Graphics Processing Unit Version Research Articles

Adiabatic/Nonadiabatic State-to-State Reactive Scattering Dynamics Implemented on Graphics Processing Units

An efficient graphics processing units (GPUs) version of time-dependent wavepacket code is developed for the atom-diatom state-to-state reactive scattering processes. The propagation of the wavepacket is entirely calculated on GPUs employing the split-operator method after preparation of the initial wavepacket on the central processing unit (CPU). An additional split-operator method is introduced in the rotational part of the Hamiltonian to decrease communication of GPUs without losing accuracy of state-to-state information. The code is tested to calculate the differential cross sections of H + H2 reaction and state-resolved reaction probabilities of nonadiabatic triplet-singlet transitions of O((3)P,(1)D) + H2 for the total angular momentum J = 0. The global speedups of 22.11, 38.80, and 44.80 are found comparing the parallel computation of one GPU, two GPUs by exact rotational operator, and two GPU versions by an approximate rotational operator with serial computation of the CPU, respectively.

Accelerating VASP electronic structure calculations using graphic processing units

We present a way to improve the performance of the electronic structure Vienna Ab initio Simulation Package (VASP) program. We show that high-performance computers equipped with graphics processing units (GPUs) as accelerators may reduce drastically the computation time when offloading these sections to the graphic chips. The procedure consists of (i) profiling the performance of the code to isolate the time-consuming parts, (ii) rewriting these so that the algorithms become better-suited for the chosen graphic accelerator, and (iii) optimizing memory traffic between the host computer and the GPU accelerator. We chose to accelerate VASP with NVIDIA GPU using CUDA. We compare the GPU and original versions of VASP by evaluating the Davidson and RMM-DIIS algorithms on chemical systems of up to 1100 atoms. In these tests, the total time is reduced by a factor between 3 and 8 when running on n (CPU core + GPU) compared to n CPU cores only, without any accuracy loss.

Tunable, mixed-resolution modeling using library-based Monte Carlo and graphics processing units.

Building on our recently introduced library-based Monte Carlo (LBMC) approach, we describe a flexible protocol for mixed coarse-grained (CG)/all-atom (AA) simulation of proteins and ligands. In the present implementation of LBMC, protein side chain configurations are pre-calculated and stored in libraries, while bonded interactions along the backbone are treated explicitly. Because the AA side chain coordinates are maintained at minimal run-time cost, arbitrary sites and interaction terms can be turned on to create mixed-resolution models. For example, an AA region of interest such as a binding site can be coupled to a CG model for the rest of the protein. We have additionally developed a hybrid implementation of the generalized Born/surface area (GBSA) implicit solvent model suitable for mixed-resolution models, which in turn was ported to a graphics processing unit (GPU) for faster calculation. The new software was applied to study two systems: (i) the behavior of spin labels on the B1 domain of protein G (GB1) and (ii) docking of randomly initialized estradiol configurations to the ligand binding domain of the estrogen receptor (ERα). The performance of the GPU version of the code was also benchmarked in a number of additional systems.

A comparison of GPU strategies for unstructured mesh physics

SUMMARYThere have been few efforts to date to write physics algorithms for general unstructured meshes (meshes composed of arbitrary polygons/polyhedra) on graphics processing units (GPUs). Typical strategies for GPU memory management, such as double‐buffering and coalescing memory accesses, are difficult to apply to the irregular memory storage patterns of unstructured meshes. This paper presents results from an initial GPU version of a typical unstructured mesh kernel. Three different memory management strategies are described and implemented. Timing results for all three strategies are presented, in some cases showing speedups of over 20 times compared with the original CPU code.Copyright © 2012 John Wiley & Sons, Ltd.

SU-E-T-493: Accelerated Monte Carlo Methods for Photon Dosimetry Using a Dual-GPU System and CUDA.

To develop a Graphics Processing Unit (GPU) based Monte Carlo (MC) code that accelerates dose calculations on a dual-GPU system. We simulated a clinical case of prostate cancer treatment. A voxelized abdomen phantom derived from 120 CT slices was used containing 218×126×60 voxels, and a GE LightSpeed 16-MDCT scanner was modeled. A CPU version of the MC code was first developed in C++ and tested on Intel Xeon X5660 2.8GHz CPU, then it was translated into GPU version using CUDA C 4.1 and run on a dual Tesla m2 090 GPU system. The code was featured with automatic assignment of simulation task to multiple GPUs, as well as accurate calculation of energy- and material- dependent cross-sections. Double-precision floating point format was used for accuracy. Doses to the rectum, prostate, bladder and femoral heads were calculated. When running on a single GPU, the MC GPU code was found to be ×19 times faster than the CPU code and ×42 times faster than MCNPX. These speedup factors were doubled on the dual-GPU system. The dose Result was benchmarked against MCNPX and a maximum difference of 1% was observed when the relative error is kept below 0.1%. A GPU-based MC code was developed for dose calculations using detailed patient and CT scanner models. Efficiency and accuracy were both guaranteed in this code. Scalability of the code was confirmed on the dual-GPU system.

TH-F-211-04: A Fast Finite Size Pencil Beam Algorithm for Dose Calculation Using GPUs

Purpose: To facilitate real-time dose calculation for online adaptive radiation therapy, we have implemented the Monte Carlo-based finite size pencil beam (FSPB) dose calculation algorithm on Graphics Processing Units (GPUs). Methods: The FSPB dose calculation algorithm is based on the analytical kernel calculated from Monte Carlo simulations using the EGSnrc Monte Carlo code. This provides the benefit of Monte Carlo accuracy while the GPU implementation and the exponential kernel significantly reduces execution time. The GPU-based FSPB implementation was executed on the NVidia GTX480 GPU card. Both CPU and GPU versions of the FSPB algorithm were implemented operationally identically and benchmarked for accuracy and execution time. The dose calculation was performed in heterogeneous phantoms consisting of water, lung and bone slabs, and in clinical cases. The GPU-based FSPB implementation was compared with the CPU version and with a commercial treatment planning system for different situations including IMRT plans. Results: The FSPB algorithm with heterogeneity correction and anisotropic analytical algorithm (AAA) in heterogeneous phantom and clinical treatment planning cases agree within 2%. The GPU implementation was found to be up to 477 times faster than the single threaded CPU while achieving identical result in terms of dosimetric accuracy. The GPU-based FSPB was close to six times faster than the AAA, with single beam calculations completed within 0.45 s for typical scenarios. Conclusions: This work describes the implementation of a GPU-based FSPB algorithm for dose calculation in radiation therapy. The results demonstrated that the GPU-based FSPB algorithm can provide accurate 3D dose distributions in heterogeneous media in less than a second per beam for clinically relevant cases. This research is supported by CPRIT Individual Investigator Award RP110329.

Optimization schemes and performance evaluation of Smith–Waterman algorithm on CPU, GPU and FPGA

SUMMARYWith fierce competition between CPU and graphics processing unit (GPU) platforms, performance evaluation has become the focus of various sectors. In this paper, we take a well‐known algorithm in the field of biosequence matching and database searching, the Smith–Waterman (S‐W) algorithm as an example, and demonstrate approaches that fully exploit its performance potentials on CPU, GPU, and field‐programmable gate array (FPGA) computing platforms. For CPU platforms, we perform two optimizations, single instruction, multiple data and multithread, with compiler options, to gain over 70 × speedups over naive CPU versions on quad‐core CPU platforms. For GPU platforms, we propose the combination of coalesced global memory accesses, shared memory tiles, and loop unfolding, achieving 50 × speedups over initial GPU versions on an NVIDIA GeForce GTX 470 card. Experimental results show that the GPU GTX 470 gains 12 × speedups, instead of 100 × reported by some studies, over Intel quadcore CPU Q9400, under the same manufacturing technology and both with fully optimized schemes. In addition, for FPGA platforms, we customize a linear systolic array for the S‐W algorithm in a 45‐nm FPGA chip from Xilinx (XC6VLX760), with up to 1024 processing elements. Under only 133 MHz clock rate, the FPGA platform reaches the highest performance and becomes the most power‐efficient platform, using only 25 W compared with 190 W of the GPU GTX 470. Copyright © 2011 John Wiley & Sons, Ltd.

GPU-Based Fast Iterative Reconstruction of Fully 3-D PET Sinograms

This work presents a graphics processing unit (GPU)-based implementation of a fully 3-D PET iterative reconstruction code, FIRST (Fast Iterative Reconstruction Software for [PET] Tomography), which was developed by our group. We describe the main steps followed to convert the FIRST code (which can run on several CPUs using the message passing interface [MPI] protocol) into a code where the main time-consuming parts of the reconstruction process (forward and backward projection) are massively parallelized on a GPU. Our objective was to obtain significant acceleration of the reconstruction without compromising the image quality or the flexibility of the CPU implementation. Therefore, we implemented a GPU version using an abstraction layer for the GPU, namely, CUDA C. The code reconstructs images from sinogram data, and with the same System Response Matrix obtained from Monte Carlo simulations than the CPU version. The use of memory was optimized to ensure good performance in the GPU. The code was adapted for the VrPET small-animal PET scanner. The CUDA version is more than 70 times faster than the original code running in a single core of a high-end CPU, with no loss of accuracy.

연속 영상 기반 실시간 객체 분할

This paper shows an approach for real-time object segmentation on GPU (Graphics Processing Unit) using CUDA (Compute Unified Device Architecture). Recently, many applications that is monitoring system, motion analysis, object tracking or etc require real-time processing. It is not suitable for object segmentation to procedure real-time in CPU. NVIDIA provide CUDA platform for Parallel Processing for General Computation to upgrade limit of Hardware Graphic. In this paper, we use adaptive Gaussian Mixture Background Modeling in the step of object extraction and CCL(Connected Component Labeling) for classification. The speed of GPU and CPU is compared and evaluated with implementation in Core2 Quad processor with 2.4GHz.The GPU version achieved a speedup of 3x-4x over the CPU version.

Importance of explicit vectorization for CPU and GPU software performance

Much of the current focus in high-performance computing is on multi-threading, multi-computing, and graphics processing unit (GPU) computing. However, vectorization and non-parallel optimization techniques, which can often be employed additionally, are less frequently discussed. In this paper, we present an analysis of several optimizations done on both central processing unit (CPU) and GPU implementations of a particular computationally intensive Metropolis Monte Carlo algorithm. Explicit vectorization on the CPU and the equivalent, explicit memory coalescing, on the GPU are found to be critical to achieving good performance of this algorithm in both environments. The fully-optimized CPU version achieves a 9× to 12× speedup over the original CPU version, in addition to speedup from multi-threading. This is 2× faster than the fully-optimized GPU version, indicating the importance of optimizing CPU implementations.

Application of graphics processing unit (GPU) to software in elementary particle/high energy physics field

Abstract We apply graphics processing unit (GPU) to software in elementary particle and high energy physics field. We develop GPU versions of programs for fast calculations of cross sections of Standard Model (SM) processes at the Large Hadron Colliders (LHC) and those for numerical integration of multi-dimensional differential cross sections of these processes via the Monte Carlo method. We summarize our activity on the development of programs which run on GPU and the results on their performances. Compared with the CPU programs, we obtain around 50 times on average better performance on GPU. We expect that application of GPU to more general programs which are used in jobs on GRID will greatly reduce the computing resources necessary for large scale reconstructions and simulations for the LHC analysis.

On developing B-spline registration algorithms for multi-core processors

Spline-based deformable registration methods are quite popular within the medical-imaging community due to their flexibility and robustness. However, they require a large amount of computing time to obtain adequate results. This paper makes two contributions towards accelerating B-spline-based registration. First, we propose a grid-alignment scheme and associated data structures that greatly reduce the complexity of the registration algorithm. Based on this grid-alignment scheme, we then develop highly data parallel designs for B-spline registration within the stream-processing model, suitable for implementation on multi-core processors such as graphics processing units (GPUs). Particular attention is focused on an optimal method for performing analytic gradient computations in a data parallel fashion. CPU and GPU versions are validated for execution time and registration quality. Performance results on large images show that our GPU algorithm achieves a speedup of 15 times over the single-threaded CPU implementation whereas our multi-core CPU algorithm achieves a speedup of 8 times over the single-threaded implementation. The CPU and GPU versions achieve near-identical registration quality in terms of RMS differences between the generated vector fields.

Accelerating Smith-Waterman Algorithm for Biological Database Search on CUDA-Compatible GPUs

This paper presents a fast method capable of accelerating the Smith-Waterman algorithm for biological database search on a cluster of graphics processing units (GPUs). Our method is implemented using compute unified device architecture (CUDA), which is available on the nVIDIA GPU. As compared with previous methods, our method has four major contributions. (1) The method efficiently uses on-chip shared memory to reduce the data amount being transferred between off-chip video memory and processing elements in the GPU. (2) It also reduces the number of data fetches by applying a data reuse technique to query and database sequences. (3) A pipelined method is also implemented to overlap GPU execution with database access. (4) Finally, a master/worker paradigm is employed to accelerate hundreds of database searches on a cluster system. In experiments, the peak performance on a GeForce GTX 280 card reaches 8.32 giga cell updates per second (GCUPS). We also find that our method reduces the amount of data fetches to 1/140, achieving approximately three times higher performance than a previous CUDA-based method. Our 32-node cluster version is approximately 28 times faster than a single GPU version. Furthermore, the effective performance reaches 75.6 giga instructions per second (GIPS) using 32 GeForce 8800 GTX cards.

SU-FF-T-622: Fast GPU-Based Raytracing Dose Calculations for Brachytherapy in Heterogeneous Media

Purpose: To develop a fast dose calculation algorithm for permanent implant brachytherapy in heterogeneous media based on raytracing. Method and Materials: We have implemented a modified version of the TG‐43 dosimetry protocol based on an incremental version of Siddon's raytracing algorithm that accounts for tissue composition and interseed attenuation. Raytracing along straight lines from sources to dose calculation points is used to evaluate the length traveled in each voxel, for which the density and composition are known. The water‐equivalent distance corresponding to this radiological length is then used to compute the dose with the TG‐43 formalism. The raytracing being numerically intensive, we have implemented the algorithm on a Graphics Processing Unit (GPU) in order to exploit its parallel processing features and therefore reduce the overall dose computation time. For this purpose we used a NVIDIA 8800 GT GPU and the CUDA programming language. The algorithm was tested with a simple voxelized phantom and the results were compared to regular TG‐43 dose calculations. Results: Tissue composition and interseed attenuation were shown to impact significantly the dose calculation in a simple scenario. The GPU version of the modified TG‐43 algorithm was up to 34× faster than its Central Processing Unit (CPU) version. The accuracy of the results was the same on the GPU as on the CPU. Conclusion:Brachytherapydose calculations can potentially be more accurate by accounting for heterogeneities with raytracing‐based algorithms. These algorithms are numerically intensive but can exploit the parallel architecture of stream processors such as GPUs for acceleration. The GPU implementation presented here allows for execution times that are more acceptable for a clinical use. Future work will include the evaluation of the accuracy of the algorithm with Monte Carlo simulations.Conflict of Interest: Research sponsored by Varian Medical Systems Inc.

Stream processors: a new platform for Monte Carlo calculations

Graphics processing units (GPUs) and similar stream processors are increasingly used for general-purpose calculations. Their pipelined architecture can be exploited to accelerate various algorithms, sometimes with spectacular results. Monte Carlo codes, being computationally intensive, are likely to benefit from the development of stream processing platforms. We explore this potential here with a simple subroutine sometimes used in Monte Carlo techniques. More specifically, a ray tracing algorithm that computes the exact radiological path in a voxel grid was implemented in CPU and GPU versions, which then were compared in terms of execution speed. This benchmarking experiment was conducted under various conditions, in order to assess the memory and bandwidth limitations of each platform. The results show that the GPU provides a significant speed improvement factor over the CPU. For the specific hardware used in this work, namely a nVidia 7600 GS GPU, a speed increase factor up to 6 was achieved over an Xeon 2.4 GHz CPU. With the development of faster stream processors, this factor is expected to reach levels that can potentially change how Monte Carlo techniques are used, for example in radiation therapy planning. The ongoing development of simpler language extensions and programming interfaces also promises to increase the accessibility of these devices. Overall, stream processors are likely to play an increasingly larger role in scientific computing, and in particular in Monte Carlo techniques.

UAV Swarm Control: Calculating Digital Pheromone Fields with the GPU

Our future military force will be complex: a highly integrated mix of manned and unmanned units. These unmanned units could function individually or within a swarm. The readiness of future warfighters to work alongside and utilize these new forces depends on the creation of usable interfaces and training simulators. The difficulty is that current unmanned aerial vehicle (UAV) control interfaces require too much operator attention, and common swarm control methods require expensive computational power. This paper begins with a discussion on how to improve upon current user interfaces and then reviews a swarm control method, the digital pheromone field. This method uses digital pheromones to bias the movements of individual units within a swarm toward areas that are attractive and away from areas that are dangerous or unattractive. Next, a more efficient method for performing pheromone field calculations is introduced, one that harnesses the power of the graphics processing unit (GPU) in today's graphics cards by reshaping the ADAPTIV swarm control algorithm into a form acceptable to the GPU's pipeline [1]. The GPU ADAPTIV implementation is tested in scenarios that involve up to 50,000 virtual UAVs. When compared to its counterpart CPU implementation, the GPU version performed over 30 times faster than the CPU version. This gain translates directly into lower costs for training the future warfighter today and fielding the swarms of tomorrow. Finally, this paper presents a vision of how to combine these new interface ideas and performance enhancements into an effective swarm control interface and training simulator.