series of unpack and interleave ops.

llvm-svn: 27119

llvm-svn: 27118

Add a bunch of patterns for different datatypes, e.g. bit_convert, undef and
zero vector support.

llvm-svn: 27117

llvm-svn: 27116

llvm-svn: 27115

llvm-svn: 27114

llvm-svn: 27113

llvm-svn: 27112

llvm-svn: 27111

Cannot yet select: intrinsic %llvm.ppc.altivec.lvx

instead of this:

Cannot yet select: 0x9b047e0: v4i32,ch = INTRINSIC 0x9b04540:1, 0x9b04710, 0x9b04790, 0x9b04540

llvm-svn: 27110

llvm-svn: 27109

llvm-svn: 27108

llvm-svn: 27107

 <int -1, int -1, int -1, int -1>
and
 <int 65537, int 65537, int 65537, int 65537>

Using things like:
  vspltisb v0, -1
and:
  vspltish v0, 1

instead of using constant pool loads.

This implements CodeGen/PowerPC/vec_splat.ll:splat_imm_i{32|16}.

llvm-svn: 27106

llvm-svn: 27105

llvm-svn: 27104

llvm-svn: 27103

llvm-svn: 27102

llvm-svn: 27101

llvm-svn: 27100

llvm-svn: 27099

llvm-svn: 27098

llvm-svn: 27097

llvm-svn: 27096

llvm-svn: 27095

llvm-svn: 27094

llvm-svn: 27093

llvm-svn: 27092

llvm-svn: 27091

intrinsics that don't take pointer arguments now work.  For example, we can
compile this:

int test3( __m128d *A) {
  return _mm_movemask_pd(*A);
}
int test4( __m128 *A) {
  return _mm_movemask_ps(*A);
}

to this:

_test3:
        movl 4(%esp), %eax
        movapd (%eax), %xmm0
        movmskpd %xmm0, %eax
        ret
_test4:
        movl 4(%esp), %eax
        movaps (%eax), %xmm0
        movmskps %xmm0, %eax
        ret

llvm-svn: 27090

llvm-svn: 27089

llvm-svn: 27088

llvm-svn: 27087

llvm-svn: 27086

llvm-svn: 27085

llvm-svn: 27084

llvm-svn: 27083

emit the code to select intrinsics, but that is next :)

llvm-svn: 27082

llvm-svn: 27081

llvm-svn: 27080