Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/C203B439-1E39-4BAC-9A7D-5E592A3D469E | ||
| Year | 2022 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Michail-Iason Chatzakis, "Design and implementation of Monte Carlo Path Tracing system on a reconfigurable logic-based platform", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2022 https://doi.org/10.26233/heallink.tuc.92901 | ||
| Appears in Collections |
In recent years the subject of ray tracing has caused great interest becauseit has many applications, ranging from the movie industry to flight simulators.Major companies develop specific hardware for ray tracing acceleration.Even though work has been done in the past, the interest of ray tracingon FPGAs has decreased as GPUs were deemed to be more suitable for thetask at hand. In this thesis we re-visit the topic, albeit from the baseline ofthe implementation of tree structures which have not been implemented inhardware for ray tracing in the past. We aim to evaluate the performance ofthis ray tracing algorithm across 3 different platforms: CPU, GPU and FPGA.We also aim to introduce a ray tracing FPGA design to exploit the parallelismthat the platform can provide. The first step in the process was the conversionof data structures and algorithms so that they are suitable to FPGA platforms,by converting recursive structures to iterative. Secondly, we create an architecturein which we used streaming pipeline logic in order to achieve goodperformance. With our introduced architecture, based on the given scenewe managed to achieve speedup of up to 2x against an AMD 5600xt GPU,whereas both GPUs and FPGAs perform much better than CPUs even whenmulti-threading is used. To conclude, the use of FPGAs yielded a satisfactoryspeedup and further work on the introduced architecture can lead to betterspeedups.
Copyright © DIAS 2013
