Experiences with OpenMP, PGI, HMPP and OpenACC Directives on ISO/TTI Kernels

Ghosh, Sayan; Liao, Terrence; Calandra, Henri; Chapman, Barbara

doi:10.1109/sc.companion.2012.95

Cited by 17 publications

(12 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…New hardware technologies (such as GPUs, FPGA, etc) have motivated researchers to further investigate how to re-purpose stencil algorithms to take advantage of their unique characteristics, following two major approaches: programming models and pure algorithmic optimization. On the programming models front, in 2012, Ghosh et al [17] analyzed the performance and programmability of three high-level directive-based GPU programming models (PGI, CAPS, and OpenACC) on an NVIDIA GPU on kernels for reverse time migration (RTM). In 2017, Qawasmeh et al [32] implemented an MPI + OpenACC approach for seismic modeling and RTM.…”

Section: Related Work 31 Stencil Computationmentioning

confidence: 99%

Porting and Evaluation of a Distributed Task-driven Stencil-based Application

Raut

Anderson

Araya‐Polo

et al. 2021

Proceedings of the 12th International Workshop on Programming Models and Applications for Multicores and Manycores

View full text Add to dashboard Cite

Alternative programming models and runtimes are increasing in popularity and maturity. This allows porting and comparing, on competitive grounds, emerging parallel approaches against the traditional MPI+X paradigm. In this work, an implementation of distributed task-based stencil computation is compared with a traditional MPI+X implementation of the same application. The Legion task-based parallel programming system is used as an alternative to MPI, but the underlying OpenMP approach is kept at the subdomain level. Overall results are promising toward making this alternative method competitive to the traditional MPI approach. In future work, extensions to other applications will be explored, as well as the use of GPUs. CCS CONCEPTS• Computing methodologies → Parallel programming languages.

show abstract

Section: Related Work 31 Stencil Computationmentioning

confidence: 99%

Porting and Evaluation of a Distributed Task-driven Stencil-based Application

Raut

Anderson

Araya‐Polo

et al. 2021

Proceedings of the 12th International Workshop on Programming Models and Applications for Multicores and Manycores

View full text Add to dashboard Cite

show abstract

“…Several attempts were previously conducted to port RTM and seismic modelling to GPUs. Ghosh et al (2012) had experiences in analysing the performance and programmability of three high-level directive-based GPU programming models - PGI, HMPP and OpenACC on a single Nvidia GPU against Isotropic (ISO) and Tilted Transversely Isotropic (TTI) finite difference kernels. This work only focused on the computational kernels of the modelling phase without taking care of the other phases of RTM.…”

Section: Related Workmentioning

confidence: 99%

Performance portability in reverse time migration and seismic modelling via OpenACC

Qawasmeh

Hugues²,

Calandra³

et al. 2017

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

Heterogeneity among the computational resources within a single machine has significantly increased in high performance computing to exploit the tremendous potential of graphics processing units (GPUs). Portability in terms of code development and performance has been a challenge due to major differences between GPU programming and memory models from one side and conventional central processing units (CPUs) from another side. Performance characteristics of compilers and processors also vary between machines. Emerging high-level directive-based programming models such as OpenACC has been proposed to target this challenge. In this work, we develop OpenACC implementations for both seismic modelling and reverse time migration algorithms that solve the isotropic, acoustic, and elastic wave equations. We employ OpenACC to take advantage of the computational power of two Nvidia GPU cards: (1) M2090 and (2) K40, residing in IBM and CRAY XC30 clusters respectively. We also explore the main aspects of hybridization seismic modelling and reverse time migration by implementing an Message Passing Interface (MPI)+OpenACC approach. We expose various mapping techniques to develop a portable code that maximizes performance regardless of compiler or platform. Depending on the intensity of the computations, different propagators exhibited different speedup behaviours against a full socket CPU MPI implementation. A performance enhancement of ~10× was obtained, when the acoustic model was ported to a single GPU, compared with a 1.7× speedup obtained using the isotropic model. Our MPI+OpenACC implementation of reverse time migration and seismic modelling shows promising scaling when multiple GPUs were used.

show abstract

“…Abdelkhalek et al [1] employed CUDA to solve the acoustic wave equation in RTM and seismic modeling on GPU. Ghosh et al [9] had experiences in analyzing the performance and programmability of three high-level directive-based GPU programming models -PGI, HMPP and OpenACC on an Nvidia GPU against Isotropic (ISO)/Tilted Transversely Isotropic (TTI) finite difference kernels. A prediction-based performance tuning mechanism was proposed in [13] to tune OpenACC gang and vector clauses.…”

Section: Related Workmentioning

confidence: 99%

GPU technology applied to reverse time migration and seismic modeling via OpenACC

Qawasmeh

Chapman

Hugues³

et al. 2015

Proceedings of the Sixth International Workshop on Programming Models and Applications for Multicores and Manycores

Self Cite

View full text Add to dashboard Cite

GPU computing offers tremendous potential to accelerate complex scientific applications and is becoming a leading force in speeding up seismic imaging and velocity analysis techniques. Developing portable code is a challenge that can be overcome using emerging high-level directive-based programming model such as OpenACC. In this paper, we develop OpenACC implementations for both seismic modeling and Reverse Time Migration (RTM) algorithms that solve the isotropic, acoustic, and elastic wave equations. We employ OpenACC to take advantage of the computational power of two Nvidia GPU cards: 1) M2090 and 2) K40, residing in IBM and CRAY XC30 clusters respectively. Although we implement a hybrid OpenACC-MPI approach to parallelize seismic modeling and RTM on multiple GPUs, in this paper, we focus on developing mapping techniques to exploit potentials of one GPU. We observe an incremental improvement in performance while exploring different optimization techniques. Adequate code restructuring to tap GPU's potential seems critical. Depending on the intensity of computations, different propagators exhibit different speedup behaviors. A performance enhancement of ∼ 10x was obtained, when the acoustic model was ported to a single GPU, compared with a 1.3x speedup obtained using the isotropic model.

show abstract

Experiences with OpenMP, PGI, HMPP and OpenACC Directives on ISO/TTI Kernels

Cited by 17 publications

References 4 publications

Porting and Evaluation of a Distributed Task-driven Stencil-based Application

Porting and Evaluation of a Distributed Task-driven Stencil-based Application

Performance portability in reverse time migration and seismic modelling via OpenACC

GPU technology applied to reverse time migration and seismic modeling via OpenACC

Contact Info

Product

Resources

About