Abstract
In prior work, we introduced the Pipeline Array architecture—a distributed parallel data processing
framework based on the five core characteristics of Assignment, State Holding, Compare, Block, and
Update. The architecture was presented with a two-dimensional (2D) grid as a typical embodiment for
spatially structured data such as image pixels and sensor grids.
This paper clarifies and extends the topological understanding of the Pipeline Array. We emphasize that
the defining characteristic of an Ordered Array is fixed spatial coordinates and neighbor relationships
—not the specific geometric shape of the array. The 2D grid is one valid embodiment, but not a limitation.
The architecture naturally supports other ordered topologies including toroidal (cylindrical), spherical, and
three-dimensional (3D) grid arrangements, all of which preserve the essential properties of fixed
coordinates, neighbor relationships, and O(1) local communication.
We discuss the theoretical feasibility of these topological extensions, their potential advantages for specific
application domains (e.g., seamless boundary handling, volumetric data processing), and their relationship
to the existing Pipeline Array architecture. This note serves to clarify that the Pipeline Array architecture is
not confined to 2D grids and that any ordered arrangement satisfying the fixed-coordinate and neighbor-
relationship criteria falls within the scope of the architecture.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Wu, Jinhui, "Pipeline Array Architecture: Topological Freedom Beyond 2D Grid", Technical Disclosure Commons, (August 12, 2026)
https://www.tdcommons.org/dpubs_series/11341