diff options
| author | Sanjay Patel <spatel@rotateright.com> | 2017-04-19 21:23:09 +0000 |
|---|---|---|
| committer | Sanjay Patel <spatel@rotateright.com> | 2017-04-19 21:23:09 +0000 |
| commit | ae382bb6af2c8ac70d0c24ee32418f0980d7f2c8 (patch) | |
| tree | 770daacf8a282c27f3442381da5561d469213c43 /llvm/test/CodeGen/PowerPC/andc.ll | |
| parent | ada0888a111750ff4caec49208d11de4e29bfb61 (diff) | |
| download | bcm5719-llvm-ae382bb6af2c8ac70d0c24ee32418f0980d7f2c8.tar.gz bcm5719-llvm-ae382bb6af2c8ac70d0c24ee32418f0980d7f2c8.zip | |
[DAG] add splat vector support for 'xor' in SimplifyDemandedBits
This allows forming more 'not' ops, so we get improvements for ISAs that have and-not.
Follow-up to:
https://reviews.llvm.org/rL300725
llvm-svn: 300763
Diffstat (limited to 'llvm/test/CodeGen/PowerPC/andc.ll')
| -rw-r--r-- | llvm/test/CodeGen/PowerPC/andc.ll | 6 |
1 files changed, 2 insertions, 4 deletions
diff --git a/llvm/test/CodeGen/PowerPC/andc.ll b/llvm/test/CodeGen/PowerPC/andc.ll index f9b1e4876fd..df47bfc1e38 100644 --- a/llvm/test/CodeGen/PowerPC/andc.ll +++ b/llvm/test/CodeGen/PowerPC/andc.ll @@ -43,10 +43,8 @@ define i1 @foo(i32 %i) { define <4 x i32> @hidden_not_v4i32(<4 x i32> %x) { ; CHECK-LABEL: hidden_not_v4i32: ; CHECK: # BB#0: -; CHECK-NEXT: vspltisw 3, 15 -; CHECK-NEXT: vspltisw 4, 6 -; CHECK-NEXT: xxlxor 0, 34, 35 -; CHECK-NEXT: xxland 34, 0, 36 +; CHECK-NEXT: vspltisw 3, 6 +; CHECK-NEXT: xxlandc 34, 35, 34 ; CHECK-NEXT: blr %xor = xor <4 x i32> %x, <i32 15, i32 15, i32 15, i32 15> %and = and <4 x i32> %xor, <i32 6, i32 6, i32 6, i32 6> |
