R600/SI: Try to use scalar BFE.
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Fri, 18 Apr 2014 05:19:26 +0000 (05:19 +0000)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Fri, 18 Apr 2014 05:19:26 +0000 (05:19 +0000)
commit746734df1accfeb5f336de211655c304d58cb7b3
treea8dd5416388f375301241808d8c2833c4b4d2e82
parent532a5ffe4c264030d6e350641947fe69fe7babad
R600/SI: Try to use scalar BFE.

Use scalar BFE with constant shift and offset when possible.
This is complicated by the fact that the scalar version packs
the two operands of the vector version into one.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206558 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/R600/AMDGPUISelDAGToDAG.cpp
lib/Target/R600/SIInstrInfo.cpp
test/CodeGen/R600/sext-in-reg.ll