PowerPC: Lowering floor intrinsic for Altivec

This patch lowers the llvm.floor, llvm.ceil, llvm.trunc, and
llvm.nearbyint to Altivec instruction when using 4 single-precision
float vectors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168086 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/PowerPC/PPCISelLowering.cpp b/lib/Target/PowerPC/PPCISelLowering.cpp
index d51baa6a576a85dc201a0eea3dbdcc6ddc8e8857..7d97450676ea1f5403ad8ab512927ffc4994e985 100644 (file)
--- a/lib/Target/PowerPC/PPCISelLowering.cpp
+++ b/lib/Target/PowerPC/PPCISelLowering.cpp
@@ -402,6 +402,10 @@ PPCTargetLowering::PPCTargetLowering(PPCTargetMachine &TM)
     setOperationAction(ISD::FP_TO_UINT, MVT::v4i32, Legal);
     setOperationAction(ISD::SINT_TO_FP, MVT::v4i32, Legal);
     setOperationAction(ISD::UINT_TO_FP, MVT::v4i32, Legal);
+    setOperationAction(ISD::FFLOOR, MVT::v4f32, Legal);
+    setOperationAction(ISD::FCEIL, MVT::v4f32, Legal);
+    setOperationAction(ISD::FTRUNC, MVT::v4f32, Legal);
+    setOperationAction(ISD::FNEARBYINT, MVT::v4f32, Legal);
 
     addRegisterClass(MVT::v4f32, &PPC::VRRCRegClass);
     addRegisterClass(MVT::v4i32, &PPC::VRRCRegClass);

diff --git a/lib/Target/PowerPC/PPCInstrAltivec.td b/lib/Target/PowerPC/PPCInstrAltivec.td
index ba58c3e4ac8809d74e6df4d942259255148cdf6e..87758e90fbd5dcebfca580ac0c07d013adf1acf2 100644 (file)
--- a/lib/Target/PowerPC/PPCInstrAltivec.td
+++ b/lib/Target/PowerPC/PPCInstrAltivec.td
@@ -721,3 +721,13 @@ def : Pat<(v4f32 (sint_to_fp (v4i32 VRRC:$vA))),
            (VCFSX_0 VRRC:$vA)>;
 def : Pat<(v4f32 (uint_to_fp (v4i32 VRRC:$vA))),
            (VCFUX_0 VRRC:$vA)>;
+
+// Floating-point rounding
+def : Pat<(v4f32 (ffloor (v4f32 VRRC:$vA))),
+          (VRFIM VRRC:$vA)>;
+def : Pat<(v4f32 (fceil (v4f32 VRRC:$vA))),
+          (VRFIP VRRC:$vA)>;
+def : Pat<(v4f32 (ftrunc (v4f32 VRRC:$vA))),
+          (VRFIZ VRRC:$vA)>;
+def : Pat<(v4f32 (fnearbyint (v4f32 VRRC:$vA))),
+          (VRFIN VRRC:$vA)>;

diff --git a/test/CodeGen/PowerPC/vec_rounding.ll b/test/CodeGen/PowerPC/vec_rounding.ll
new file mode 100644 (file)
index 0000000..f41faa0
--- /dev/null
+++ b/test/CodeGen/PowerPC/vec_rounding.ll
@@ -0,0 +1,172 @@
+; RUN: llc -mcpu=pwr6 -mattr=+altivec < %s | FileCheck %s
+
+; Check vector round to single-precision toward -infinity (vrfim)
+; instruction generation using Altivec.
+
+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 double> @llvm.floor.v2f64(<2 x double> %p)
+define <2 x double> @floor_v2f64(<2 x double> %p)
+{
+  %t = call <2 x double> @llvm.floor.v2f64(<2 x double> %p)
+  ret <2 x double> %t
+}
+; CHECK: floor_v2f64:
+; CHECK: bl floor
+; CHECK: bl floor
+
+declare <4 x double> @llvm.floor.v4f64(<4 x double> %p)
+define <4 x double> @floor_v4f64(<4 x double> %p)
+{
+  %t = call <4 x double> @llvm.floor.v4f64(<4 x double> %p)
+  ret <4 x double> %t
+}
+; CHECK: floor_v4f64:
+; CHECK: bl floor
+; CHECK: bl floor
+; CHECK: bl floor
+; CHECK: bl floor
+
+declare <2 x double> @llvm.ceil.v2f64(<2 x double> %p)
+define <2 x double> @ceil_v2f64(<2 x double> %p)
+{
+  %t = call <2 x double> @llvm.ceil.v2f64(<2 x double> %p)
+  ret <2 x double> %t
+}
+; CHECK: ceil_v2f64:
+; CHECK: bl ceil
+; CHECK: bl ceil
+
+declare <4 x double> @llvm.ceil.v4f64(<4 x double> %p)
+define <4 x double> @ceil_v4f64(<4 x double> %p)
+{
+  %t = call <4 x double> @llvm.ceil.v4f64(<4 x double> %p)
+  ret <4 x double> %t
+}
+; CHECK: ceil_v4f64:
+; CHECK: bl ceil
+; CHECK: bl ceil
+; CHECK: bl ceil
+; CHECK: bl ceil
+
+declare <2 x double> @llvm.trunc.v2f64(<2 x double> %p)
+define <2 x double> @trunc_v2f64(<2 x double> %p)
+{
+  %t = call <2 x double> @llvm.trunc.v2f64(<2 x double> %p)
+  ret <2 x double> %t
+}
+; CHECK: trunc_v2f64:
+; CHECK: bl trunc
+; CHECK: bl trunc
+
+declare <4 x double> @llvm.trunc.v4f64(<4 x double> %p)
+define <4 x double> @trunc_v4f64(<4 x double> %p)
+{
+  %t = call <4 x double> @llvm.trunc.v4f64(<4 x double> %p)
+  ret <4 x double> %t
+}
+; CHECK: trunc_v4f64:
+; CHECK: bl trunc
+; CHECK: bl trunc
+; CHECK: bl trunc
+; CHECK: bl trunc
+
+declare <2 x double> @llvm.nearbyint.v2f64(<2 x double> %p)
+define <2 x double> @nearbyint_v2f64(<2 x double> %p)
+{
+  %t = call <2 x double> @llvm.nearbyint.v2f64(<2 x double> %p)
+  ret <2 x double> %t
+}
+; CHECK: nearbyint_v2f64:
+; CHECK: bl nearbyint
+; CHECK: bl nearbyint
+
+declare <4 x double> @llvm.nearbyint.v4f64(<4 x double> %p)
+define <4 x double> @nearbyint_v4f64(<4 x double> %p)
+{
+  %t = call <4 x double> @llvm.nearbyint.v4f64(<4 x double> %p)
+  ret <4 x double> %t
+}
+; CHECK: nearbyint_v4f64:
+; CHECK: bl nearbyint
+; CHECK: bl nearbyint
+; CHECK: bl nearbyint
+; CHECK: bl nearbyint
+
+
+declare <4 x float> @llvm.floor.v4f32(<4 x float> %p)
+define <4 x float> @floor_v4f32(<4 x float> %p)
+{
+  %t = call <4 x float> @llvm.floor.v4f32(<4 x float> %p)
+  ret <4 x float> %t
+}
+; CHECK: floor_v4f32:
+; CHECK: vrfim
+
+declare <8 x float> @llvm.floor.v8f32(<8 x float> %p)
+define <8 x float> @floor_v8f32(<8 x float> %p)
+{
+  %t = call <8 x float> @llvm.floor.v8f32(<8 x float> %p)
+  ret <8 x float> %t
+}
+; CHECK: floor_v8f32:
+; CHECK: vrfim
+; CHECK: vrfim
+
+declare <4 x float> @llvm.ceil.v4f32(<4 x float> %p)
+define <4 x float> @ceil_v4f32(<4 x float> %p)
+{
+  %t = call <4 x float> @llvm.ceil.v4f32(<4 x float> %p)
+  ret <4 x float> %t
+}
+; CHECK: ceil_v4f32:
+; CHECK: vrfip
+
+declare <8 x float> @llvm.ceil.v8f32(<8 x float> %p)
+define <8 x float> @ceil_v8f32(<8 x float> %p)
+{
+  %t = call <8 x float> @llvm.ceil.v8f32(<8 x float> %p)
+  ret <8 x float> %t
+}
+; CHECK: ceil_v8f32:
+; CHECK: vrfip
+; CHECK: vrfip
+
+declare <4 x float> @llvm.trunc.v4f32(<4 x float> %p)
+define <4 x float> @trunc_v4f32(<4 x float> %p)
+{
+  %t = call <4 x float> @llvm.trunc.v4f32(<4 x float> %p)
+  ret <4 x float> %t
+}
+; CHECK: trunc_v4f32:
+; CHECK: vrfiz
+
+declare <8 x float> @llvm.trunc.v8f32(<8 x float> %p)
+define <8 x float> @trunc_v8f32(<8 x float> %p)
+{
+  %t = call <8 x float> @llvm.trunc.v8f32(<8 x float> %p)
+  ret <8 x float> %t
+}
+; CHECK: trunc_v8f32:
+; CHECK: vrfiz
+; CHECK: vrfiz
+
+declare <4 x float> @llvm.nearbyint.v4f32(<4 x float> %p)
+define <4 x float> @nearbyint_v4f32(<4 x float> %p)
+{
+  %t = call <4 x float> @llvm.nearbyint.v4f32(<4 x float> %p)
+  ret <4 x float> %t
+}
+; CHECK: nearbyint_v4f32:
+; CHECK: vrfin
+
+declare <8 x float> @llvm.nearbyint.v8f32(<8 x float> %p)
+define <8 x float> @nearbyint_v8f32(<8 x float> %p)
+{
+  %t = call <8 x float> @llvm.nearbyint.v8f32(<8 x float> %p)
+  ret <8 x float> %t
+}
+; CHECK: nearbyint_v8f32:
+; CHECK: vrfin
+; CHECK: vrfin