diff options
| author | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2012-10-30 18:29:42 +0000 |
|---|---|---|
| committer | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2012-10-30 18:29:42 +0000 |
| commit | c83b5dc625bd85276a23c36c1fbad193203d2bc7 (patch) | |
| tree | 877728d5f2993c57f1a92196910a7e5fe7f03e10 /test/CodeGen/PowerPC/vec_sqrt.ll | |
| parent | cc03331caaeee00b1238654feda8c5a517e48c3a (diff) | |
| download | llvm-c83b5dc625bd85276a23c36c1fbad193203d2bc7.tar.gz llvm-c83b5dc625bd85276a23c36c1fbad193203d2bc7.tar.bz2 llvm-c83b5dc625bd85276a23c36c1fbad193203d2bc7.tar.xz | |
PowerPC: Expand FSRQT for vector types
This patch expands FSQRT for floating point vector types when altivec is
used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167034 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/PowerPC/vec_sqrt.ll')
| -rw-r--r-- | test/CodeGen/PowerPC/vec_sqrt.ll | 71 |
1 files changed, 71 insertions, 0 deletions
diff --git a/test/CodeGen/PowerPC/vec_sqrt.ll b/test/CodeGen/PowerPC/vec_sqrt.ll new file mode 100644 index 0000000000..055da1a229 --- /dev/null +++ b/test/CodeGen/PowerPC/vec_sqrt.ll @@ -0,0 +1,71 @@ +; RUN: llc -mcpu=pwr6 -mattr=+altivec,+fsqrt < %s | FileCheck %s + +; Check for vector sqrt expansion using floating-point types, since altivec +; does not provide an fsqrt instruction for vector. + +target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v128:128:128-n32:64" +target triple = "powerpc64-unknown-linux-gnu" + +declare <2 x float> @llvm.sqrt.v2f32(<2 x float> %val) +declare <4 x float> @llvm.sqrt.v4f32(<4 x float> %val) +declare <8 x float> @llvm.sqrt.v8f32(<8 x float> %val) +declare <2 x double> @llvm.sqrt.v2f64(<2 x double> %val) +declare <4 x double> @llvm.sqrt.v4f64(<4 x double> %val) + +define <2 x float> @v2f32_sqrt(<2 x float> %x) nounwind readnone { +entry: + %sqrt = call <2 x float> @llvm.sqrt.v2f32 (<2 x float> %x) + ret <2 x float> %sqrt +} +; sqrt (<2 x float>) is promoted to sqrt (<4 x float>) +; CHECK: v2f32_sqrt: +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} + +define <4 x float> @v4f32_sqrt(<4 x float> %x) nounwind readnone { +entry: + %sqrt = call <4 x float> @llvm.sqrt.v4f32 (<4 x float> %x) + ret <4 x float> %sqrt +} +; CHECK: v4f32_sqrt: +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} + +define <8 x float> @v8f32_sqrt(<8 x float> %x) nounwind readnone { +entry: + %sqrt = call <8 x float> @llvm.sqrt.v8f32 (<8 x float> %x) + ret <8 x float> %sqrt +} +; CHECK: v8f32_sqrt: +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrts {{[0-9]+}}, {{[0-9]+}} + +define <2 x double> @v2f64_sqrt(<2 x double> %x) nounwind readnone { +entry: + %sqrt = call <2 x double> @llvm.sqrt.v2f64 (<2 x double> %x) + ret <2 x double> %sqrt +} +; CHECK: v2f64_sqrt: +; CHECK: fsqrt {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrt {{[0-9]+}}, {{[0-9]+}} + +define <4 x double> @v4f64_sqrt(<4 x double> %x) nounwind readnone { +entry: + %sqrt = call <4 x double> @llvm.sqrt.v4f64 (<4 x double> %x) + ret <4 x double> %sqrt +} +; CHECK: v4f64_sqrt: +; CHECK: fsqrt {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrt {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrt {{[0-9]+}}, {{[0-9]+}} +; CHECK: fsqrt {{[0-9]+}}, {{[0-9]+}} |