# Spira: Exploiting Voxel Data Structural Properties for Efficient Sparse Convolution in Point Cloud Networks

arXiv: https://arxiv.org/abs/2511.20834 · HTML: https://mlsanity.com/paper/2511.20834/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 39 references that are in the map.
- First posted on arXiv 25 Nov 2025.

## Builds on (most cited first, 19 of 19)

- [ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes](https://mlsanity.com/paper/1702.04405/) (2017)
- [Scalability in Perception for Autonomous Driving: Waymo Open Dataset](https://mlsanity.com/paper/1912.04838/) (2019)
- [4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks](https://mlsanity.com/paper/1904.08755/) (2019)
- [Center-based 3D Object Detection and Tracking](https://mlsanity.com/paper/2006.11275/) (2020)
- [Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation](https://mlsanity.com/paper/2011.10033/) (2020)
- [TorchSparse: Efficient Point Cloud Inference Engine](https://mlsanity.com/paper/2204.10319/) (2022)
- [SparseTIR: Composable Abstractions for Sparse Compilation in Deep Learning](https://mlsanity.com/paper/2207.04606/) (2022)
- [TensorIR: An Abstraction for Automatic Tensorized Program Optimization](https://mlsanity.com/paper/2207.04296/) (2022)
- [TorchSparse++: Efficient Training and Inference Framework for Sparse Convolution on GPUs](https://mlsanity.com/paper/2311.12862/) (2023)
- [Hidet: Task-Mapping Programming Paradigm for Deep Learning Tensor Programs](https://mlsanity.com/paper/2210.09603/) (2022)
- [PointAcc: Efficient Point Cloud Accelerator](https://mlsanity.com/paper/2110.07600/) (2021)
- [Bolt: Bridging the Gap between Auto-tuners and Hardware-native Performance](https://mlsanity.com/paper/2110.15238/) (2021)
- [Mesorasi: Architecture Support for Point Cloud Analytics via Delayed-Aggregation](https://mlsanity.com/paper/2008.06967/) (2020)
- [Crescent: Taming Memory Irregularities for Accelerating Deep Point Cloud Analytics](https://mlsanity.com/paper/2204.10707/) (2022)
- [Onesweep: A Faster Least Significant Digit Radix Sort for GPUs](https://mlsanity.com/paper/2206.01784/) (2022)
- [SPLAT: A framework for optimised GPU code-generation for SParse reguLar ATtention](https://mlsanity.com/paper/2407.16847/) (2024)
- [Minuet: Accelerating 3D Sparse Convolutions on GPUs](https://mlsanity.com/paper/2401.06145/) (2023)
- [SpOctA: A 3D Sparse Convolution Accelerator with Octree-Encoding-Based Map Search and Inherent Sparsity-Aware Processing](https://mlsanity.com/paper/2308.09249/) (2023)
- [Voxel-CIM: An Efficient Compute-in-Memory Accelerator for Voxel-based Point Cloud Neural Networks](https://mlsanity.com/paper/2409.19077/) (2024)

## Abstract (arXiv)

Sparse Convolution (SpC) powers 3D point cloud networks widely used in autonomous driving and augmented/virtual reality. SpC builds a kernel map that stores mappings between input voxel coordinates, output coordinates, and weight offsets, then uses this map to compute feature vectors for output coordinates. Our work identifies three key properties of voxel coordinates: they are integer-valued, bounded within a limited spatial range, and geometrically continuous, i.e., neighboring voxels on the same object surface are highly likely to exist at small spatial offsets from each other. Prior SpC engines do not fully exploit these properties and suffer from high pre-processing and post-processing overheads during kernel map construction. To address this, we design Spira, the first voxel-property-aware SpC engine for GPUs. Spira proposes (i) a high-performance one-shot search algorithm that builds the kernel map with no pre-processing and high data locality, (ii) an effective packed-native processing scheme that accesses packed voxel coordinates at low cost, (iii) a flexible dual-dataflow execution mechanism that efficiently computes output feature vectors by adapting to layer characteristics, and (iv) a network-wide parallelization strategy that builds kernel maps for all SpC layers concurrently at network start. Our evaluation shows that Spira significantly outperforms prior state-of-the-art SpC engines by 1.68x on average and up to 3.04x for end-to-end inference, and by 2.11x on average and up to 3.44x for layer-wise execution across diverse layer configurations. The source code of Spira is freely available at github.com/SPIN-Research-Group/Spira.