Add AVX 256-bit unpack and interleave

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108017 91177308-0d34-0410-b5e6-96231b3b80d8

3 files changed, 81 insertions, 0 deletions

diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index e873652..45b4801 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -165,6 +165,10 @@ def memopv4i32 : PatFrag<(ops node:$ptr), (v4i32 (memop node:$ptr))>;
 def memopv2i64 : PatFrag<(ops node:$ptr), (v2i64 (memop node:$ptr))>;
 def memopv16i8 : PatFrag<(ops node:$ptr), (v16i8 (memop node:$ptr))>;
 
+// FIXME: move this to a more appropriate place after all AVX is done.
+def memopv8f32 : PatFrag<(ops node:$ptr), (v8f32 (memop node:$ptr))>;
+def memopv4f64 : PatFrag<(ops node:$ptr), (v4f64 (memop node:$ptr))>;
+
 // SSSE3 uses MMX registers for some instructions. They aren't aligned on a
 // 16-byte boundary.
 // FIXME: 8 byte alignment for mmx reads is not required
@@ -1539,6 +1543,19 @@ let AddedComplexity = 10 in {
     defm VUNPCKLPD: sse12_unpack_interleave<0x14, unpckl, v2f64, memopv2f64,
           VR128, f128mem, "unpcklpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                          SSEPackedDouble>, OpSize, VEX_4V;
+
+    defm VUNPCKHPSY: sse12_unpack_interleave<0x15, unpckh, v8f32, memopv8f32,
+          VR256, f256mem, "unpckhps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
+                         SSEPackedSingle>, VEX_4V;
+    defm VUNPCKHPDY: sse12_unpack_interleave<0x15, unpckh, v4f64, memopv4f64,
+          VR256, f256mem, "unpckhpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
+                         SSEPackedDouble>, OpSize, VEX_4V;
+    defm VUNPCKLPSY: sse12_unpack_interleave<0x14, unpckl, v8f32, memopv8f32,
+          VR256, f256mem, "unpcklps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
+                         SSEPackedSingle>, VEX_4V;
+    defm VUNPCKLPDY: sse12_unpack_interleave<0x14, unpckl, v4f64, memopv4f64,
+          VR256, f256mem, "unpcklpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
+                         SSEPackedDouble>, OpSize, VEX_4V;
   }
 
   let Constraints = "$src1 = $dst" in {
diff --git a/test/MC/AsmParser/X86/x86_32-encoding.s b/test/MC/AsmParser/X86/x86_32-encoding.s
index e4bf29e..35ef1fa 100644
--- a/test/MC/AsmParser/X86/x86_32-encoding.s
+++ b/test/MC/AsmParser/X86/x86_32-encoding.s
@@ -12354,3 +12354,35 @@
 // CHECK: encoding: [0xc5,0xfd,0x11,0x08]
           vmovupd  %ymm1, (%eax)
 
+// CHECK: vunpckhps  %ymm1, %ymm2, %ymm4
+// CHECK: encoding: [0xc5,0xec,0x15,0xe1]
+          vunpckhps  %ymm1, %ymm2, %ymm4
+
+// CHECK: vunpckhpd  %ymm1, %ymm2, %ymm4
+// CHECK: encoding: [0xc5,0xed,0x15,0xe1]
+          vunpckhpd  %ymm1, %ymm2, %ymm4
+
+// CHECK: vunpcklps  %ymm1, %ymm2, %ymm4
+// CHECK: encoding: [0xc5,0xec,0x14,0xe1]
+          vunpcklps  %ymm1, %ymm2, %ymm4
+
+// CHECK: vunpcklpd  %ymm1, %ymm2, %ymm4
+// CHECK: encoding: [0xc5,0xed,0x14,0xe1]
+          vunpcklpd  %ymm1, %ymm2, %ymm4
+
+// CHECK: vunpckhps  -4(%ebx,%ecx,8), %ymm2, %ymm5
+// CHECK: encoding: [0xc5,0xec,0x15,0x6c,0xcb,0xfc]
+          vunpckhps  -4(%ebx,%ecx,8), %ymm2, %ymm5
+
+// CHECK: vunpckhpd  -4(%ebx,%ecx,8), %ymm2, %ymm5
+// CHECK: encoding: [0xc5,0xed,0x15,0x6c,0xcb,0xfc]
+          vunpckhpd  -4(%ebx,%ecx,8), %ymm2, %ymm5
+
+// CHECK: vunpcklps  -4(%ebx,%ecx,8), %ymm2, %ymm5
+// CHECK: encoding: [0xc5,0xec,0x14,0x6c,0xcb,0xfc]
+          vunpcklps  -4(%ebx,%ecx,8), %ymm2, %ymm5
+
+// CHECK: vunpcklpd  -4(%ebx,%ecx,8), %ymm2, %ymm5
+// CHECK: encoding: [0xc5,0xed,0x14,0x6c,0xcb,0xfc]
+          vunpcklpd  -4(%ebx,%ecx,8), %ymm2, %ymm5
+
diff --git a/test/MC/AsmParser/X86/x86_64-encoding.s b/test/MC/AsmParser/X86/x86_64-encoding.s
index 452aa53..a875cbc 100644
--- a/test/MC/AsmParser/X86/x86_64-encoding.s
+++ b/test/MC/AsmParser/X86/x86_64-encoding.s
@@ -2428,3 +2428,35 @@ pshufb	CPI1_0(%rip), %xmm1
 // CHECK: encoding: [0xc5,0x7d,0x11,0x18]
           vmovupd  %ymm11, (%rax)
 
+// CHECK: vunpckhps  %ymm11, %ymm12, %ymm4
+// CHECK: encoding: [0xc4,0xc1,0x1c,0x15,0xe3]
+          vunpckhps  %ymm11, %ymm12, %ymm4
+
+// CHECK: vunpckhpd  %ymm11, %ymm12, %ymm4
+// CHECK: encoding: [0xc4,0xc1,0x1d,0x15,0xe3]
+          vunpckhpd  %ymm11, %ymm12, %ymm4
+
+// CHECK: vunpcklps  %ymm11, %ymm12, %ymm4
+// CHECK: encoding: [0xc4,0xc1,0x1c,0x14,0xe3]
+          vunpcklps  %ymm11, %ymm12, %ymm4
+
+// CHECK: vunpcklpd  %ymm11, %ymm12, %ymm4
+// CHECK: encoding: [0xc4,0xc1,0x1d,0x14,0xe3]
+          vunpcklpd  %ymm11, %ymm12, %ymm4
+
+// CHECK: vunpckhps  -4(%rbx,%rcx,8), %ymm12, %ymm10
+// CHECK: encoding: [0xc5,0x1c,0x15,0x54,0xcb,0xfc]
+          vunpckhps  -4(%rbx,%rcx,8), %ymm12, %ymm10
+
+// CHECK: vunpckhpd  -4(%rbx,%rcx,8), %ymm12, %ymm10
+// CHECK: encoding: [0xc5,0x1d,0x15,0x54,0xcb,0xfc]
+          vunpckhpd  -4(%rbx,%rcx,8), %ymm12, %ymm10
+
+// CHECK: vunpcklps  -4(%rbx,%rcx,8), %ymm12, %ymm10
+// CHECK: encoding: [0xc5,0x1c,0x14,0x54,0xcb,0xfc]
+          vunpcklps  -4(%rbx,%rcx,8), %ymm12, %ymm10
+
+// CHECK: vunpcklpd  -4(%rbx,%rcx,8), %ymm12, %ymm10
+// CHECK: encoding: [0xc5,0x1d,0x14,0x54,0xcb,0xfc]
+          vunpcklpd  -4(%rbx,%rcx,8), %ymm12, %ymm10
+