Scalable multi-GPU implementation of the MAGFLOW simulator
We have developed a robust and scalable multi-GPU (Graphics Processing Unit) version of the cellular-automaton-based MAGFLOW lava simulator. The cellular automaton is partitioned into strips that are assigned to different GPUs, with minimal overlapping. For each GPU, a host thread is launched to man...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Istituto Nazionale di Geofisica e Vulcanologia (INGV)
2011-12-01
|
| Series: | Annals of Geophysics |
| Subjects: | |
| Online Access: | http://www.annalsofgeophysics.eu/index.php/annals/article/view/5342 |
| _version_ | 1819129140516749312 |
|---|---|
| author | Giovanni Gallo Ciro Del Negro Alexis Hérault Giuseppe Bilotta Eugenio Rustico |
| author_facet | Giovanni Gallo Ciro Del Negro Alexis Hérault Giuseppe Bilotta Eugenio Rustico |
| author_sort | Giovanni Gallo |
| collection | DOAJ |
| description | We have developed a robust and scalable multi-GPU (Graphics Processing Unit) version of the cellular-automaton-based MAGFLOW lava simulator. The cellular automaton is partitioned into strips that are assigned to different GPUs, with minimal overlapping. For each GPU, a host thread is launched to manage allocation, deallocation, data transfer and kernel launches; the main host thread coordinates all of the GPUs, to ensure temporal coherence and data integrity. The overlapping borders and maximum temporal step need to be exchanged among the GPUs at the beginning of every evolution of the cellular automaton; data transfers are asynchronous with respect to the computations, to cover the introduced overhead. It is not required to have GPUs of the same speed or capacity; the system runs flawlessly on homogeneous and heterogeneous hardware. The speed-up factor differs from that which is ideal (#GPUs×) only for a constant overhead loss of about 4E−2 · T · #GPUs, with T as the total simulation time. |
| first_indexed | 2024-12-22T08:38:59Z |
| format | Article |
| id | doaj.art-a6277237532942b08439dc7e43ae3207 |
| institution | Directory Open Access Journal |
| issn | 1593-5213 2037-416X |
| language | English |
| last_indexed | 2024-12-22T08:38:59Z |
| publishDate | 2011-12-01 |
| publisher | Istituto Nazionale di Geofisica e Vulcanologia (INGV) |
| record_format | Article |
| series | Annals of Geophysics |
| spelling | doaj.art-a6277237532942b08439dc7e43ae32072022-12-21T18:32:16ZengIstituto Nazionale di Geofisica e Vulcanologia (INGV)Annals of Geophysics1593-52132037-416X2011-12-0154510.4401/ag-5342Scalable multi-GPU implementation of the MAGFLOW simulatorGiovanni GalloCiro Del NegroAlexis HéraultGiuseppe BilottaEugenio RusticoWe have developed a robust and scalable multi-GPU (Graphics Processing Unit) version of the cellular-automaton-based MAGFLOW lava simulator. The cellular automaton is partitioned into strips that are assigned to different GPUs, with minimal overlapping. For each GPU, a host thread is launched to manage allocation, deallocation, data transfer and kernel launches; the main host thread coordinates all of the GPUs, to ensure temporal coherence and data integrity. The overlapping borders and maximum temporal step need to be exchanged among the GPUs at the beginning of every evolution of the cellular automaton; data transfers are asynchronous with respect to the computations, to cover the introduced overhead. It is not required to have GPUs of the same speed or capacity; the system runs flawlessly on homogeneous and heterogeneous hardware. The speed-up factor differs from that which is ideal (#GPUs×) only for a constant overhead loss of about 4E−2 · T · #GPUs, with T as the total simulation time.http://www.annalsofgeophysics.eu/index.php/annals/article/view/5342GPGPU modeling, HPC, parallel programming, GPU, multi-GPU, hazard, lava. |
| spellingShingle | Giovanni Gallo Ciro Del Negro Alexis Hérault Giuseppe Bilotta Eugenio Rustico Scalable multi-GPU implementation of the MAGFLOW simulator Annals of Geophysics GPGPU modeling, HPC, parallel programming, GPU, multi-GPU, hazard, lava. |
| title | Scalable multi-GPU implementation of the MAGFLOW simulator |
| title_full | Scalable multi-GPU implementation of the MAGFLOW simulator |
| title_fullStr | Scalable multi-GPU implementation of the MAGFLOW simulator |
| title_full_unstemmed | Scalable multi-GPU implementation of the MAGFLOW simulator |
| title_short | Scalable multi-GPU implementation of the MAGFLOW simulator |
| title_sort | scalable multi gpu implementation of the magflow simulator |
| topic | GPGPU modeling, HPC, parallel programming, GPU, multi-GPU, hazard, lava. |
| url | http://www.annalsofgeophysics.eu/index.php/annals/article/view/5342 |
| work_keys_str_mv | AT giovannigallo scalablemultigpuimplementationofthemagflowsimulator AT cirodelnegro scalablemultigpuimplementationofthemagflowsimulator AT alexisherault scalablemultigpuimplementationofthemagflowsimulator AT giuseppebilotta scalablemultigpuimplementationofthemagflowsimulator AT eugeniorustico scalablemultigpuimplementationofthemagflowsimulator |