Hacker News Reader

It sounds like this has some parallels to Mesa's Gallium llvmpipe. (Both projects aim to render some version of OpenGL on the CPU.) Does anyone more familiar with either project know how much those goals overlap and if opportunities for code sharing exist?

View on HN

It also seems to have parallels to Apple's OpenGL implementation which, if the hardware doesn't support the required functionality, falls back to software rendering. It uses LLVM to dynamically generate machine code.

It's also dead slow (obviously).

Notably, that was a very early use of LLVM, before LLVM was well known, wasn't it? Mid 2000s?

Reply on news.ycombinator.com