projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: fb99fc9) | patch
[PowerPC] Improve int_to_fp(fp_to_int(x)) combining
authorHal Finkel <hfinkel@anl.gov>
Tue, 6 Jan 2015 06:01:57 +0000 (06:01 +0000)
committerHal Finkel <hfinkel@anl.gov>
Tue, 6 Jan 2015 06:01:57 +0000 (06:01 +0000)
commit5efb9188442b06d7a91e92d1b7d94d39c57ffd95
tree8012f2de1afe99984618ecd34feb4f290476e64btree | snapshot
parentfb99fc96c9c373b7b43965a0acbb4a81f09b86c5commit | diff
[PowerPC] Improve int_to_fp(fp_to_int(x)) combining

The old target DAG combine that allowed for performing int_to_fp(fp_to_int(x))
without a load/store pair is updated here with support for unsigned integers,
and to support single-precision values without a third rounding step, on newer
cores with the appropriate instructions.

llvm-svn: 225248
llvm/lib/Target/PowerPC/PPCISelLowering.cpp diff | blob | history
llvm/lib/Target/PowerPC/PPCISelLowering.h diff | blob | history
llvm/test/CodeGen/PowerPC/fp-to-int-to-fp.ll [new file with mode: 0644] blob
Domain: System / Toolchain;