projects / platform / upstream / mesa.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 2157757) | patch
radeonsi: use minnum and maxnum LLVM intrinsics for MIN and MAX opcodes
authorMarek Olšák <marek.olsak@amd.com>
Thu, 20 Nov 2014 21:16:09 +0000 (22:16 +0100)
committerMarek Olšák <marek.olsak@amd.com>
Fri, 21 Nov 2014 11:15:58 +0000 (12:15 +0100)
commit645b471d619b654d3bacfa8598f759833e08db4e
treedf1ae1582454d5b0627669998f8985cfaecc765dtree | snapshot
parent21577571b37e68edc0422fbf80932588a4614abccommit | diff
radeonsi: use minnum and maxnum LLVM intrinsics for MIN and MAX opcodes

So far it has been compiled into pretty ugly code (8 instructions or so
for either opcode).

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
src/gallium/drivers/radeonsi/si_shader.c diff | blob | history
Domain: Graphics System / GL;