Change search
ReferencesLink to record
Permanent link

Direct link
Accelerating IISPH: A Parallel GPGPU Solution Using CUDA
Blekinge Institute of Technology, Faculty of Computing, Department of Creative Technologies.
Blekinge Institute of Technology, Faculty of Computing, Department of Creative Technologies.
2015 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Context. Simulating realistic fluid behavior in incompressible fluids for computer graphics has been pioneered with the implicit incompressible smoothed particle hydrodynamics (IISPH) solver. The algorithm converges faster than other incompressible SPH-solvers, but real-time performance (in the perspective of video games, 30 frames per second) is still an issue when the particle count increases.

Objectives. This thesis aims at improving the performance of the IISPH-solver by proposing a parallel solution that runs on the GPU using CUDA. The solution should not compromise the physical accuracy of the original solution. Investigated aspects are execution time, memory usage and physical accuracy.

Methods. The proposed implementation uses a fine-grained approach where each particle is calculated on a separate thread. It is compared to a sequential and a parallel OpenMP implementation running on the CPU.

Results and Conclusions. It is shown that the parallel CUDA solution allow for real-time performance for approximately 19 times the amount of particles than that of the sequential implementation. For approximately 175 000 particles the simulation runs at the constraint of real-time performance, more particles are still considered interactive. The visual result of the proposed implementation deviated slightly from the ones on the CPU.

Place, publisher, year, edition, pages
2015. , 47 p.
Keyword [en]
implicit incompressible smoothed particle hydrodynamics, fluid simulation, real-time, GPGPU
National Category
Computer Science
URN: urn:nbn:se:bth-10429OAI: diva2:840395
Subject / course
DV2524 Degree Project in Computer Science for Engineers
Educational program
PAACI Master of Science in Game and Software Engineering
Available from: 2015-08-05 Created: 2015-07-08 Last updated: 2015-08-05Bibliographically approved

Open Access in DiVA

fulltext(11183 kB)384 downloads
File information
File name FULLTEXT02.pdfFile size 11183 kBChecksum SHA-512
Type fulltextMimetype application/pdf

Search in DiVA

By author/editor
Eliasson, AndréFranzén, Pontus
By organisation
Department of Creative Technologies
Computer Science

Search outside of DiVA

GoogleGoogle Scholar
Total: 384 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 577 hits
ReferencesLink to record
Permanent link

Direct link