Global Memory Access Modelling for Efficient Implementation of the LBM on GPUs

Abstract : In this work, we investigate the global memory access mechanism on recent GPUs. For the purpose of this study, we created specific benchmark programs, which allowed us to explore the scheduling of global memory transactions. Thus, we formulate a model capable of estimating the execution time for a large class of applications. Our main goal is to facilitate optimisation of regular data-parallel applications on GPUs. As an example, we finally describe our CUDA implementations of LBM flow solvers on which our model was able to estimate performance with less than 5% relative error.
Complete list of metadatas

Cited literature [12 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01003059
Contributor : Laboratoire Cethil <>
Submitted on : Monday, June 9, 2014 - 3:51:25 PM
Last modification on : Wednesday, December 19, 2018 - 1:08:13 PM
Long-term archiving on : Tuesday, September 9, 2014 - 10:40:45 AM

File

ACL21.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01003059, version 1

Citation

Christian Obrecht, Frederic Kuznik, Bernard Tourancheau, Jean-Jacques Roux. Global Memory Access Modelling for Efficient Implementation of the LBM on GPUs. Lecture Notes in Computer Science, Springer, 2011, 6449, pp.151-161. ⟨hal-01003059⟩

Share

Metrics

Record views

433

Files downloads

232