Influence of data amount, data type and implementation packages in GPU coding

Xu, Peng; Sun, Mingyan; Gao, Yin-Jun; Du, Tai-Jiao; Hu, Jinming; Zhang, Jun-Jie

doi:10.1016/j.array.2022.100261

Cited by 3 publications

(4 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also, unless specified explicitly, the Numba and CuPy platforms assigned double-precision types by default for all floating-point variables and constants. In this regard, we explicitly specified a type for single-precision calculations [39,81].…”

Section: Methodsmentioning

confidence: 99%

“…In this section, we study the impact of precision by performing computations in single and double precision. For a similar investigation in the context of the finite-difference approach to Burgers' equation, see [81]. Extensive tests of Numba and CUDA C for matrixmatrix multiplication, parallel reduction, and 3D stencil application in single and double precisions can be found in [39].…”

Section: Impact Of Precisionmentioning

confidence: 99%

“…In this work, in the context of MC radiation transport, we assess the performance of two popular platforms that allow GPU computing from the Python programming language: Numba [14] and CuPy [16,17]. Both Numba and CuPy are examples of high-level popular platforms (e.g., [39,81]). Their main advantage is their simplicity.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Exploring Numba and CuPy for GPU-Accelerated Monte Carlo Radiation Transport

Askar,

Yergaliyev,

Shukirgaliyev

et al. 2024

Computation

View full text Add to dashboard Cite

This paper examines the performance of two popular GPU programming platforms, Numba and CuPy, for Monte Carlo radiation transport calculations. We conducted tests involving random number generation and one-dimensional Monte Carlo radiation transport in plane-parallel geometry on three GPU cards: NVIDIA Tesla A100, Tesla V100, and GeForce RTX3080. We compared Numba and CuPy to each other and our CUDA C implementation. The results show that CUDA C, as expected, has the fastest performance and highest energy efficiency, while Numba offers comparable performance when data movement is minimal. While CuPy offers ease of implementation, it performs slower for compute-heavy tasks.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Impact Of Precisionmentioning

confidence: 99%

See 1 more Smart Citation

Exploring Numba and CuPy for GPU-Accelerated Monte Carlo Radiation Transport

Askar,

Yergaliyev,

Shukirgaliyev

et al. 2024

Computation

View full text Add to dashboard Cite

show abstract

“…It is important to note that threads within the same block can share memory, enabling efficient data interchange. In the context of CUDA, the most common computation involves transferring data from the CPU to the GPU [19] The main steps of the CUDA program flow, as depicted in Fig. 4, are: the data is loaded into the host CPU memory and then transferred to the GPU memory using a function called "cudaMemcpy".…”

Section: B Compute Unified Device Architecturementioning

confidence: 99%

Enhancing Particle Swarm Optimization Performance Through CUDA and Tree Reduction Algorithm

Younis,

Eleyat

2024

IJACSA

View full text Add to dashboard Cite

In this paper, we present an enhancement for Particle Swarm Optimization performance by utilizing CUDA and a Tree Reduction Algorithm. PSO is a widely used metaheuristic algorithm that has been adapted into a CUDA version known as CPSO. The tree reduction algorithm is employed to efficiently compute the global best position. To evaluate our approach, we compared the speedup achieved by our CUDA version against the standard version of PSO, observing a maximum speedup of 37x. Additionally, we identified a linear relationship between the size of swarm particles and execution time; as the number of particles increases, so does computational loadhighlighting the efficiency of parallel implementations in reducing execution time. Our proposed parallel PSOs have demonstrated significant reductions in execution time along with improvements in convergence speed and local optimization performance -particularly beneficial for solving large-scale problems with high computational loads.

show abstract

Utilization of the RBG-Maxwell Framework for Collisionless Plasma at Atmospheric Scales

Sun,

Xu,

et al. 2024

IEEE Trans. Plasma Sci.

View full text Add to dashboard Cite

Influence of data amount, data type and implementation packages in GPU coding

Cited by 3 publications

References 20 publications

Exploring Numba and CuPy for GPU-Accelerated Monte Carlo Radiation Transport

Exploring Numba and CuPy for GPU-Accelerated Monte Carlo Radiation Transport

Enhancing Particle Swarm Optimization Performance Through CUDA and Tree Reduction Algorithm

Utilization of the RBG-Maxwell Framework for Collisionless Plasma at Atmospheric Scales

Contact Info

Product

Resources

About