Massively parallel simulations of relativistic fluid dynamics on graphics processing units with CUDA
Not provided.
Computer Science↗
SEARCH · Engineering Papers
Results for “CUDA”
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.
CUDA and MPI performances on the computation of Rossi-α distribution from pulsed neutron sources
Not Available
Talamo, Alberto↗
Experience of Migrating a Parallel Graph Coloring Program from CUDA to SYCL
We describe the experience of converting a CUDA implementation of a parallel graph coloring algorithm to SYCL. The goals are for our work to be useful to application and compiler developers by providing a detailed description of migration paths between CUDA and SYCL. We will describe how CUDA functions are mapped to SYCL functions. Evaluating the CUDA and SYCL implementations of the algorithm shows that the performance of SYCL and CUDA kernels are comparable over the test graph set on NVIDIA P100 and V100 GPUs. The SYCL program also allows for performance evaluation with the OpenCL and Level Zero interfaces and power profiling on an Intel GPU computing platform.
97 MATHEMATICS AND COMPUTING↗
Porting numerical integration codes from CUDA to oneAPI: a case study
We present our experience in porting optimized CUDA implementations to oneAPI. We focus on the use case of numerical integration, particularly the CUDA implementations of PAGANI and $m$-Cubes. We faced several challenges that caused performance degradation in the oneAPI ports. These include differences in utilized registers per thread, compiler optimizations, and mappings of CUDA library calls to oneAPI equivalents. After addressing those challenges, we tested both the PAGANI and m-Cubes integrators on numerous integrands of various characteristics. To evaluate the quality of the ports, we collected performance metrics of the CUDA and oneAPI implementations on the Nvidia V100 GPU. We found that the oneAPI ports often achieve comparable performance to the CUDA versions, and that they are at most 10% slower.
97 MATHEMATICS AND COMPUTING↗
A case study of CUDA FORTRAN and OpenACC for an atmospheric climate kernel
The porting of a key kernel in the tracer advection routines of the Community Atmosphere Model – Spectral Element (CAM-SE) to use Graphics Processing Units (GPUs) using OpenACC is considered in comparison to an existing CUDA FORTRAN port. The development of the OpenACC kernel for GPUs was substantially simpler than that of the CUDA port. Also, OpenACC performance was about 1.5× slower than the optimized CUDA version. Particular focus is given to compiler maturity regarding OpenACC implementation for modern FORTRAN, and it is found that the Cray implementation is currently more mature than the PGI implementation. Still, for the case that ran successfully on PGI, the PGI OpenACC runtime was slightly faster than Cray. The results show encouraging performance for OpenACC implementation compared to CUDA while also exposing some issues that may be necessary before the implementations are suitable for porting all of CAM-SE. Furthermore, most notable are that GPU shared memory should be used by future OpenACC implementations and that derived type support should be expanded.
54 ENVIRONMENTAL SCIENCES↗
Automatic Offloading C++ Expression Templates to CUDA Enabled GPUs
In the last few years, many scientific applications have been developed for powerful graphics processing units (GPUs) and have achieved remarkable speedups. This success can be partially attributed to high performance host callable GPU library routines that are offloaded to GPUs at runtime. These library routines are based on C/C++-like programming toolkits such as CUDA from NVIDIA and have the same calling signatures as their CPU counterparts. Recently, with the sufficient support of C++ templates from CUDA, the emergence of template libraries have enabled further advancement in code reusability and rapid software development for GPUs. However, Expression Templates (ET), which have been very popular for implementing data parallel scientific software for host CPUs because of their intuitive and mathematics-like syntax, have been underutilized by GPU development libraries. The lack of ET usage is caused by the difficulty of offloading expression templates from hosts to GPUs due to the inability to pass instantiated expressions to GPU kernels as well as the absence of the exact form of the expressions for the templates at the time of coding. This paper presents a general approach that enables automatic offloading of C++ expression templates to CUDA enabled GPUs by using the C++ metaprogramming technique and Just-In-Time (JIT) compilation methodology to generate and compile CUDA kernels for corresponding expression templates followed by executing the kernels with appropriate arguments. This approach allows developers to port applications to run on GPUs with virtually no code modifications. More specifically, this paper uses a large ET based data parallel physics library called QDP++ as an example to illustrate many aspects of the approach to offload expression templates automatically and to demonstrate very good speedups for typical QDP++ applications running on GPUs against running on CPUs using this method of offloading. In addition, this approach of automatic offlo- ding expression templates could be applied to other many-core accelerators that provide C++ programming toolkits with the support of C++ template.
97 MATHEMATICS AND COMPUTING↗
Evaluating CUDA Portability with HIPCL and DPCT
HIPCL is expanding the scope of the CUDA portability route from an AMD platform to an OpenCL platform. In the meantime, the Intel DPC++ Compatibility Tool (DPCT) is migrating a CUDA program to a data parallel C++ (DPC++) program. Towards the goal of portability enhancement, we evaluate the performance of the CUDA applications from Rodinia, SHOC, and proxy applications ported using HIPCL and DPCT on Intel GPUs. After profiling the ported programs, we aim to understand their performance gaps, and optimize codes converted by DPCT to improve their performance. The open-source repository for the CUDA, HIP, and DPCT programs will be useful for the development of a translator.
Jin, Zheming↗
Enhancing ChatPORT with CUDA-to-SYCL Kernel Translation Capability
Large Language Models (LLMs) have shown strong capabilities in general code translation. However, code translation involving parallel programming models remains largely unexplored. This work enhances the capabilities of code LLMs in CUDA-to-SYCL kernel translation with parameter-efficient fine-tuning. The resultant fine-tuned LLM, called ChatPORT, is an effort to provide high-fidelity translations from one programming model to another. We describe the preparation of datasets from heterogeneous computing benchmarks for model fine-tuning and testing, the parameter-efficient fine-tuning of 19 open-source code models ranging in size from 0.5 to 34 billion parameters and evaluate the correctness rates of the SYCL kernels by the fine-tuned models. The experimental results show that most code models fail to translate CUDA codes to SYCL correctly. However, fine-tuning these models using a small set of CUDA and SYCL kernels can enhance the capabilities of these models in kernel translation. Depending on the sizes of the models, the correctness rate ranges from 19.9% to 81.7% for a test dataset of 62 CUDA kernels.
Jin, Zheming [ORNL] (ORCID:000000027197780X)↗
Performance of CUDA Unified Memory in CMS Heterogeneous Pixel Reconstruction
The management of separate memory spaces of CPUs and GPUs brings an additional burden to the development of software for GPUs. To help with this, CUDA unified memory provides a single address space that can be accessed from both CPU and GPU. The automatic data transfer mechanism is based on page faults generated by the memory accesses. This mechanism has a performance cost, that can be with explicit memory prefetch requests. Various hints on the inteded usage of the memory regions can also be given to further improve the performance. The overall effect of unified memory compared to an explicit memory management can depend heavily on the application. In this paper we evaluate the performance impact of CUDA unified memory using the heterogeneous pixel reconstruction code from the CMS experiment as a realistic use case of a GPU-targeting HEP reconstruction software. We also compare the programming model using CUDA unified memory to the explicit management of separate CPU and GPU memory spaces.
Kortelainen, Matti J.↗
A Flexible CUDA LU-based Solver for Small, Batched Linear Systems
This chapter presents the implementation of a batched CUDA solver based on LU factorization for small linear systems. This solver may be used in applications such as reactive flow transport models, which apply the Newton-Raphson technique to linearize and iteratively solve the sets of non linear equations that represent the reactions for ten of thousands to millions of physical locations. The implementation exploits somewhat counterintuitive GPGPU programming techniques: it assigns the solution of a matrix (representing a system) to a single CUDA thread, does not exploit shared memory and employs dynamic memory allocation on the GPUs. These techniques enable our implementation to simultaneously solve sets of systems with over 100 equations and to employ LU decomposition with complete pivoting, providing the higher numerical accuracy required by certain applications. Other currently available solutions for batched linear solvers are limited by size and only support partial pivoting, although they may result faster in certain conditions. We discuss the code of our implementation and present a comparison with the other implementations, discussing the various tradeoffs in terms of performance and flexibility. This work will enable developers that need batched linear solvers to choose whichever implementation is more appropriate to the features and the requirements of their applications, and even to implement dynamic switching approaches that can choose the best implementation depending on the input data.
bached linear solvers↗
Using Hybrid Model OpenSHMEM + CUDA to Implement the SHOC Benchmark Suite
This work describes the process of porting the Scalable HeterOgeneous Computing (SHOC) benchmark suite from the hybrid MPI + CUDA implementation to OpenSHMEM + CUDA. SHOC includes a wide variety of benchmark kernels used to measure accelerator performance in both single node and cluster configurations. The hybrid model implementation attempts to place all major computation on accelerator devices, and uses MPI to synchronize and aggregate results. In some cases, MPI Groups are used to gradually reduce the number of accelerators used for computation as the problem size drops. Porting this behavior to OpenSHMEM required implementing several synchronizing collective operations, and using SHMEM teams to replace MPI Group functionality. Benchmark results on a Cray XK7 system with one GPU per compute node show that SHMEM performance is equal to MPI performance in these hybrid tasks. These results and porting experience show that using OpenSHMEM for accelerator devices benefits from adding functionality for synchronization and teams, and would further benefit from adding support for communication within accelerator kernels.
97 MATHEMATICS AND COMPUTING↗
Nekbone performance on GPUs with OpenACC and CUDA Fortran implementations
We present a hybrid GPU implementation and performance analysis of Nekbone, which represents one of the core kernels of the incompressible Navier–Stokes solver Nek5000. The implementation is based on OpenACC and CUDA Fortran for local parallelization of the compute-intensive matrix–matrix multiplication part, which significantly minimizes the modification of the existing CPU code while extending the simulation capability of the code to GPU architectures. Our discussion includes the GPU results of OpenACC interoperating with CUDA Fortran and the gather–scatter operations with GPUDirect communication. We demonstrate performance of up to 552 Tflops on 16, 384 GPUs of the OLCF Cray XK7 Titan.
Computer Science↗
OpenMP 4.5 and CUDA C++ and FORTRAN code samples
Brief Description for Public Release: A collection of general code samples demonstrating issues encountered with using OpenMP4.5 (and some CUDA) during efforts to port codes to a GPU platform using IBM XL. These independent examples include both C++ and FORTRAN language usage, along with attempts to use OpenMP and CUDA in an interoperable manner. This suite of code samples is meant to supplement the examples provided by openmp.org at https://github.com/OpenMP/Examples, especially in regards to FORTRAN language support and OpenMP/CUDA interoperability.
Black, AaronC.↗
CUDA Computation of the Feynman Distribution
In 2006, NVIDIA Corporation introduced the Computer Unified Device Architecture (CUDA) for C, C++, and Fortran programming languages. The CUDA parallel computing platform uses the Graphics Processing Unit (GPU) of NVIDIA graphics card, and it is designed for parallel computations. In these computations, the same algorithms are executed on the data in parallel. The GPU contains many Streaming Multiprocessors (SM) that can independently execute blocks of threads. A thread is a sequence of serial programmed instructions. The SM executes threads in groups of 32 and a group of 32 threads is called warp. The threads within a block can be synchronized and have access to a shared memory that is visible only to them. Conversely, blocks of threads cannot be synchronized and have access only to the global memory of the graphics card. This memory is called device memory and is independent of the host memory controlled by the Central Processing Unit (CPU). A single thread has access to a local memory (register) that is not visible to any other thread. The part of the program that is executed on the graphics card is called kernel. Each thread of the kernel has a unique identification index that is accessible by the in-built variable threadIdx; the latter is a 3-component vector. A threads block can contain up to 1024 threads. Blocks of threads have a unique identification index that is accessible by the in-built variable blockIdx; the latter is a 3-component vector. The in-built variable blockDim contains the number of threads per block; this variable is a 3-component vector. Threads and blocks represent different levels of parallelism; a kernel is executed as a grid of blocks of threads. Figure 1 illustrates a grid of thread blocks for a GPU with 2 SM. A SM can execute multiple blocks of threads in parallel; a block of threads is executed on a single SM. A GPU can execute multiple kernels concurrently. This work has used the Quadro M5000 graphics card with the following technical specifications: - 2048 cores with 861 MHz frequency, - 16 streaming multiprocessors, - 5.2 trillion transistors, - 8.2 Gb device memory, - 64 kb register memory (per thread), - 48 kb shared memory (per block) - 211.6 GB/s memory bandwidth, and - 3.5 GFLOPS floating point performance.
97 MATHEMATICS AND COMPUTING↗
GPU-acceleration of tensor renormalization with PyTorch using CUDA
We show that numerical computations based on tensor renormalization group (TRG) methods can be significantly accelerated with PyTorch on graphics processing units (GPUs) by leveraging NVIDIA's Compute Unified Device Architecture (CUDA). Here we find improvement in the runtime and its scaling with bond dimension for two-dimensional systems. Our results establish that the utilization of GPU resources is essential for future precision computations with TRG.
97 MATHEMATICS AND COMPUTING↗
Tausch: A halo exchange library for large heterogeneous computing systems using MPI, OpenCL, and CUDA
Exchanging halo data is a common task in modern scientific computing applications and efficient handling of this operation is critical for the performance of the overall simulation. Tausch is a novel header-only library that provides a simple API for efficiently handling these types of data movements. Tausch supports both simple CPU-only systems, but also more complex heterogeneous systems with both CPUs and GPUs. It currently supports both OpenCL and CUDA for communicating with GPGPU devices, and allows for communication between GPGPUs and CPUs. The API allows for drop-in replacement in existing codes and can be used for the communication layer in new codes. This paper provides an overview of the approach taken in Tausch, and a performance analysis that demonstrates expected and achieved performance. Here, we highlight the ease of use and performance with three applications: First Tausch is compared to the halo exchange framework from two Mantevo applications, HPCCG and miniFE, and then it is used to replace a legacy halo exchange library in the flexible multigrid solver framework Cedar.
97 MATHEMATICS AND COMPUTING↗
FOCUS: A full-orbit CUDA solver for particle simulations in magnetized plasmas
A Full-Orbit CUDA Solver for particle simulations in plasmas, FOCUS, has been developed. The code follows exact particle trajectories by solving Newton’s equations with the Lorentz force. The code can use fields calculated analytically or numerically and it is also possible to use the information provided by equilibrium reconstruction and transport codes. FOCUS has an elastic collisions module which covers the whole particle energy range in magnetic fusion devices. Moreover, an atomic collision module was also included to simulate the interaction of the test particles with neutral or partially ionized species. Regarding performance, the main feature is that the code runs on GPUs, allowing the simulation of a large number of particles using moderate computational resources. Different versions of the code have been used in several papers. Here we present a complete description of its capabilities, the basis physics included and detailed information about the numerical algorithms employed. Finally, some examples are presented.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
Accelerating solidification process simulation for large-sized system of liquid metal atoms using GPU with CUDA
Molecular dynamics simulation is a powerful tool to simulate and analyze complex physical processes and phenomena at atomic characteristic for predicting the natural time-evolution of a system of atoms. Precise simulation of physical processes has strong requirements both in the simulation size and computing timescale. Therefore, finding available computing resources is crucial to accelerate computation. However, a tremendous computational resource (GPGPU) are recently being utilized for general purpose computing due to its high performance of floating-point arithmetic operation, wide memory bandwidth and enhanced programmability. As for the most time-consuming component in MD simulation calculation during the case of studying liquid metal solidification processes, this paper presents a fine-grained spatial decomposition method to accelerate the computation of update of neighbor lists and interaction force calculation by take advantage of modern graphics processors units (GPU), enlarging the scale of the simulation system to a simulation system involving 10 000 000 atoms. In addition, a number of evaluations and tests, ranging from executions on different precision enabled-CUDA versions, over various types of GPU (NVIDIA 480GTX, 580GTX and M2050) to CPU clusters with different number of CPU cores are discussed. The experimental results demonstrate that GPU-based calculations are typically 9∼11 times faster than the corresponding sequential execution and approximately 1.5∼2 times faster than 16 CPU cores clusters implementations. On the basis of the simulated results, the comparisons between the theoretical results and the experimental ones are executed, and the good agreement between the two and more complete and larger cluster structures in the actual macroscopic materials are observed. Moreover, different nucleation and evolution mechanism of nano-clusters and nano-crystals formed in the processes of metal solidification is observed with large-sized system.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗