projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 88e9008) | patch
NEON use vmov.i32 to splat some f32 values into vectors.
authorJim Grosbach <grosbach@apple.com>
Fri, 20 Jan 2012 18:09:51 +0000 (18:09 +0000)
committerJim Grosbach <grosbach@apple.com>
Fri, 20 Jan 2012 18:09:51 +0000 (18:09 +0000)
commit51222d1551383dd7b95ba356b1a5ed89df69e789
tree4fcc07b235155ab75ee642f659ebedc970ee228ftree | snapshot
parent88e900808adf49ae35395544a169c15d2dc836f7commit | diff
NEON use vmov.i32 to splat some f32 values into vectors.

For bit patterns that aren't representable using the 8-bit floating point
representation for vmov.f32, but are representable via vmov.i32, treat
the .f32 syntax as an alias. Most importantly, this covers the case
'vmov.f32 Vd, #0.0'.

rdar://10616677

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@148556 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/ARM/ARMInstrNEON.td diff | blob | history
lib/Target/ARM/AsmParser/ARMAsmParser.cpp diff | blob | history
test/MC/ARM/simple-fp-encoding.s diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors