If an altivec predicate compare is used immediately by a branch, don't
use a (serializing) MFCR instruction to read the CR6 register, which requires
a compare to get it back to CR's.  Instead, just branch on CR6 directly. :)

For example, for:
void foo2(vector float *A, vector float *B) {
  if (!vec_any_eq(*A, *B))
    *B = (vector float){0,0,0,0};
}

We now generate:

_foo2:
        mfspr r2, 256
        oris r5, r2, 12288
        mtspr 256, r5
        lvx v2, 0, r4
        lvx v3, 0, r3
        vcmpeqfp. v2, v3, v2
        bne cr6, LBB1_2 ; UnifiedReturnBlock
LBB1_1: ; cond_true
        vxor v2, v2, v2
        stvx v2, 0, r4
        mtspr 256, r2
        blr
LBB1_2: ; UnifiedReturnBlock
        mtspr 256, r2
        blr

instead of:

_foo2:
        mfspr r2, 256
        oris r5, r2, 12288
        mtspr 256, r5
        lvx v2, 0, r4
        lvx v3, 0, r3
        vcmpeqfp. v2, v3, v2
        mfcr r3, 2
        rlwinm r3, r3, 27, 31, 31
        cmpwi cr0, r3, 0
        beq cr0, LBB1_2 ; UnifiedReturnBlock
LBB1_1: ; cond_true
        vxor v2, v2, v2
        stvx v2, 0, r4
        mtspr 256, r2
        blr
LBB1_2: ; UnifiedReturnBlock
        mtspr 256, r2
        blr

This implements CodeGen/PowerPC/vec_br_cmp.ll.

llvm-svn: 27804

Canonicalize BUILD_VECTOR's that match VSPLTI's into a single type for each
form, eliminating a bunch of Pat patterns in the .td file and allowing us to
CSE stuff more aggressively.  This implements
PowerPC/buildvec_canonicalize.ll:VSPLTI

llvm-svn: 27614

No functionality changes.

llvm-svn: 27536

Convert vsldoi(x,x) to work the same way other (x,x) cases work.

llvm-svn: 27467

llvm-svn: 27463

llvm-svn: 27457

lower it and LLVM to have one fewer intrinsic.  This implements
CodeGen/PowerPC/vec_shuffle.ll

llvm-svn: 27450

vperm with a perm mask lvx'd from the constant pool.

llvm-svn: 27448

handle all 4 PPC vector types.   This simplifies the matching code and allows
us to eliminate a bunch of patterns.  This also adds cases we were missing,
such as CodeGen/PowerPC/vec_splat.ll:splat_h.

llvm-svn: 27400

llvm-svn: 27359

predicates to VCMPo nodes.

llvm-svn: 27285

same thing and we have a dag node for the former.

llvm-svn: 27205

llvm-svn: 27151

llvm-svn: 27149

 <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

Regression/CodeGen/PowerPC/vec_zero.ll

llvm-svn: 27059

_foo2:
        extsw r2, r3
        std r2, -8(r1)
        lfd f0, -8(r1)
        fcfid f0, f0
        frsp f1, f0
        blr

instead of this:

_foo2:
        lis r2, ha16(LCPI2_0)
        lis r4, 17200
        xoris r3, r3, 32768
        stw r3, -4(r1)
        stw r4, -8(r1)
        lfs f0, lo16(LCPI2_0)(r2)
        lfd f1, -8(r1)
        fsub f0, f1, f0
        frsp f1, f0
        blr

This speeds up Misc/pi from 2.44s->2.09s with LLC and from 3.01->2.18s
with llcbeta (16.7% and 38.1% respectively).

llvm-svn: 26943

llvm-svn: 26896

instructions

llvm-svn: 26894

TODO: leave specific ones as VECTOR_SHUFFLE's and turn them into specialized
operations like vsplt*

llvm-svn: 26887

llvm-svn: 26868

llvm-svn: 26742

void foo(float a, int *b) { *b = a; }

to this:

_foo:
        fctiwz f0, f1
        stfiwx f0, 0, r4
        blr

instead of this:

_foo:
        fctiwz f0, f1
        stfd f0, -8(r1)
        lwz r2, -4(r1)
        stw r2, 0(r4)
        blr

This implements CodeGen/PowerPC/stfiwx.ll, and also incidentally does the
right thing for GCC bugzilla 26505.

llvm-svn: 26447

llvm-svn: 26445

llvm-svn: 26308

llvm-svn: 26305

llvm-svn: 26042

llvm-svn: 26027

llvm-svn: 25853

llvm-svn: 25717

llvm-svn: 25710

the same functionality.  This addresses another piece of bug 680.  Next,
on to fixing Alpha VAARG, which I broke last time.

llvm-svn: 25696

Remove TLI.LowerVA* and replace it with SDNodes that are lowered the same
way as everything else.

llvm-svn: 25606

llvm-svn: 25166

llvm-svn: 24874

llvm-svn: 24700

amount handling that PPC provides.  These are generated by the lowering code
and prevents the dag combiner from assuming (rightfully) that the shifts
don't only look at 5 bits.  This fixes a miscompilation of crafty with
the new front-end.

llvm-svn: 24615

on Darwin to remove smarts from the isel.  This is currently disabled by
default (uncomment setOperationAction(ISD::GlobalAddress to enable it).
tblgen needs to become smarter about tglobaladdr nodes and bigger patterns
needed to be added to the .td file.  However, we can currently emit stuff like
this:  :)

        li r2, lo16(L_x$non_lazy_ptr)
        lis r3, ha16(L_x$non_lazy_ptr)
        lwzx r2, r3, r2

The obvious improvements will follow.

llvm-svn: 24390

allows us to lower legal return types to something else, to meet ABI
requirements (such as that i64 be returned in two i32 regs on Darwin/ppc).

llvm-svn: 23802

redundant after the change.

llvm-svn: 23759