Abstract
Alias-Wavefront OBJ meshes are a common text file type for transferring 3D mesh data between applications made by different vendors. However, as the mesh complexity gets higher and denser, the files become larger and slower to import. This paper explores the use of GPUs to accelerate the importing and parsing of OBJ files by studying file read-time, runtime, and load resistance. We propose a new method of reading and parsing that circumvents GPU architecture limitations and improves performance, seeing the new GPU method outperforms CPU methods with a 6× - 8× speedup. When running on a heavily loaded system, the new method only received an 80% performance hit, compared to the 160% that the CPU methods received. The loaded GPU speedup compared to unloaded CPU methods was 3.5×, and, when compared to loaded CPU methods, 8×. These results demonstrate that the time is right for further research into the use of data-parallel GPU acceleration beyond that of computer graphics and high performance computing.
Highlights
Graphics processing units (GPUs) have seen a lot of interest, outside their original purpose of rendering computer graphics, as they offer considerable computation speedups over theirManuscript received: 2015-08-25; accepted: 2015-08-29CPU counterparts in particular use cases [1,2,3]
While research into N-body simulations, global illumination, fluid dynamics, and other exciting simulations have drawn the majority of the attention [1,2,3]; this paper focuses on the more “mundane” elements of programming, such as file importing and parsing, to show that these, too, can take advantage of the modern GPU and their impressive potential for parallelization
The area of importing and parsing has seen relatively little interest as GPU architecture and runtime differences mean that algorithms either are unsuitable, or require heavy rework to see any marginal speedup
Summary
Graphics processing units (GPUs) have seen a lot of interest, outside their original purpose of rendering computer graphics, as they offer considerable computation speedups over their. The area of importing and parsing has seen relatively little interest as GPU architecture and runtime differences mean that algorithms either are unsuitable, or require heavy rework to see any marginal speedup. There has been research into natural language parsing [4, 5] and integrating the GPU into the file system under Linux [6], but the closest related research has been limited to optimizing and running queries in SQL or on data stored in XML [7, 8]. Sort and reduction algorithms can run 30 or more times faster than their linear counterparts (https://developer.nvidia.com/Thrust). In this research we look at a task (importing an OBJ mesh to OpenGL) and investigate how a very linear task on the CPU can be re-written so that it can take advantage of the data-driven parallelization that the GPU provides.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
