6 files changed, 58 insertions, 85 deletions
diff --git a/test/CodeGen/X86/SIMD/dg.exp b/test/CodeGen/X86/SIMD/dg.exp
deleted file mode 100644
index 629a147736..0000000000
--- a/test/CodeGen/X86/SIMD/dg.exp
+++ /dev/null
@@ -1,5 +0,0 @@
-load_lib llvm.exp
-
-if { [llvm_supports_target X86] } {
-  RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll,c,cpp}]]
-}
diff --git a/test/CodeGen/X86/SIMD/notvunpcklpd.ll b/test/CodeGen/X86/SIMD/notvunpcklpd.ll
deleted file mode 100644
index 3817ee7999..0000000000
--- a/test/CodeGen/X86/SIMD/notvunpcklpd.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llc < %s -mattr=+avx | FileCheck %s
-
-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-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
-target triple = "x86_64-unknown-linux-gnu"
-
-define void @try_([2 x <4 x double>]* noalias %incarray, [2 x <4 x double>]* noalias %incarrayb ) {
-entry:
-	%incarray1 = alloca [2 x <4 x double>]*, align 8
-	%incarrayb1 = alloca [2 x <4 x double>]*, align 8
-	%carray = alloca [2 x <4 x double>], align 16
-	%r = getelementptr [2 x <4 x double>]* %incarray, i32 0, i32 0
-	%rb = getelementptr [2 x <4 x double>]* %incarrayb, i32 0, i32 0
-	%r3 = load <4 x double>* %r, align 8
-	%r4 = load <4 x double>* %rb, align 8
-	%r11 = shufflevector <4 x double> %r3, <4 x double> %r4, <4 x i32> < i32 0, i32 4, i32 1, i32 5 >		; <<4 x double>> [#uses=1]
-; CHECK-NOT: vunpcklpd %ymm
-	%r12 = getelementptr [2 x <4 x double>]* %carray, i32 0, i32 1
-	store <4 x double> %r11, <4 x double>* %r12, align 4
-	ret void
-}
diff --git a/test/CodeGen/X86/SIMD/notvunpcklps.ll b/test/CodeGen/X86/SIMD/notvunpcklps.ll
deleted file mode 100644
index e3b115f6ba..0000000000
--- a/test/CodeGen/X86/SIMD/notvunpcklps.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llc < %s -mattr=+avx | FileCheck %s
-
-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-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
-target triple = "x86_64-unknown-linux-gnu"
-
-define void @try_([2 x <8 x float>]* noalias %incarray, [2 x <8 x float>]* noalias %incarrayb ) {
-enmtry:
-	%incarray1 = alloca [2 x <8 x float>]*, align 8
-	%incarrayb1 = alloca [2 x <8 x float>]*, align 8
-	%carray = alloca [2 x <8 x float>], align 16
-	%r = getelementptr [2 x <8 x float>]* %incarray, i32 0, i32 0
-	%rb = getelementptr [2 x <8 x float>]* %incarrayb, i32 0, i32 0
-	%r3 = load <8 x float>* %r, align 8
-	%r4 = load <8 x float>* %rb, align 8
-	%r8 = shufflevector <8 x float> %r3, <8 x float> %r4, <8 x i32> < i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11 >		; <<8 x float>> [#uses=1]
-; CHECK-NOT: vunpcklps %ymm
-	%r9 = getelementptr [2 x <8 x float>]* %carray, i32 0, i32 0
-	store <8 x float> %r8, <8 x float>* %r9, align 4
-	ret void
-}
diff --git a/test/CodeGen/X86/SIMD/vunpcklpd.ll b/test/CodeGen/X86/SIMD/vunpcklpd.ll
deleted file mode 100644
index 60d23a4f67..0000000000
--- a/test/CodeGen/X86/SIMD/vunpcklpd.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llc < %s -mattr=+avx | FileCheck %s
-
-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-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
-target triple = "x86_64-unknown-linux-gnu"
-
-define void @try_([2 x <4 x double>]* noalias %incarray, [2 x <4 x double>]* noalias %incarrayb ) {
-entry:
-	%incarray1 = alloca [2 x <4 x double>]*, align 8
-	%incarrayb1 = alloca [2 x <4 x double>]*, align 8
-	%carray = alloca [2 x <4 x double>], align 16
-	%r = getelementptr [2 x <4 x double>]* %incarray, i32 0, i32 0
-	%rb = getelementptr [2 x <4 x double>]* %incarrayb, i32 0, i32 0
-	%r3 = load <4 x double>* %r, align 8
-	%r4 = load <4 x double>* %rb, align 8
-	%r11 = shufflevector <4 x double> %r3, <4 x double> %r4, <4 x i32> < i32 0, i32 4, i32 2, i32 6 >		; <<4 x double>> [#uses=1]
-; CHECK: vunpcklpd
-	%r12 = getelementptr [2 x <4 x double>]* %carray, i32 0, i32 1
-	store <4 x double> %r11, <4 x double>* %r12, align 4
-	ret void
-}
diff --git a/test/CodeGen/X86/SIMD/vunpcklps.ll b/test/CodeGen/X86/SIMD/vunpcklps.ll
deleted file mode 100644
index a87b29965e..0000000000
--- a/test/CodeGen/X86/SIMD/vunpcklps.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llc < %s -mattr=+avx | FileCheck %s
-
-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-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
-target triple = "x86_64-unknown-linux-gnu"
-
-define void @try_([2 x <8 x float>]* noalias %incarray, [2 x <8 x float>]* noalias %incarrayb ) {
-entry:
-	%incarray1 = alloca [2 x <8 x float>]*, align 8
-	%incarrayb1 = alloca [2 x <8 x float>]*, align 8
-	%carray = alloca [2 x <8 x float>], align 16
-	%r = getelementptr [2 x <8 x float>]* %incarray, i32 0, i32 0
-	%rb = getelementptr [2 x <8 x float>]* %incarrayb, i32 0, i32 0
-	%r3 = load <8 x float>* %r, align 8
-	%r4 = load <8 x float>* %rb, align 8
-	%r11 = shufflevector <8 x float> %r3, <8 x float> %r4, <8 x i32> < i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13 >		; <<8 x float>> [#uses=1]
-; CHECK: vunpcklps
-	%r12 = getelementptr [2 x <8 x float>]* %carray, i32 0, i32 1
-	store <8 x float> %r11, <8 x float>* %r12, align 4
-	ret void
-}
diff --git a/test/CodeGen/X86/avx-256-unpack.ll b/test/CodeGen/X86/avx-256-unpack.ll
new file mode 100644
index 0000000000..4e906ee155
--- /dev/null
+++ b/test/CodeGen/X86/avx-256-unpack.ll
@@ -0,0 +1,58 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx -mattr=+avx | FileCheck %s
+
+; CHECK: vunpckhps
+define <8 x float> @unpackhips(<8 x float> %src1, <8 x float> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <8 x float> %src1, <8 x float> %src2, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 6, i32 14, i32 7, i32 15>
+  ret <8 x float> %shuffle.i
+}
+
+; CHECK: vunpckhpd
+define <4 x double> @unpackhipd(<4 x double> %src1, <4 x double> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <4 x double> %src1, <4 x double> %src2, <4 x i32> <i32 1, i32 5, i32 3, i32 7>
+  ret <4 x double> %shuffle.i
+}
+
+; CHECK: vunpcklps
+define <8 x float> @unpacklops(<8 x float> %src1, <8 x float> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <8 x float> %src1, <8 x float> %src2, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 4, i32 12, i32 5, i32 13>
+  ret <8 x float> %shuffle.i
+}
+
+; CHECK: vunpcklpd
+define <4 x double> @unpacklopd(<4 x double> %src1, <4 x double> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <4 x double> %src1, <4 x double> %src2, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+  ret <4 x double> %shuffle.i
+}
+
+; CHECK-NOT: vunpcklps %ymm
+define <8 x float> @unpacklops-not(<8 x float> %src1, <8 x float> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <8 x float> %src1, <8 x float> %src2, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
+  ret <8 x float> %shuffle.i
+}
+
+; CHECK-NOT: vunpcklpd %ymm
+define <4 x double> @unpacklopd-not(<4 x double> %src1, <4 x double> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <4 x double> %src1, <4 x double> %src2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
+  ret <4 x double> %shuffle.i
+}
+
+; CHECK-NOT: vunpckhps %ymm
+define <8 x float> @unpackhips-not(<8 x float> %src1, <8 x float> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <8 x float> %src1, <8 x float> %src2, <8 x i32> <i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13>
+  ret <8 x float> %shuffle.i
+}
+
+; CHECK-NOT: vunpckhpd %ymm
+define <4 x double> @unpackhipd-not(<4 x double> %src1, <4 x double> %src2) nounwind uwtable readnone ssp {
+entry:
+  %shuffle.i = shufflevector <4 x double> %src1, <4 x double> %src2, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
+  ret <4 x double> %shuffle.i
+}
+