Hardware Acceleration of Neural Graphics
Preprint
Abstract
Rendering and inverse-rendering algorithms that drive conventional computer graphics have recently been superseded by neural representations (NR). NRs have recently been used to learn the geometric and the material properties of the scenes and use the information to synthesize photorealistic imagery, thereby promising a replacement for traditional rendering algorithms with scalable quality and predictable performance. In this work we ask the question: Does neural graphics (NG) need hardware support? We studied representative NG applications showing that, if we want to render 4k res. at 60FPS there is a gap of 1.5X-55X in the desired performance on current GPUs. For AR/VR applications, there is an even larger gap of 2-4 OOM between the desired performance and the required system power. We identify that the input encoding and the MLP kernels are the performance bottlenecks, consuming 72%,60% and 59% of application time for multi res. hashgrid, multi res. densegrid and low res. densegrid encodings, respectively. We propose a NG processing cluster, a scalable and flexible hardware architecture that directly accelerates the input encoding and MLP kernels through dedicated engines and supports a wide range of NG applications. We also accelerate the rest of the kernels by fusing them together in Vulkan, which leads to 9.94X kernel-level performance improvement compared to un-fused implementation of the pre-processing and the post-processing kernels. Our results show that, NGPC gives up to 58X end-to-end application-level performance improvement, for multi res. hashgrid encoding on average across the four NG applications, the performance benefits are 12X,20X,33X and 39X for the scaling factor of 8,16,32 and 64, respectively. Our results show that with multi res. hashgrid encoding, NGPC enables the rendering of 4k res. at 30FPS for NeRF and 8k res. at 120FPS for all our other NG applications.Keywords
All Related Versions
This publication has 11 references indexed in Scilit:
- Instant neural graphics primitives with a multiresolution hash encodingACM Transactions on Graphics, 2022
- Advances in Neural RenderingComputer Graphics Forum, 2022
- Large steps in inverse rendering of geometryACM Transactions on Graphics, 2021
- Deep Learning for 3D Point Clouds: A SurveyIEEE Transactions on Pattern Analysis and Machine Intelligence, 2021
- DONeRF: Towards Real‐Time Rendering of Compact Neural Radiance Fields using Depth Oracle NetworksComputer Graphics Forum, 2021
- NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis2020
- State of the Art on Neural RenderingComputer Graphics Forum, 2020
- Scaling equations for the accurate prediction of CMOS device performance from 180 nm to 7 nmIntegration, 2017
- A Survey on Hair Modeling: Styling, Simulation, and RenderingIEEE Transactions on Visualization and Computer Graphics, 2007
- Graphsplatting: visualizing graphs as continuous fieldsIEEE Transactions on Visualization and Computer Graphics, 2003