRAGON codes. NESTLE is a multi-dimensional static and kinetic neutronic code developed at North Carolina State University. It solves up to four energy groups neutron diffusion equations utilizing the Nodal Expansion Method (NEM) in Cartesian or hexagonal geometry. The DRAGON code, developed at Ecole Polytechnique de Montreal, performs lattice physics calculations based on the neutron transport equation and is capable of using very fine energy group structures. In this work, we have developed a coupling approach to exchange data among the various modules. In the coupling process, the generated nuclear data (in fine multigroup energy structure) are collapsed down into two- or four-group energy structures for use in NESTLE. The neutron kinetics and thermal-hydraulics modules are coupled at each time step by using the cross-section data. The power distribution results of the neutronic calculations are transmitted to the thermal-hydraulics code. The spatial distribution of coolant density and the fuel-moderator temperature, which result from the thermal-hydraulic calculations, are transmitted back to the neutron kinetics codes and then the loop is closed using new neutronics results. Details of the actual data transfers will be described in the full length paper.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Machine-learning-enabled on-the-fly analysis of RHEED patterns during thin film deposition by molecular beam epitaxy

Thin film deposition is a fundamental technology for the discovery, optimization, and manufacturing of functional materials. Deposition by molecular beam epitaxy (MBE) typically employs reflection high-energy electron diffraction (RHEED) as a real-time in situ probe of the growing film. However, the state-of-the-art for RHEED analysis during deposition requires human observation. Here, we present an approach using machine learning (ML) methods to monitor, analyze, and interpret RHEED images on-the-fly during thin film deposition. In the analysis workflow, RHEED pattern images are collected at one frame per second and featurized using a pretrained deep convolutional neural network. The feature vectors are then statistically analyzed to identify changepoints; these changepoints can be related to changes in the deposition mode from initial film nucleation to a transition regime, smooth film deposition, and in some cases, an additional transition to a rough, islanded deposition regime. The feature vectors are additionally analyzed via graph analysis and community classification. The graph is quantified as a stabilization plot, and we show that inflection points in the stabilization plot correspond to changes in the growth regime. The full RHEED analysis workflow is termed RHAAPsody and includes data transfer and output to a visual dashboard. We demonstrate the functionality of RHAAPsody by analyzing the precaptured RHEED images from epitaxial depositions of anatase TiO2 on SrTiO3(001) and show that the analysis workflow can be executed in less than 1 s. Our approach shows promise as one component of ML-enabled real-time feedback control of the MBE deposition process.

36 MATERIALS SCIENCE↗

Modeling and Analysis of Hybrid Pixel Detector Deficiencies for Scientific Applications

Semiconductor hybrid pixel detectors often consist of a pixellated sensor layer bump bonded to a matching pixelated readout integrated circuit (ROIC). The sensor can range from high resistivity Si to III-V materials, whereas a Si CMOS process is typically used to manufacture the ROIC. Independent, device physics and electronic design automation (EDA) tools are used to determine sensor characteristics and verify functional performance of ROICs respectively with significantly different solvers. Some physics solvers provide the capability of transferring data to the EDA tool. However, single pixel transient simulations are either not feasible due to convergence difficulties or are prohibitively long. A simplified sensor model, which includes a current pulse in parallel with detector equivalent capacitor, is often used; even then, spice type top-level (entire array) simulations range from days to weeks. In order to analyze detector deficiencies for a particular scientific application, accurately defined transient behavioral models of all the functional blocks are required. Furthermore, various simulations, such as transient, noise, Monte Carlo, inter-pixel effects, etc. of the entire array need to be performed within a reasonable time frame without trading off accuracy. The sensor and the analog front-end can be modeling using a real number modeling language, as complex mathematical functions or detailed data can be saved to text files, for further top-level digital simulations. Parasitically aware digital timing is extracted in a standard delay format (sdf) from the pixel digital back-end layout as well as the periphery of the ROIC. For any given input, detector level worst-case and best-case simulations are performed using a Verilog simulation environment to determine the output. Each top-level transient simulation takes no more than 10-15 minutes. The impact of changing key parameters such as sensor Poissonian shot noise, analog front-end bandwidth, jitter due to clock distribution etc. can be accurately analyzed to determine ROIC architectural viability and bottlenecks. Hence the impact of the detector parameters on the scientific application can be studied.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

The sensitivity of patient specific IMRT QC to systematic MLC leaf bank offset errors

Purpose: Patient specific IMRT QC is performed routinely in many clinics as a safeguard against errors and inaccuracies which may be introduced during the complex planning, data transfer, and delivery phases of this type of treatment. The purpose of this work is to evaluate the feasibility of detecting systematic errors in MLC leaf bank position with patient specific checks. Methods: 9 head and neck (H and N) and 14 prostate IMRT beams were delivered using MLC files containing systematic offsets ({+-}1 mm in two banks, {+-}0.5 mm in two banks, and 1 mm in one bank of leaves). The beams were measured using both MAPCHECK (Sun Nuclear Corp., Melbourne, FL) and the aS1000 electronic portal imaging device (Varian Medical Systems, Palo Alto, CA). Comparisons with calculated fields, without offsets, were made using commonly adopted criteria including absolute dose (AD) difference, relative dose difference, distance to agreement (DTA), and the gamma index. Results: The criteria most sensitive to systematic leaf bank offsets were the 3% AD, 3 mm DTA for MAPCHECK and the gamma index with 2% AD and 2 mm DTA for the EPID. The criterion based on the relative dose measurements was the least sensitive to MLC offsets. More highly modulated fields, i.e., H and N, showed greater changes in the percentage of passing points due to systematic MLC inaccuracy than prostate fields. Conclusions: None of the techniques or criteria tested is sufficiently sensitive, with the population of IMRT fields, to detect a systematic MLC offset at a clinically significant level on an individual field. Patient specific QC cannot, therefore, substitute for routine QC of the MLC itself.

60 APPLIED LIFE SCIENCES↗

In vivo dosimetry in external beam radiotherapy

In vivo dosimetry (IVD) is in use in external beam radiotherapy (EBRT) to detect major errors, to assess clinically relevant differences between planned and delivered dose, to record dose received by individual patients, and to fulfill legal requirements. After discussing briefly the main characteristics of the most commonly applied IVD systems, the clinical experience of IVD during EBRT will be summarized. Advancement of the traditional aspects of in vivo dosimetry as well as the development of currently available and newly emerging noninterventional technologies are required for large-scale implementation of IVD in EBRT. These new technologies include the development of electronic portal imaging devices for 2D and 3D patient dosimetry during advanced treatment techniques, such as IMRT and VMAT, and the use of IVD in proton and ion radiotherapy by measuring the decay of radiation-induced radionuclides. In the final analysis, we will show in this Vision 20/20 paper that in addition to regulatory compliance and reimbursement issues, the rationale for in vivo measurements is to provide an accurate and independent verification of the overall treatment procedure. It will enable the identification of potential errors in dose calculation, data transfer, dose delivery, patient setup, and changes in patient anatomy. It is the authors' opinion that all treatments with curative intent should be verified through in vivo dose measurements in combination with pretreatment checks.

61 RADIATION PROTECTION AND DOSIMETRY↗

SU-E-T-395: Multi-GPU-Based VMAT Treatment Plan Optimization Using a Column-Generation Approach

Purpose: GPU has been employed to speed up VMAT optimizations from hours to minutes. However, its limited memory capacity makes it difficult to handle cases with a huge dose-deposition-coefficient (DDC) matrix, e.g. those with a large target size, multiple arcs, small beam angle intervals and/or small beamlet size. We propose multi-GPU-based VMAT optimization to solve this memory issue to make GPU-based VMAT more practical for clinical use. Methods: Our column-generation-based method generates apertures sequentially by iteratively searching for an optimal feasible aperture (referred as pricing problem, PP) and optimizing aperture intensities (referred as master problem, MP). The PP requires access to the large DDC matrix, which is implemented on a multi-GPU system. Each GPU stores a DDC sub-matrix corresponding to one fraction of beam angles and is only responsible for calculation related to those angles. Broadcast and parallel reduction schemes are adopted for inter-GPU data transfer. MP is a relatively small-scale problem and is implemented on one GPU. One headand- neck cancer case was used for test. Three different strategies for VMAT optimization on single GPU were also implemented for comparison: (S1) truncating DDC matrix to ignore its small value entries for optimization; (S2) transferring DDC matrix part by part to GPU during optimizations whenever needed; (S3) moving DDC matrix related calculation onto CPU. Results: Our multi-GPU-based implementation reaches a good plan within 1 minute. Although S1 was 10 seconds faster than our method, the obtained plan quality is worse. Both S2 and S3 handle the full DDC matrix and hence yield the same plan as in our method. However, the computation time is longer, namely 4 minutes and 30 minutes, respectively. Conclusion: Our multi-GPU-based VMAT optimization can effectively solve the limited memory issue with good plan quality and high efficiency, making GPUbased ultra-fast VMAT planning practical for real clinical use.

60 APPLIED LIFE SCIENCES↗

MO-A-BRD-10: A Fast and Accurate GPU-Based Proton Transport Monte Carlo Simulation for Validating Proton Therapy Treatment Plans

Purpose: To build a GPU-based Monte Carlo (MC) simulation of proton transport with detailed modeling of elastic and non-elastic (NE) protonnucleus interactions, for use in a very fast and cost-effective proton therapy treatment plan verification system. Methods: Using the CUDA framework, we implemented kernels for the following tasks: (1) Simulation of beam spots from our possible scanning nozzle configurations, (2) Proton propagation through CT geometry, taking into account nuclear elastic and multiple scattering, as well as energy straggling, (3) Bertini-style modeling of the intranuclear cascade stage of NE interactions, and (4) Simulation of nuclear evaporation. To validate our MC, we performed: (1) Secondary particle yield calculations in NE collisions with therapeutically-relevant nuclei, (2) Pencil-beam dose calculations in homogeneous phantoms, (3) A large number of treatment plan dose recalculations, and compared with Geant4.9.6p2/TOPAS. A workflow was devised for calculating plans from a commercially available treatment planning system, with scripts for reading DICOM files and generating inputs for our MC. Results: Yields, energy and angular distributions of secondaries from NE collisions on various nuclei are in good agreement with the Geant4.9.6p2 Bertini and Binary cascade models. The 3D-gamma pass rate at 2%–2mm for 70–230 MeV pencil-beam dose distributions in water, soft tissue, bone and Ti phantoms is 100%. The pass rate at 2%–2mm for treatment plan calculations is typically above 98%. The net computational time on a NVIDIA GTX680 card, including all CPU-GPU data transfers, is around 20s for 1×10{sup 7} proton histories. Conclusion: Our GPU-based proton transport MC is the first of its kind to include a detailed nuclear model to handle NE interactions on any nucleus. Dosimetric calculations demonstrate very good agreement with Geant4.9.6p2/TOPAS. Our MC is being integrated into a framework to perform fast routine clinical QA of pencil-beam treatment plans. Hardware for such a system will cost under $5000. This work was funded in part by a grant from Varian Medical Systems, Inc.

61 RADIATION PROTECTION AND DOSIMETRY↗

Quality assurance for online adapted treatment plans: Benchmarking and delivery monitoring simulation

Purpose: An important challenge facing online adaptive radiation therapy is the development of feasible and efficient quality assurance (QA). This project aimed to validate the deliverability of online adapted plans and develop a proof-of-concept online delivery monitoring system for online adaptive radiation therapy QA. Methods: The first part of this project benchmarked automatically online adapted prostate treatment plans using traditional portal dosimetry IMRT QA. The portal dosimetry QA results of online adapted plans were compared to original (unadapted) plans as well as randomly selected prostate IMRT plans from our clinic. In the second part, an online delivery monitoring system was designed and validated via a simulated treatment with intentional multileaf collimator (MLC) errors. This system was based on inputs from the dynamic machine information (DMI), which continuously reports actual MLC positions and machine monitor units (MUs) at intervals of 50 ms or less during delivery. Based on the DMI, the system performed two levels of monitoring/verification during the delivery: (1) dynamic monitoring of cumulative fluence errors resulting from leaf position deviations and visualization using fluence error maps (FEMs); and (2) verification of MLC positions against the treatment plan for potential errors in MLC motion and data transfer at each control point. Validation of the online delivery monitoring system was performed by introducing intentional systematic MLC errors (ranging from 0.5 to 2 mm) to the DMI files for both leaf banks. These DMI files were analyzed by the proposed system to evaluate the system’s performance in quantifying errors and revealing the source of errors, as well as to understand patterns in the FEMs. In addition, FEMs from 210 actual prostate IMRT beams were analyzed using the proposed system to further validate its ability to catch and identify errors, as well as establish error magnitude baselines for prostate IMRT delivery. Results: Online adapted plans were found to have similar delivery accuracy in comparison to clinical IMRT plans when validated with portal dosimetry IMRT QA. FEMs for the simulated deliveries with intentional MLC errors exhibited distinct patterns for different MLC error magnitudes and directions, indicating that the proposed delivery monitoring system is highly specific in detecting the source of errors. Implementing the proposed QA system for online adapted plans revealed excellent delivery accuracy: over 99% of leaf position differences were within 0.5 mm, and >99% of pixels in the FEMs had fluence errors within 0.5 MU. Patterns present in the FEMs and MLC control point analysis for actual patient cases agreed with the error pattern analysis results, further validating the system’s ability to reveal and differentiate MLC deviations. Calculation of the fluence map based on the DMI was performed within 2 ms after receiving each DMI input. Conclusions: The proposed online delivery monitoring system requires minimal additional resources and time commitment to the current clinical workflow while still maintaining high sensitivity to leaf position errors and specificity to error types. The presented online delivery monitoring system therefore represents a promising QA system candidate for online adaptive radiation therapy.

07 ISOTOPE AND RADIATION SOURCES↗

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↗

MO-B-BRD-02: 3D Printing in the Clinic

This session is designed so that the learning objectives are practical. The intent is that the attendee may take home an understanding of not just the technology, but also the logistical steps necessary to execute these 3D printing techniques in the clinic. Four practical 3D printing topics will be discussed: (i) Creating bolus and compensators for photon machines; (ii) tools for proton therapy; (iii) clinical applications in imaging; (iv) custom phantom design for clinic and research use. The use of 3D printers within the radiation oncology setting is proving to be a useful tool for creating patient specific bolus and compensators with the added benefit of cost savings. Creating the proper protocol is essential to ensuring that the desired effect is achieved and modeled in the treatment planning system. The critical choice of printer material (since it determines the interaction with the radiation) will be discussed. Selection of 3D printer type, design methods, verification of dose calculation, and the printing process will be detailed to give the basis for establishing your own protocol for electron and photon fields. A practical discussion of likely obstacles that may be encountered will be included. The diversity of systems and techniques in proton facilities leads to different facilities having very different requirements for beam modifying hardware and quality assurance devices. Many departments find the need to design and fabricate facility-specific equipment, making 3D printing an attractive technology. 3D printer applications in proton therapy will be discussed, including beam filters and compensators, and the design of proton therapy specific quality assurance tools. Quality control specific to 3D printing in proton therapy will be addressed. Advantages and disadvantages of different printing technology for these applications will also be discussed. 3D printing applications using high-resolution radiology-based imaging data will be presented. This data is used to 3D print individualized physical models of patient’s unique anatomy for aid in planning complex and challenging surgical procedures. Methods, techniques and imaging requirements for 3D printing anatomic models from imaging data will be discussed. Specific applications currently being used in the radiology clinic will be detailed. Standardized phantoms for radiation therapy are abundant. However, custom phantom designs can be advantageous for both clinical tasks and research. 3D printing is a useful method of custom fabrication that allows one to construct custom objects relatively quickly. Possibilities for custom radiotherapy phantoms range from 3D printing a hollow shell and filling the shell with tissue equivalent materials to fully printing the entire phantom with materials that are tissue equivalent as well as suitable for 3D printing. A range of materials available for use in radiotherapy phantoms and in the case of phantoms for dosimetric measurements, this choice is critical. The necessary steps required will be discussed including: modalities of 3D model generation, 3D model requirements for 3D printing, generation of machine instructions from the 3D model, and 3D printing techniques, choice of phantoms material, and troubleshooting techniques for each step in the process. Case examples of 3D printed phantoms will be shown. Learning Objectives: Understand the types of 3D modeling software required to design your device, the file formats required for data transfer from design software to 3D printer, and general troubleshooting techniques for each step of the process. Learn the differences between materials and design for photons vs. electrons vs. protons. Understand the importance of material choice and design geometries for your custom phantoms. Learn specific steps of quality assurance and quality control for 3D printed beam filters and compensators for proton therapy. Learn of special 3D printing applications for imaging. Cunha: Research support from Phillips Healthcare.

60 APPLIED LIFE SCIENCES↗

MO-B-BRD-03: Principles, Pitfalls and Techniques of 3D Printing for Bolus and Compensators

This session is designed so that the learning objectives are practical. The intent is that the attendee may take home an understanding of not just the technology, but also the logistical steps necessary to execute these 3D printing techniques in the clinic. Four practical 3D printing topics will be discussed: (i) Creating bolus and compensators for photon machines; (ii) tools for proton therapy; (iii) clinical applications in imaging; (iv) custom phantom design for clinic and research use. The use of 3D printers within the radiation oncology setting is proving to be a useful tool for creating patient specific bolus and compensators with the added benefit of cost savings. Creating the proper protocol is essential to ensuring that the desired effect is achieved and modeled in the treatment planning system. The critical choice of printer material (since it determines the interaction with the radiation) will be discussed. Selection of 3D printer type, design methods, verification of dose calculation, and the printing process will be detailed to give the basis for establishing your own protocol for electron and photon fields. A practical discussion of likely obstacles that may be encountered will be included. The diversity of systems and techniques in proton facilities leads to different facilities having very different requirements for beam modifying hardware and quality assurance devices. Many departments find the need to design and fabricate facility-specific equipment, making 3D printing an attractive technology. 3D printer applications in proton therapy will be discussed, including beam filters and compensators, and the design of proton therapy specific quality assurance tools. Quality control specific to 3D printing in proton therapy will be addressed. Advantages and disadvantages of different printing technology for these applications will also be discussed. 3D printing applications using high-resolution radiology-based imaging data will be presented. This data is used to 3D print individualized physical models of patient’s unique anatomy for aid in planning complex and challenging surgical procedures. Methods, techniques and imaging requirements for 3D printing anatomic models from imaging data will be discussed. Specific applications currently being used in the radiology clinic will be detailed. Standardized phantoms for radiation therapy are abundant. However, custom phantom designs can be advantageous for both clinical tasks and research. 3D printing is a useful method of custom fabrication that allows one to construct custom objects relatively quickly. Possibilities for custom radiotherapy phantoms range from 3D printing a hollow shell and filling the shell with tissue equivalent materials to fully printing the entire phantom with materials that are tissue equivalent as well as suitable for 3D printing. A range of materials available for use in radiotherapy phantoms and in the case of phantoms for dosimetric measurements, this choice is critical. The necessary steps required will be discussed including: modalities of 3D model generation, 3D model requirements for 3D printing, generation of machine instructions from the 3D model, and 3D printing techniques, choice of phantoms material, and troubleshooting techniques for each step in the process. Case examples of 3D printed phantoms will be shown. Learning Objectives: Understand the types of 3D modeling software required to design your device, the file formats required for data transfer from design software to 3D printer, and general troubleshooting techniques for each step of the process. Learn the differences between materials and design for photons vs. electrons vs. protons. Understand the importance of material choice and design geometries for your custom phantoms. Learn specific steps of quality assurance and quality control for 3D printed beam filters and compensators for proton therapy. Learn of special 3D printing applications for imaging. Cunha: Research support from Phillips Healthcare.

60 APPLIED LIFE SCIENCES↗