Instant on-device Neural Radiance Fields (NeRFs) are in growing demand for unleashing the promise of immersive AR/VR experiences, but are still limited by their prohibitive training time. Our profiling analysis reveals a memory-bound inefficiency in NeRF training. To tackle this inefficiency, near-memory processing (NMP) promises to be an effective solution, but also faces challenges due to the unique workloads of NeRFs, including the random hash table lookup, random point processing sequence, and heterogeneous bottleneck steps. Therefore, we propose the first NMP framework, Instant-NeRF, dedicated to enabling instant on-device NeRF training. Experiments on eight datasets consistently validate the effectiveness of Instant-NeRF.
@article{arxiv.2305.05766,
title = {Instant-NeRF: Instant On-Device Neural Radiance Field Training via Algorithm-Accelerator Co-Designed Near-Memory Processing},
author = {Yang Zhao and Shang Wu and Jingqun Zhang and Sixu Li and Chaojian Li and Yingyan Lin},
journal= {arXiv preprint arXiv:2305.05766},
year = {2023}
}