AArch64: implement efficient f16 bitcasts
authorTim Northover <tnorthover@apple.com>
Fri, 18 Jul 2014 13:07:05 +0000 (13:07 +0000)
committerTim Northover <tnorthover@apple.com>
Fri, 18 Jul 2014 13:07:05 +0000 (13:07 +0000)
commite72ff8829e60e156b787fce12b7cedfc8d30d7f8
tree4e0b33423f85fe2e39dc57eebd5188c484f12055
parentb41b1d4bacf0acc7919fbb7d30c773f8dfc32521
AArch64: implement efficient f16 bitcasts

Because i16 is illegal, there's no native DAG method to
represent a bitcast to or from an f16 type. This meant LLVM was
inserting a stack store/load pair which is really not ideal.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@213378 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/AArch64/AArch64ISelLowering.cpp
test/CodeGen/AArch64/half.ll