diff options
| -rw-r--r-- | lib/Target/X86/X86ISelLowering.cpp | 11 | ||||
| -rw-r--r-- | lib/Target/X86/X86InstrSSE.td | 30 | ||||
| -rw-r--r-- | test/CodeGen/X86/avx-256-splat.ll | 13 |
3 files changed, 34 insertions, 20 deletions
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp index 1c827f0dd3..6ce531a960 100644 --- a/lib/Target/X86/X86ISelLowering.cpp +++ b/lib/Target/X86/X86ISelLowering.cpp @@ -970,11 +970,12 @@ X86TargetLowering::X86TargetLowering(X86TargetMachine &TM) setOperationAction(ISD::VSETCC, MVT::v2i64, Custom); if (!UseSoftFloat && Subtarget->hasAVX()) { - addRegisterClass(MVT::v8f32, X86::VR256RegisterClass); - addRegisterClass(MVT::v4f64, X86::VR256RegisterClass); - addRegisterClass(MVT::v8i32, X86::VR256RegisterClass); - addRegisterClass(MVT::v4i64, X86::VR256RegisterClass); - addRegisterClass(MVT::v32i8, X86::VR256RegisterClass); + addRegisterClass(MVT::v32i8, X86::VR256RegisterClass); + addRegisterClass(MVT::v16i16, X86::VR256RegisterClass); + addRegisterClass(MVT::v8i32, X86::VR256RegisterClass); + addRegisterClass(MVT::v8f32, X86::VR256RegisterClass); + addRegisterClass(MVT::v4i64, X86::VR256RegisterClass); + addRegisterClass(MVT::v4f64, X86::VR256RegisterClass); setOperationAction(ISD::LOAD, MVT::v8f32, Legal); setOperationAction(ISD::LOAD, MVT::v4f64, Legal); diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td index 73e465fe29..47aa7da68c 100644 --- a/lib/Target/X86/X86InstrSSE.td +++ b/lib/Target/X86/X86InstrSSE.td @@ -3668,20 +3668,22 @@ let Predicates = [HasXMMInt] in { } let Predicates = [HasAVX] in { - def : Pat<(v4f64 (bitconvert (v8f32 VR256:$src))), (v4f64 VR256:$src)>; - def : Pat<(v4f64 (bitconvert (v4i64 VR256:$src))), (v4f64 VR256:$src)>; - def : Pat<(v4f64 (bitconvert (v32i8 VR256:$src))), (v4f64 VR256:$src)>; - def : Pat<(v8f32 (bitconvert (v4i64 VR256:$src))), (v8f32 VR256:$src)>; - def : Pat<(v8f32 (bitconvert (v4f64 VR256:$src))), (v8f32 VR256:$src)>; - def : Pat<(v8f32 (bitconvert (v32i8 VR256:$src))), (v8f32 VR256:$src)>; - def : Pat<(v4i64 (bitconvert (v8f32 VR256:$src))), (v4i64 VR256:$src)>; - def : Pat<(v4i64 (bitconvert (v4f64 VR256:$src))), (v4i64 VR256:$src)>; - def : Pat<(v4i64 (bitconvert (v32i8 VR256:$src))), (v4i64 VR256:$src)>; - def : Pat<(v32i8 (bitconvert (v4f64 VR256:$src))), (v32i8 VR256:$src)>; - def : Pat<(v32i8 (bitconvert (v4i64 VR256:$src))), (v32i8 VR256:$src)>; - def : Pat<(v32i8 (bitconvert (v8f32 VR256:$src))), (v32i8 VR256:$src)>; - def : Pat<(v32i8 (bitconvert (v8i32 VR256:$src))), (v32i8 VR256:$src)>; - def : Pat<(v8i32 (bitconvert (v32i8 VR256:$src))), (v8i32 VR256:$src)>; + def : Pat<(v4f64 (bitconvert (v8f32 VR256:$src))), (v4f64 VR256:$src)>; + def : Pat<(v4f64 (bitconvert (v4i64 VR256:$src))), (v4f64 VR256:$src)>; + def : Pat<(v4f64 (bitconvert (v32i8 VR256:$src))), (v4f64 VR256:$src)>; + def : Pat<(v8f32 (bitconvert (v4i64 VR256:$src))), (v8f32 VR256:$src)>; + def : Pat<(v8f32 (bitconvert (v4f64 VR256:$src))), (v8f32 VR256:$src)>; + def : Pat<(v8f32 (bitconvert (v32i8 VR256:$src))), (v8f32 VR256:$src)>; + def : Pat<(v8f32 (bitconvert (v16i16 VR256:$src))), (v8f32 VR256:$src)>; + def : Pat<(v4i64 (bitconvert (v8f32 VR256:$src))), (v4i64 VR256:$src)>; + def : Pat<(v4i64 (bitconvert (v4f64 VR256:$src))), (v4i64 VR256:$src)>; + def : Pat<(v4i64 (bitconvert (v32i8 VR256:$src))), (v4i64 VR256:$src)>; + def : Pat<(v32i8 (bitconvert (v4f64 VR256:$src))), (v32i8 VR256:$src)>; + def : Pat<(v32i8 (bitconvert (v4i64 VR256:$src))), (v32i8 VR256:$src)>; + def : Pat<(v32i8 (bitconvert (v8f32 VR256:$src))), (v32i8 VR256:$src)>; + def : Pat<(v32i8 (bitconvert (v8i32 VR256:$src))), (v32i8 VR256:$src)>; + def : Pat<(v8i32 (bitconvert (v32i8 VR256:$src))), (v8i32 VR256:$src)>; + def : Pat<(v16i16 (bitconvert (v8f32 VR256:$src))), (v16i16 VR256:$src)>; } // Move scalar to XMM zero-extended diff --git a/test/CodeGen/X86/avx-256-splat.ll b/test/CodeGen/X86/avx-256-splat.ll index 27ff9268b4..39230fe4fb 100644 --- a/test/CodeGen/X86/avx-256-splat.ll +++ b/test/CodeGen/X86/avx-256-splat.ll @@ -1,6 +1,6 @@ ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx -mattr=+avx | FileCheck %s -; FIXME: use avx versions for punpcklbw and punpckhbw +; FIXME: use avx versions for punpcklbw, punpckhbw and punpckhwd ; CHECK: vextractf128 $0 ; CHECK-NEXT: punpcklbw @@ -14,3 +14,14 @@ entry: ret <32 x i8> %shuffle } +; CHECK: vextractf128 $0 +; CHECK-NEXT: punpckhwd +; CHECK-NEXT: vinsertf128 $0 +; CHECK-NEXT: vinsertf128 $1 +; CHECK-NEXT: vpermilps $85 +define <16 x i16> @funcB(<16 x i16> %a) nounwind uwtable readnone ssp { +entry: + %shuffle = shufflevector <16 x i16> %a, <16 x i16> undef, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> + ret <16 x i16> %shuffle +} + |