IdeaWell

Project ideas from Hacker News discussions.

No Graphics API

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

1. Complexity and Legacy Baggage in Current APIs

Vulkan and DX12 are criticized for excessive complexity, poor documentation, PSO compilation stutters, heavy synchronization, and support for outdated hardware.
"This is definitely a huge issue combined with the API complexity, leading many devs to use wrappers like the aforementioned SDL3" - torginus
"Vulkan is another mess, even if there was a 2.0, how are devs supposed to actually use it" - pjmlp

2. Need for Simplified Modern API

A lean API exposing bindless textures, pointers, and split pipelines on recent GPUs would boost productivity, reduce driver overhead, and enable easier emulation/porting.
"My prototype API fits in one screen: 150 lines of code. The blog post is titled 'No Graphics API'" - (article context, echoed by) vblanco: "Per-drawcall cost goes to nanosecond scale"
"A simplified API means higher programmer productivity, higher robustness, simplified debugging" - flohofwoe

3. Desktop-Mobile Unification Drawbacks

Vulkan's unified API fragments due to mobile limitations (tilers, drivers), holding back desktop; separate APIs or dropping mobile urged.
"Desktop and mobile are way too different for a uniform rendering API. They should be two different flavours" - m-schuetz
"Mobile is holding desktop back not just feature wise, it also fucks up the API" - m-schuetz

🚀 Project Ideas

ModernBindless RHI

Summary

  • A lightweight Rendering Hardware Interface (RHI) library implementing the "No Graphics API" proposal: bindless textures/buffers via GPU pointers, simplified barriers, split pipelines, targeting GPUs from 2018+ (NVIDIA 20xx, AMD RX 5xxx, Intel Arc, modern mobile).
  • Core value: Dramatically reduces boilerplate (150-line API), eliminates PSO explosions/stutters, boosts productivity 3-5x for indie devs/engineers vs. raw Vulkan/DX12.

Details

Key Value
Target Audience Indie game devs, engine authors, non-AAA graphics programmers frustrated with Vulkan/DX12 complexity (e.g., "800 lines for simple tasks" - tonis2).
Core Feature Cross-backend (Vulkan/DX12/Metal) abstraction with raw GPU addresses, auto-cycling sync, programmable vertex fetch; WebGPU backend for browser prototyping.
Tech Stack Rust (safety for pointers/sync), wgpu-rs or Vulkan-Hpp for backends, WGSL/SPIR-V shaders.
Difficulty Medium
Monetization Revenue-ready: Freemium (core free, pro caching/profiling $10/mo).

Notes

  • HN users crave simplicity: "vastly simpler API" (eek2121), "per-drawcall cost to nanoseconds" (vblanco), SDL3 overlap but missing bindless (cyber_kinetist).
  • High discussion potential as open-source drop-in for SDL3_gpu; practical for tools like Piet-GPU/Vello needing unified desktop/mobile compute.

ShaderStutter Buster

Summary

  • Automated shader precompiler and PSO manager: scans app shaders at build-time, generates optimal PSOs for target GPUs, caches globally via Steam-like depot, eliminates runtime hitches.
  • Core value: Ends "compiling shaders" screens/stutters plaguing DX12/Vulkan games (e.g., Doom/Indy Jones require RT hardware).

Details

Key Value
Target Audience Game devs/porting to DX12/Vulkan, engine users (Unreal/Unity forks), Linux Proton users.
Core Feature JIT analyzer + Vulkan Shader Object/DX12 PSO feeder; cloud/offline cache sharing; auto-permutations for multivendor (NVIDIA/AMD/Intel).
Tech Stack C++/Rust CLI, DXC/Glslang compilers, Redis for cache, wasm for WebGPU preview.
Difficulty High
Monetization Revenue-ready: SaaS ($20/mo per project, free OSS tier).

Notes

  • Directly solves "PSO problem" (apitman, modeless: "reduce/eliminate compiling shaders"), "stutters from new objects" (torginus).
  • Utility gold for HN: sparks threads on perf; integrable into engines/RHIs like NVRHI (klaussilveira).

GPU API DocForge

Summary

  • AI-powered interactive docs platform: unifies Vulkan/DX12/Metal tutorials/samples, auto-generates code from natural language ("bindless texture sampler"), runs live via WebGPU/WASM.
  • Core value: Fixes fragmented docs ("DX docs bad state" - pjmlp, "Vulkan mess on Android"), enables rapid prototyping without setup.

Details

Key Value
Target Audience Beginner-to-mid graphics devs, mobile/Android porters, WebGPU users.
Core Feature Semantic search across APIs, diff viewer (e.g., legacy vs. dynamic rendering), embedded playground with extension detection.
Tech Stack Next.js/Svelte, LLM (Ollama/Claude) for query-to-code, Emscripten/WGPU for live samples.
Difficulty Medium
Monetization Hobby

Notes

  • Addresses "hunting through Learn/Github" (pjmlp), "no good primer" (wg0), "terrible documentation" (starkparker).
  • HN love: "seamless interoperability" (ablob), sparks API debates; practical for "reuse components" without full engines.

Read Later