AVX-512: gather-scatter tests; added foldable instructions;

Specify GATHER/SCATTER as heavy instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@189736 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86InstrInfo.cpp b/lib/Target/X86/X86InstrInfo.cpp
index c4c090b..c9942c5 100644 (file)
--- a/lib/Target/X86/X86InstrInfo.cpp
+++ b/lib/Target/X86/X86InstrInfo.cpp
@@ -375,7 +375,9 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm)
     { X86::VMOVAPSYrr,  X86::VMOVAPSYmr,    TB_FOLDED_STORE | TB_ALIGN_32 },
     { X86::VMOVDQAYrr,  X86::VMOVDQAYmr,    TB_FOLDED_STORE | TB_ALIGN_32 },
     { X86::VMOVUPDYrr,  X86::VMOVUPDYmr,    TB_FOLDED_STORE },
     { X86::VMOVAPSYrr,  X86::VMOVAPSYmr,    TB_FOLDED_STORE | TB_ALIGN_32 },
     { X86::VMOVDQAYrr,  X86::VMOVDQAYmr,    TB_FOLDED_STORE | TB_ALIGN_32 },
     { X86::VMOVUPDYrr,  X86::VMOVUPDYmr,    TB_FOLDED_STORE },

-    { X86::VMOVUPSYrr,  X86::VMOVUPSYmr,    TB_FOLDED_STORE }
+    { X86::VMOVUPSYrr,  X86::VMOVUPSYmr,    TB_FOLDED_STORE },
+    // AVX-512 foldable instructions
+    { X86::VMOVPDI2DIZrr,X86::VMOVPDI2DIZmr,  TB_FOLDED_STORE }

   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl0); i != e; ++i) {
   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl0); i != e; ++i) {


@@ -581,6 +583,14 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm)
     { X86::TZCNT16rr,       X86::TZCNT16rm,           0 },
     { X86::TZCNT32rr,       X86::TZCNT32rm,           0 },
     { X86::TZCNT64rr,       X86::TZCNT64rm,           0 },
     { X86::TZCNT16rr,       X86::TZCNT16rm,           0 },
     { X86::TZCNT32rr,       X86::TZCNT32rm,           0 },
     { X86::TZCNT64rr,       X86::TZCNT64rm,           0 },

+
+    // AVX-512 foldable instructions
+    { X86::VMOV64toPQIZrr,  X86::VMOVQI2PQIZrm,       0 },
+    { X86::VMOVDI2SSZrr,    X86::VMOVDI2SSZrm,        0 },
+    { X86::VMOVDQA32rr,     X86::VMOVDQA32rm,          TB_ALIGN_64 },
+    { X86::VMOVDQA64rr,     X86::VMOVDQA64rm,          TB_ALIGN_64 },
+    { X86::VMOVDQU32rr,     X86::VMOVDQU32rm,          0 },
+    { X86::VMOVDQU64rr,     X86::VMOVDQU64rm,          0 },

   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl1); i != e; ++i) {
   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl1); i != e; ++i) {


@@ -1180,12 +1190,35 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm)
     { X86::PEXT64rr,          X86::PEXT64rm,            0 },
 
     // AVX-512 foldable instructions
     { X86::PEXT64rr,          X86::PEXT64rm,            0 },
 
     // AVX-512 foldable instructions

+    { X86::VPADDDZrr,         X86::VPADDDZrm,           0 },
+    { X86::VPADDQZrr,         X86::VPADDQZrm,           0 },
+    { X86::VADDPSZrr,         X86::VADDPSZrm,           0 },
+    { X86::VADDPDZrr,         X86::VADDPDZrm,           0 },
+    { X86::VSUBPSZrr,         X86::VSUBPSZrm,           0 },
+    { X86::VSUBPDZrr,         X86::VSUBPDZrm,           0 },
+    { X86::VMULPSZrr,         X86::VMULPSZrm,           0 },
+    { X86::VMULPDZrr,         X86::VMULPDZrm,           0 },
+    { X86::VDIVPSZrr,         X86::VDIVPSZrm,           0 },
+    { X86::VDIVPDZrr,         X86::VDIVPDZrm,           0 },
+    { X86::VMINPSZrr,         X86::VMINPSZrm,           0 },
+    { X86::VMINPDZrr,         X86::VMINPDZrm,           0 },
+    { X86::VMAXPSZrr,         X86::VMAXPSZrm,           0 },
+    { X86::VMAXPDZrr,         X86::VMAXPDZrm,           0 },

     { X86::VPERMPDZri,        X86::VPERMPDZmi,          0 },
     { X86::VPERMPSZrr,        X86::VPERMPSZrm,          0 },
     { X86::VPERMI2Drr,        X86::VPERMI2Drm,          0 },
     { X86::VPERMI2Qrr,        X86::VPERMI2Qrm,          0 },
     { X86::VPERMI2PSrr,       X86::VPERMI2PSrm,         0 },
     { X86::VPERMI2PDrr,       X86::VPERMI2PDrm,         0 },
     { X86::VPERMPDZri,        X86::VPERMPDZmi,          0 },
     { X86::VPERMPSZrr,        X86::VPERMPSZrm,          0 },
     { X86::VPERMI2Drr,        X86::VPERMI2Drm,          0 },
     { X86::VPERMI2Qrr,        X86::VPERMI2Qrm,          0 },
     { X86::VPERMI2PSrr,       X86::VPERMI2PSrm,         0 },
     { X86::VPERMI2PDrr,       X86::VPERMI2PDrm,         0 },

+    { X86::VPSLLVDZrr,        X86::VPSLLVDZrm,          0 },
+    { X86::VPSLLVQZrr,        X86::VPSLLVQZrm,          0 },
+    { X86::VPSRAVDZrr,        X86::VPSRAVDZrm,          0 },
+    { X86::VPSRLVDZrr,        X86::VPSRLVDZrm,          0 },
+    { X86::VPSRLVQZrr,        X86::VPSRLVQZrm,          0 },
+    { X86::VSHUFPDZrri,       X86::VSHUFPDZrmi,         0 },
+    { X86::VSHUFPSZrri,       X86::VSHUFPSZrmi,         0 },
+    { X86::VALIGNQrri,        X86::VALIGNQrmi,          0 },
+    { X86::VALIGNDrri,        X86::VALIGNDrmi,          0 },

   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl2); i != e; ++i) {
   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl2); i != e; ++i) {


@@ -4010,6 +4043,8 @@ static bool hasPartialRegUpdate(unsigned Opcode) {
   case X86::Int_VCVTSD2SSrr:
   case X86::VCVTSS2SDrr:
   case X86::Int_VCVTSS2SDrr:
   case X86::Int_VCVTSD2SSrr:
   case X86::VCVTSS2SDrr:
   case X86::Int_VCVTSS2SDrr:

+  case X86::VCVTSD2SSZrr:
+  case X86::VCVTSS2SDZrr:

   case X86::VRCPSSr:
   case X86::VROUNDSDr:
   case X86::VROUNDSDr_Int:
   case X86::VRCPSSr:
   case X86::VROUNDSDr:
   case X86::VROUNDSDr_Int:


@@ -5064,6 +5099,15 @@ bool X86InstrInfo::isHighLatencyDef(int opc) const {
   case X86::VSQRTSSm:
   case X86::VSQRTSSm_Int:
   case X86::VSQRTSSr:
   case X86::VSQRTSSm:
   case X86::VSQRTSSm_Int:
   case X86::VSQRTSSr:

+
+  case X86::VGATHERQPSZrm:
+  case X86::VGATHERQPDZrm:
+  case X86::VGATHERDPDZrm:
+  case X86::VGATHERDPSZrm:
+  case X86::VPGATHERQDZrm:
+  case X86::VPGATHERQQZrm:
+  case X86::VPGATHERDDZrm:
+  case X86::VPGATHERDQZrm:  

     return true;
   }
 }
     return true;
   }
 }

diff --git a/test/CodeGen/X86/avx512-gather-scatter-intrin.ll b/test/CodeGen/X86/avx512-gather-scatter-intrin.ll
index 2b87d44..0321e95 100644 (file)
--- a/test/CodeGen/X86/avx512-gather-scatter-intrin.ll
+++ b/test/CodeGen/X86/avx512-gather-scatter-intrin.ll
@@ -10,7 +10,7 @@ declare void @llvm.x86.avx512.scatter.qps.mask.512 (i8*, i8, <8 x i64>, <8 x flo
 declare <8 x double> @llvm.x86.avx512.gather.qpd.mask.512 (<8 x double>, i8, <8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpd.mask.512 (i8*, i8, <8 x i64>, <8 x double>, i32)
 
 declare <8 x double> @llvm.x86.avx512.gather.qpd.mask.512 (<8 x double>, i8, <8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpd.mask.512 (i8*, i8, <8 x i64>, <8 x double>, i32)
 

-;CHECK: gather_mask_dps
+;CHECK-LABEL: gather_mask_dps

 ;CHECK: kmovw
 ;CHECK: vgatherdps
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vgatherdps
 ;CHECK: vpadd


@@ -23,7 +23,7 @@ define void @gather_mask_dps(<16 x i32> %ind, <16 x float> %src, i16 %mask, i8*
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_mask_dpd
+;CHECK-LABEL: gather_mask_dpd

 ;CHECK: kmovw
 ;CHECK: vgatherdpd
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vgatherdpd
 ;CHECK: vpadd


@@ -36,7 +36,7 @@ define void @gather_mask_dpd(<8 x i32> %ind, <8 x double> %src, i8 %mask, i8* %b
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_mask_qps
+;CHECK-LABEL: gather_mask_qps

 ;CHECK: kmovw
 ;CHECK: vgatherqps
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vgatherqps
 ;CHECK: vpadd


@@ -49,7 +49,7 @@ define void @gather_mask_qps(<8 x i64> %ind, <8 x float> %src, i8 %mask, i8* %ba
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_mask_qpd
+;CHECK-LABEL: gather_mask_qpd

 ;CHECK: kmovw
 ;CHECK: vgatherqpd
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vgatherqpd
 ;CHECK: vpadd


@@ -74,7 +74,7 @@ declare void @llvm.x86.avx512.scatter.qpi.mask.512 (i8*, i8, <8 x i64>, <8 x i32
 declare <8 x i64> @llvm.x86.avx512.gather.qpq.mask.512 (<8 x i64>, i8, <8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpq.mask.512 (i8*, i8, <8 x i64>, <8 x i64>, i32)
 
 declare <8 x i64> @llvm.x86.avx512.gather.qpq.mask.512 (<8 x i64>, i8, <8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpq.mask.512 (i8*, i8, <8 x i64>, <8 x i64>, i32)
 

-;CHECK: gather_mask_dd
+;CHECK-LABEL: gather_mask_dd

 ;CHECK: kmovw
 ;CHECK: vpgatherdd
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vpgatherdd
 ;CHECK: vpadd


@@ -87,7 +87,7 @@ define void @gather_mask_dd(<16 x i32> %ind, <16 x i32> %src, i16 %mask, i8* %ba
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_mask_qd
+;CHECK-LABEL: gather_mask_qd

 ;CHECK: kmovw
 ;CHECK: vpgatherqd
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vpgatherqd
 ;CHECK: vpadd


@@ -100,7 +100,7 @@ define void @gather_mask_qd(<8 x i64> %ind, <8 x i32> %src, i8 %mask, i8* %base,
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_mask_qq
+;CHECK-LABEL: gather_mask_qq

 ;CHECK: kmovw
 ;CHECK: vpgatherqq
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vpgatherqq
 ;CHECK: vpadd


@@ -113,7 +113,7 @@ define void @gather_mask_qq(<8 x i64> %ind, <8 x i64> %src, i8 %mask, i8* %base,
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_mask_dq
+;CHECK-LABEL: gather_mask_dq

 ;CHECK: kmovw
 ;CHECK: vpgatherdq
 ;CHECK: vpadd
 ;CHECK: kmovw
 ;CHECK: vpgatherdq
 ;CHECK: vpadd


@@ -135,7 +135,7 @@ declare void @llvm.x86.avx512.scatter.qps.512 (i8*, <8 x i64>, <8 x float>, i32)
 declare <8 x double> @llvm.x86.avx512.gather.qpd.512 (<8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpd.512 (i8*, <8 x i64>, <8 x double>, i32)
 
 declare <8 x double> @llvm.x86.avx512.gather.qpd.512 (<8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpd.512 (i8*, <8 x i64>, <8 x double>, i32)
 

-;CHECK: gather_dps
+;CHECK-LABEL: gather_dps

 ;CHECK: kxnorw
 ;CHECK: vgatherdps
 ;CHECK: vscatterdps
 ;CHECK: kxnorw
 ;CHECK: vgatherdps
 ;CHECK: vscatterdps


@@ -147,7 +147,7 @@ define void @gather_dps(<16 x i32> %ind, i8* %base, i8* %stbuf)  {
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_qps
+;CHECK-LABEL: gather_qps

 ;CHECK: kxnorw
 ;CHECK: vgatherqps
 ;CHECK: vscatterqps
 ;CHECK: kxnorw
 ;CHECK: vgatherqps
 ;CHECK: vscatterqps


@@ -159,7 +159,7 @@ define void @gather_qps(<8 x i64> %ind, i8* %base, i8* %stbuf)  {
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_qpd
+;CHECK-LABEL: gather_qpd

 ;CHECK: kxnorw
 ;CHECK: vgatherqpd
 ;CHECK: vpadd
 ;CHECK: kxnorw
 ;CHECK: vgatherqpd
 ;CHECK: vpadd


@@ -184,7 +184,7 @@ declare void @llvm.x86.avx512.scatter.qpi.512 (i8*, <8 x i64>, <8 x i32>, i32)
 declare <8 x i64> @llvm.x86.avx512.gather.qpq.512 (<8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpq.512 (i8*, <8 x i64>, <8 x i64>, i32)
 
 declare <8 x i64> @llvm.x86.avx512.gather.qpq.512 (<8 x i64>, i8*, i32)
 declare void @llvm.x86.avx512.scatter.qpq.512 (i8*, <8 x i64>, <8 x i64>, i32)
 

-;CHECK: gather_dpi
+;CHECK-LABEL: gather_dpi

 ;CHECK: kxnorw
 ;CHECK: vpgatherdd
 ;CHECK: vpscatterdd
 ;CHECK: kxnorw
 ;CHECK: vpgatherdd
 ;CHECK: vpscatterdd


@@ -196,7 +196,8 @@ define void @gather_dpi(<16 x i32> %ind, i8* %base, i8* %stbuf)  {
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_qpq
+;CHECK-LABEL: gather_qpq
+;CHECK: vpxord  %zmm

 ;CHECK: kxnorw
 ;CHECK: vpgatherqq
 ;CHECK: vpadd
 ;CHECK: kxnorw
 ;CHECK: vpgatherqq
 ;CHECK: vpadd


@@ -209,7 +210,8 @@ define void @gather_qpq(<8 x i64> %ind, i8* %base, i8* %stbuf)  {
   ret void
 }
 
   ret void
 }
 

-;CHECK: gather_qpi
+;CHECK-LABEL: gather_qpi
+;CHECK: vpxor %ymm

 ;CHECK: kxnorw
 ;CHECK: vpgatherqd
 ;CHECK: vpadd
 ;CHECK: kxnorw
 ;CHECK: vpgatherqd
 ;CHECK: vpadd