llvmpipe: increase number of subpixel bits to eight
authorZack Rusin <zackr@vmware.com>
Thu, 19 Sep 2013 18:10:08 +0000 (14:10 -0400)
committerZack Rusin <zackr@vmware.com>
Mon, 23 Sep 2013 18:53:07 +0000 (14:53 -0400)
commit755c11dc5e94f17097c186edaaa39d818396f14c
tree7b84e6163cf9970cd3e91b28080c534c45921f64
parent6d29db715b8d60718ada1ab8ad19d969cac43caf
llvmpipe: increase number of subpixel bits to eight

Unfortunately d3d10 requires a lot higher precision (e.g.
wgf11clipping tests for it). The smallest number of precision
bits with which it passes is 8. That means that we need to
decrease the maximum length of an edge that we can handle without
subdivision by 4 bits. Abstracted the code a bit to make it easier
to change once to switch to 64bit rasterization.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
src/gallium/drivers/llvmpipe/lp_rast.h
src/gallium/drivers/llvmpipe/lp_setup.c
src/gallium/drivers/llvmpipe/lp_setup_tri.c