LLVMpipe's Geometry Processing Pipeline Kicks

A month ago we talked about Gallium3D's LLVMpipe performing well and providing a much better software rasterizer than what is available with classic Mesa. Using LLVMpipe and a modest CPU for acceleration, the OpenArena was just about playable without any GPU assistance. Now a month later LLVMpipe is becoming a even more serious performer. LLVMpipe now is able to tap into the new geometry processing pipeline and it's causing some major performance gains.

LLVMpipe leverages the Low-Level Virtual Machine (LLVM) for optimizations and it -- along with Gallium3D itself -- has been in development for some time.

In this blog post by Zack Rusin he details the work that he's done recently on Gallium3D's LLVMpipe with the geometry processing pipeline. Below is the best part.

Anyway, with the new geometry paths the improvements are quite substantial. For vertex processing dominated tests (like geartrain which went from 35fps to 110fps) the improvements are between 2x to about 6x, for cases which are dominated by fragment processing it's obviously a lot less (e.g. openarena went from about 25fps to about 32fps). All in all llvmpipe is looking real good.