[llvm] [PowerPC] fold i128 equality/inequality compares of two loads into a vectorized compare using vcmpequb.p when Altivec is available (PR #158657)

[zhijian lin via llvm-commits]

================
@@ -35,18 +35,13 @@ define signext i32 @zeroEqualityTest02(ptr %x, ptr %y) {
 define signext i32 @zeroEqualityTest01(ptr %x, ptr %y) {
 ; CHECK-LABEL: zeroEqualityTest01:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    ld 5, 0(3)
-; CHECK-NEXT:    ld 6, 0(4)
-; CHECK-NEXT:    cmpld 5, 6
-; CHECK-NEXT:    bne 0, .LBB1_2
-; CHECK-NEXT:  # %bb.1: # %loadbb1
-; CHECK-NEXT:    ld 5, 8(3)
-; CHECK-NEXT:    ld 4, 8(4)
-; CHECK-NEXT:    li 3, 0
-; CHECK-NEXT:    cmpld 5, 4
-; CHECK-NEXT:    beqlr 0
-; CHECK-NEXT:  .LBB1_2: # %res_block
-; CHECK-NEXT:    li 3, 1
+; CHECK-NEXT:    lxvd2x 34, 0, 4
+; CHECK-NEXT:    lxvd2x 35, 0, 3
+; CHECK-NEXT:    vcmpequb. 2, 3, 2
+; CHECK-NEXT:    mfocrf 3, 2
+; CHECK-NEXT:    rlwinm 3, 3, 25, 31, 31
+; CHECK-NEXT:    cntlzw 3, 3
+; CHECK-NEXT:    srwi 3, 3, 5
----------------
diggerlin wrote:

in the patch , we just make the 

```
#include <memory.h>
int cmp16(const void *a, const void *b)
{
    return memcmp(a, b, 16) == 0;
}

```
equal to 

```
#include <altivec.h>
bool cmpeq16_2(const void *a, const void *b)
{
    const vector unsigned char va = vec_xl(0, (unsigned char *)a);
    const vector unsigned char vb = vec_xl(0, (unsigned char *)b);
    return vec_all_eq(va, vb);
}

```
that is 

 Following code transforms the DAG
   ```
 t0: ch,glue = EntryToken
    t2: i64,ch = CopyFromReg t0, Register:i64 %0
     t3: i128,ch = load<(load (s128) from %ir.a, align 1)> t0, t2,
     undef:i64 t4: i64,ch = CopyFromReg t0, Register:i64 %1 t5: i128,ch =
      load<(load (s128) from %ir.b, align 1)> t0, t4, undef:i64 t6: i1 =
       setcc t3, t5, setne:ch

```
  ---->

```
t0: ch,glue = EntryToken
     t2: i64,ch = CopyFromReg t0, Register:i64 %0
    t3: v16i8,ch = load<(load (s128) from %ir.a, align 1)> t0, t2,
     undef:i64 t4: i64,ch = CopyFromReg t0, Register:i64 %1 t5: v16i8,ch =
      load<(load (s128) from %ir.b, align 1)> t0, t4, undef:i64 t6: i32 =
       llvm.ppc.altivec.vcmpequb.p TargetConstant:i32<10505>,
     Constant:i32<2>, t3, t5 t7: i1 = setcc t6, Constant:i32<0>, seteq:ch

```

I think we can have another patch to let 

```
 llvm.ppc.altivec.vcmpequb.p TargetConstant:i32<10505>,
     Constant:i32<2>, t3, t5 t7: i1 = setcc t6, Constant:i32<0>, seteq:ch
```
     
  convert to your instructions.
  

https://github.com/llvm/llvm-project/pull/158657