Rasterization with Data-Parallel Primitives. Makani Buckley, Kenneth Moreland, and Hank Childs. In Eurographics Symposium on Parallel Graphics and Visualization (EGPGV), 2026. (To appear in).
Abstract
Parallel rasterization can suffer from race conditions during fragment generation, which is traditionally addressed by using specialized hardware accessible via vendor graphics APIs. Unfortunately, graphics APIs are increasingly problematic on high-performance computers, either because they are not provided or because of concerns about dependencies with in situ visualization. In response, we present a hardware-agnostic rasterization algorithm that handles race conditions using only data-parallel primitives (DPPs), enabling efficient rendering on HPC systems without graphics API dependencies and aligning with recent efforts to deliver visualization software with DPPs. Our evaluation consists of three phases: (1) evaluating portability across different CPU and GPU architectures, (2) evaluating competitiveness with a community standard, and (3) evaluating performance across varying workloads and available parallelism. The supporting experiments run on both AMD and Nvidia GPUs, considering data sets as large as 460 million triangles and 160 million pixels. While the program performs worse in general when compared to community standards for rasterization, we conclude that the program is portably performant across different architectures without the need for specialized vendor support, and therefore useful for rasterization on high-performance computers.
Full Paper
Coming Soon
Supplemental Materials
The following tarball contains the source code used for the experiments in the paper. This code was compiled against CUDA Toolkit 11.5 and Viskores 1.1.0 in our experiments.
The following archive contains the models tested in the paper.