(vextract (v4f32 s2v (f32 load $addr)), 0) -> (f32 load $addr) 
(vextract (v4i32 bc (v4f32 s2v (f32 load $addr))), 0) -> (i32 load $addr)
Remove x86 specific patterns.

llvm-svn: 42677

although not well, and shortening FP converts.

llvm-svn: 42672

llvm-svn: 42671

llvm-svn: 42670

part widths.  Also, return the number of parts actually required to
hold the result's value.
Remove an over-cautious condition from rounding of float->hex conversion.

llvm-svn: 42669

llvm-svn: 42668

bindings headers themselves, hidden behind #ifdef __cplusplus.

llvm-svn: 42666

llvm-svn: 42664

to its inverse.
Testing this as llcbeta

llvm-svn: 42661

llvm-svn: 42656

llvm-svn: 42655

unswitch when frontier includes basic blocks that 
are not inside loop.

llvm-svn: 42654

cause performance degradation.

llvm-svn: 42653

leal 1(%ecx), %edi, which requires 67H prefix.

llvm-svn: 42647

basic arithmetic works.
Rename RTLIB long double functions to distinguish
different flavors of long double; the lib functions
have different names, alas.

llvm-svn: 42644

llvm-svn: 42642

llvm-svn: 42640

llvm-svn: 42639

llvm-svn: 42638

for consistency.

llvm-svn: 42637

llvm-svn: 42636

llvm-svn: 42635

llvm-svn: 42634

llvm-svn: 42633

produce two results.

llvm-svn: 42632

llvm-svn: 42631

llvm-svn: 42630

Optimize |= case for empty RHS

llvm-svn: 42629

llvm-svn: 42628

llvm-svn: 42627

llvm-svn: 42626

scheduler will try a number of tricks in order to avoid generating the
copies. This may not be possible in case the node produces a chain value
that prevent movement. Try unfolding the load from the node before to allow
it to be moved / cloned.

llvm-svn: 42625

llvm-svn: 42624

llvm-svn: 42623

llvm-svn: 42622

address (not just from / to frameindexes).
- Added target hooks to unfold load / store instructions / SDNodes into separate
load, data processing, store instructions / SDNodes.

llvm-svn: 42621

llvm-svn: 42620

llvm-svn: 42619

llvm-svn: 42613

llvm-svn: 42612