53KB SoftwareRenderingEngine Hitting 300+FPS a Celeron(Dvtrga–Neuro‑OS Genesis)

  • Posted 1 month ago by neuro-os
  • 3 points
I’ve been working on a minimal, low‑level graphics engine called DVTRGA (Direct Visual Transport & Render Graphics Architecture), originally built as the official graphics subsystem for an experimental OS project (Neuro‑OS Genesis).

What makes it interesting is not the feature set, but the architecture and the size:

53 KB total engine size

Pure software rendering (no GPU, no shaders, no hardware acceleration)

Runs on a basic Intel Celeron

302 FPS sustained in desktop rendering

151.7 FPS while drawing 1,000,000 particles per frame

8.85 GB/s throughput measured in software

Framebuffer‑level rendering, no external dependencies beyond SDL for the demo layer

The goal wasn’t to compete with modern engines, but to explore how far a clean, minimal, low‑level architecture can go when you remove layers of abstraction and focus on memory behavior, cache locality, and predictable pipelines.

The public repo contains:

The base DVTRGA driver (multiplatform, clean, safe to inspect)

A standalone visual demo

A full performance report

Stress test captures

Diagnostic modes (null driver, parallax simulation, 1M‑particle stress test, etc.)

Repo: https://github.com/cyberenigma-lgtm/DVTRGA-Official-Graphics-Engine-of-Neuro-OS-Genesis (github.com in Bing)

If you’re into OSDev, software rendering, framebuffer engines, or extreme minimalism in graphics design, I’d love feedback or critique. The project is intentionally small and transparent so others can inspect the architecture without wading through gigabytes of code.

0 comments