Pure 2D Tri-State Topology Grid GPU Architecture: Zero-Float Integer Simulation & FPGA Hardware Equation System
Year: 2026
Type: preprint
Abstract: This repository presents the mathematical equation system and Python hardware-behavioral simulation for a Pure 2D Tri-State Topology Grid GPU Architecture, designed for ultra-low-resource FPGA implementation and zero-framebuffer scanline rendering. Key Architectural Features: • Strict Integer Arithmetic: Eliminates all floating-point operations, trigonometric functions, square roots, division circuits, and 3D transformation matrices. • Tri-State Spatial Topology: Classifies 2D circle primitives into three topological states—Isolated (State 01), Colliding/Overlapping (State 10), and Nested (State 11)—using pure squared-distance integer comparisons. • Tile-Based Parallelism & Overdraw Culling: Partitions the display into independent parallel hardware tiles with AABB coarse culling and dead-zone overdraw skipping for occluded nested primitives. • Bit-Shift Radial Shading: Replaces traditional anti-aliasing (MSAA/TAA) and shadow mapping with hardware bit-shift radial tiers and outer halo rings. Developer-Customizable SRAM Palette Interface: The simulation script intentionally leaves the 8-index SRAM Color Lookup Table (DEFAULT_PALETTE) open as a sandbox configuration for developers. Researchers and developers can freely define, customize, and hot-swap their own RGB color schemes to test visual themes, collision highlights, and nested shadow tiers without modifying the underlying integer topology engine. Included Files: 1. fpga_gpu_engine.py(txt) – Cycle-accurate behavioral simulation in Python (NumPy/Pillow) with an open SRAM palette interface. 2. equations.md – Complete formal specification of the spatial grid, tri-state topology, overdraw culling, and LUT equations.
Author: TAN, KIAN CHENG
Language: English
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7234556539
