diff options
| author | Michael Liao <michael.liao@intel.com> | 2013-10-15 17:51:02 +0000 |
|---|---|---|
| committer | Michael Liao <michael.liao@intel.com> | 2013-10-15 17:51:02 +0000 |
| commit | dc8c044a9a3e00cab52ca204717de7aee9dab1be (patch) | |
| tree | 89ad409fb774f860488c22c9bf0af76e96f9d7ad /test | |
| parent | d45b3c4653dc4b18074b04662b6d0009880214e3 (diff) | |
| download | external_llvm-dc8c044a9a3e00cab52ca204717de7aee9dab1be.zip external_llvm-dc8c044a9a3e00cab52ca204717de7aee9dab1be.tar.gz external_llvm-dc8c044a9a3e00cab52ca204717de7aee9dab1be.tar.bz2 | |
Fix PR16807
- Lower signed division by constant powers-of-2 to target-independent
DAG operators instead of target-dependent ones to support them better
on targets where vector types are legal but shift operators on that
types are illegal. E.g., on AVX, PSRAW is only available on <8 x i16>
though <16 x i16> is a legal type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192721 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test')
| -rw-r--r-- | test/CodeGen/X86/pr16807.ll | 18 |
1 files changed, 18 insertions, 0 deletions
diff --git a/test/CodeGen/X86/pr16807.ll b/test/CodeGen/X86/pr16807.ll new file mode 100644 index 0000000..6d55d99 --- /dev/null +++ b/test/CodeGen/X86/pr16807.ll @@ -0,0 +1,18 @@ +; RUN: llc < %s -mtriple=x86_64-linux-gnu -mcpu=core-avx-i | FileCheck %s + +define <16 x i16> @f_fu(<16 x i16> %bf) { +allocas: + %avg.i.i = sdiv <16 x i16> %bf, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4> + ret <16 x i16> %avg.i.i +} + +; CHECK: f_fu +; CHECK: psraw +; CHECK: psrlw +; CHECK: paddw +; CHECK: psraw +; CHECK: psraw +; CHECK: psrlw +; CHECK: paddw +; CHECK: psraw +; CHECK: ret |