It'd probably be a good idea to perform the higher-dimensional vector transformations of the primitive renderer on the graphics card. This would require transferring a lot more per-vertex data, but as far as I can tell we're not currently memory-bandwidth bound, so this should allow the renderer to perform a lot more smoothly.
This will require the GLSL synthesiser in issue #6.
It'd probably be a good idea to perform the higher-dimensional vector transformations of the primitive renderer on the graphics card. This would require transferring a lot more per-vertex data, but as far as I can tell we're not currently memory-bandwidth bound, so this should allow the renderer to perform a lot more smoothly.
This will require the GLSL synthesiser in issue #6.