AVX512 : VGETMANT implemantation

[igorb]

Summary

AVX512: Implemented encoding and intrinsics for VGETMANTPD/S , VGETMANTSD/S instructions
Added tests for intrinsics and encoding.

[delena]

LGTM

llvm/trunk/include/llvm/IR/IntrinsicsX86.td

let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
 
   def int_x86_avx512_mask_getexp_ss : GCCBuiltin<"__builtin_ia32_getexpss_mask">,
         Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty,
                                     llvm_i8_ty, llvm_i32_ty], [IntrNoMem]>;
   def int_x86_avx512_mask_getexp_sd : GCCBuiltin<"__builtin_ia32_getexpsd_mask">,
         Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty,
                                     llvm_i8_ty, llvm_i32_ty], [IntrNoMem]>;
 
+  def int_x86_avx512_mask_getmant_pd_128 :
+         GCCBuiltin<"__builtin_ia32_getmantpd128_mask">,
+          Intrinsic<[llvm_v2f64_ty],
+          [llvm_v2f64_ty,llvm_i32_ty, llvm_v2f64_ty,  llvm_i8_ty],
+          [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_pd_256 :
+         GCCBuiltin<"__builtin_ia32_getmantpd256_mask">,
+          Intrinsic<[llvm_v4f64_ty],
+          [llvm_v4f64_ty,llvm_i32_ty, llvm_v4f64_ty,  llvm_i8_ty],
+          [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_pd_512 :
+         GCCBuiltin<"__builtin_ia32_getmantpd512_mask">,
+          Intrinsic<[llvm_v8f64_ty],
+          [llvm_v8f64_ty,llvm_i32_ty, llvm_v8f64_ty,  llvm_i8_ty,llvm_i32_ty ],
+          [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_ps_128 :
+         GCCBuiltin<"__builtin_ia32_getmantps128_mask">,
+          Intrinsic<[llvm_v4f32_ty],
+          [llvm_v4f32_ty, llvm_i32_ty, llvm_v4f32_ty,  llvm_i8_ty],
+          [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_ps_256 :
+         GCCBuiltin<"__builtin_ia32_getmantps256_mask">,
+          Intrinsic<[llvm_v8f32_ty],
+          [llvm_v8f32_ty, llvm_i32_ty, llvm_v8f32_ty,  llvm_i8_ty],
+          [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_ps_512 :
+         GCCBuiltin<"__builtin_ia32_getmantps512_mask">,
+          Intrinsic<[llvm_v16f32_ty],
+          [llvm_v16f32_ty,llvm_i32_ty, llvm_v16f32_ty,llvm_i16_ty,llvm_i32_ty],
+          [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_ss :
+         GCCBuiltin<"__builtin_ia32_getmantss_mask">,
+          Intrinsic<[llvm_v4f32_ty],
+          [llvm_v4f32_ty, llvm_v4f32_ty, llvm_i32_ty, llvm_v4f32_ty,
+           llvm_i8_ty, llvm_i32_ty], [IntrNoMem]>;
+
+  def int_x86_avx512_mask_getmant_sd :
+         GCCBuiltin<"__builtin_ia32_getmantsd_mask">,
+          Intrinsic<[llvm_v2f64_ty],
+          [llvm_v2f64_ty, llvm_v2f64_ty, llvm_i32_ty, llvm_v2f64_ty,
+           llvm_i8_ty, llvm_i32_ty], [IntrNoMem]>;
+
   def int_x86_avx512_rsqrt14_ss : GCCBuiltin<"__builtin_ia32_rsqrt14ss_mask">,
         Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty,
                                     llvm_i8_ty], [IntrNoMem]>;
   def int_x86_avx512_rsqrt14_sd : GCCBuiltin<"__builtin_ia32_rsqrt14sd_mask">,
         Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty,
                                     llvm_i8_ty], [IntrNoMem]>;
 
   def int_x86_avx512_rsqrt14_pd_512 : GCCBuiltin<"__builtin_ia32_rsqrt14pd512_mask">,

llvm/trunk/lib/Target/X86/X86ISelLowering.h

enum NodeType : unsigned {
       FMUL_RND,
       FDIV_RND,
       FMAX_RND,
       FMIN_RND,
       FSQRT_RND,
 
       // FP vector get exponent 
       FGETEXP_RND,
+      // Extract Normalized Mantissas
+      VGETMANT,
       // FP Scale
       SCALEF,
       // Integer add/sub with unsigned saturation.
       ADDUS,
       SUBUS,
       // Integer add/sub with signed saturation.
       ADDS,
       SUBS,

llvm/trunk/lib/Target/X86/X86ISelLowering.cpp

case INTR_TYPE_2OP_MASK_RM: {
       if (Op.getNumOperands() == 6)
         Rnd = Op.getOperand(5);
       else
         Rnd = DAG.getConstant(X86::STATIC_ROUNDING::CUR_DIRECTION, dl, MVT::i32);
       return getVectorMaskingNode(DAG.getNode(IntrData->Opc0, dl, VT,
                                               Src1, Src2, Rnd),
                                   Mask, PassThru, Subtarget, DAG);
     }
+    case INTR_TYPE_3OP_SCALAR_MASK_RM: {
+      SDValue Src1 = Op.getOperand(1);
+      SDValue Src2 = Op.getOperand(2);
+      SDValue Src3 = Op.getOperand(3);
+      SDValue PassThru = Op.getOperand(4);
+      SDValue Mask = Op.getOperand(5);
+      SDValue Sae  = Op.getOperand(6);
+
+      return getScalarMaskingNode(DAG.getNode(IntrData->Opc0, dl, VT, Src1,
+                                              Src2, Src3, Sae),
+                                  Mask, PassThru, Subtarget, DAG);
+    }
     case INTR_TYPE_3OP_MASK_RM: {
       SDValue Src1 = Op.getOperand(1);
       SDValue Src2 = Op.getOperand(2);
       SDValue Imm = Op.getOperand(3);
       SDValue PassThru = Op.getOperand(4);
       SDValue Mask = Op.getOperand(5);
       // We specify 2 possible modes for intrinsics, with/without rounding modes.
       // First, we check if the intrinsic have rounding mode (7 operands),
const char *X86TargetLowering::getTargetNodeName(unsigned Opcode) const {
   case X86ISD::FMADD_RND:          return "X86ISD::FMADD_RND";
   case X86ISD::FNMADD_RND:         return "X86ISD::FNMADD_RND";
   case X86ISD::FMSUB_RND:          return "X86ISD::FMSUB_RND";
   case X86ISD::FNMSUB_RND:         return "X86ISD::FNMSUB_RND";
   case X86ISD::FMADDSUB_RND:       return "X86ISD::FMADDSUB_RND";
   case X86ISD::FMSUBADD_RND:       return "X86ISD::FMSUBADD_RND";
   case X86ISD::VRNDSCALE:          return "X86ISD::VRNDSCALE";
   case X86ISD::VREDUCE:            return "X86ISD::VREDUCE";
+  case X86ISD::VGETMANT:           return "X86ISD::VGETMANT";
   case X86ISD::PCMPESTRI:          return "X86ISD::PCMPESTRI";
   case X86ISD::PCMPISTRI:          return "X86ISD::PCMPISTRI";
   case X86ISD::XTEST:              return "X86ISD::XTEST";
   case X86ISD::COMPRESS:           return "X86ISD::COMPRESS";
   case X86ISD::EXPAND:             return "X86ISD::EXPAND";
   case X86ISD::SELECT:             return "X86ISD::SELECT";
   case X86ISD::ADDSUB:             return "X86ISD::ADDSUB";
   case X86ISD::RCP28:              return "X86ISD::RCP28";

llvm/trunk/lib/Target/X86/X86InstrAVX512.td

 multiclass avx512_common_fp_sae_scalar_imm<string OpcodeStr,
                   X86VectorVTInfo _, bits<8> opc, SDNode OpNode, Predicate prd>{
   let Predicates = [prd] in {
      defm Z128 : avx512_fp_scalar_imm<opc, OpcodeStr, OpNode, _>,
                  avx512_fp_sae_scalar_imm<opc, OpcodeStr, OpNode, _>;
   }
 }
 
-multiclass avx512_common_fp_sae_packed_imm_all<string OpcodeStr, bits<8> opcPs,
-             bits<8> opcPd, SDNode OpNode, Predicate prd>{
-  defm PS : avx512_common_unary_fp_sae_packed_imm<OpcodeStr,  avx512vl_f32_info, opcPs,
-                                            OpNode, prd>, EVEX_CD8<32, CD8VF>;
-  defm PD : avx512_common_unary_fp_sae_packed_imm<OpcodeStr,  avx512vl_f64_info, opcPd,
-                                            OpNode, prd>,EVEX_CD8<64, CD8VF> , VEX_W;
+multiclass avx512_common_unary_fp_sae_packed_imm_all<string OpcodeStr,
+                    bits<8> opcPs, bits<8> opcPd, SDNode OpNode, Predicate prd>{
+  defm PS : avx512_common_unary_fp_sae_packed_imm<OpcodeStr, avx512vl_f32_info,
+                            opcPs, OpNode, prd>, EVEX_CD8<32, CD8VF>;
+  defm PD : avx512_common_unary_fp_sae_packed_imm<OpcodeStr, avx512vl_f64_info,
+                            opcPd, OpNode, prd>, EVEX_CD8<64, CD8VF>, VEX_W;
 }
 
 defm VFIXUPIMMPD : avx512_common_fp_sae_packed_imm<"vfixupimmpd",
                               avx512vl_f64_info, 0x54, X86VFixupimm, HasAVX512>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<64, CD8VF>, VEX_W;
 defm VFIXUPIMMPS : avx512_common_fp_sae_packed_imm<"vfixupimmps",
                               avx512vl_f32_info, 0x54, X86VFixupimm, HasAVX512>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<32, CD8VF>;
 
 defm VFIXUPIMMSD: avx512_common_fp_sae_scalar_imm<"vfixupimmsd", f64x_info,
                                                  0x55, X86VFixupimm, HasAVX512>,
       AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<64, CD8VT1>, VEX_W;
 defm VFIXUPIMMSS: avx512_common_fp_sae_scalar_imm<"vfixupimmss", f32x_info,
                                                  0x55, X86VFixupimm, HasAVX512>,
       AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<32, CD8VT1>;
 
-defm VREDUCE : avx512_common_fp_sae_packed_imm_all<"vreduce", 0x56, 0x56, X86VReduce, HasDQI>,AVX512AIi8Base,EVEX;
-defm VRNDSCALE : avx512_common_fp_sae_packed_imm_all<"vrndscale", 0x08, 0x09, X86VRndScale, HasAVX512>,AVX512AIi8Base, EVEX;
+defm VREDUCE   : avx512_common_unary_fp_sae_packed_imm_all<"vreduce", 0x56, 0x56,
+                              X86VReduce, HasDQI>, AVX512AIi8Base, EVEX;
+defm VRNDSCALE : avx512_common_unary_fp_sae_packed_imm_all<"vrndscale", 0x08, 0x09,
+                              X86VRndScale, HasAVX512>, AVX512AIi8Base, EVEX;
+defm VGETMANT : avx512_common_unary_fp_sae_packed_imm_all<"vgetmant", 0x26, 0x26,
+                              X86VGetMant, HasAVX512>, AVX512AIi8Base, EVEX;
+
 
 defm VRANGEPD : avx512_common_fp_sae_packed_imm<"vrangepd", avx512vl_f64_info,
                                                        0x50, X86VRange, HasDQI>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<64, CD8VF>, VEX_W;
 defm VRANGEPS : avx512_common_fp_sae_packed_imm<"vrangeps", avx512vl_f32_info,
                                                        0x50, X86VRange, HasDQI>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<32, CD8VF>;
 
 defm VRANGESD: avx512_common_fp_sae_scalar_imm<"vrangesd", f64x_info,
                                                  0x51, X86VRange, HasDQI>,
       AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<64, CD8VT1>, VEX_W;
 defm VRANGESS: avx512_common_fp_sae_scalar_imm<"vrangess", f32x_info,
                                                  0x51, X86VRange, HasDQI>,
       AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<32, CD8VT1>;
 
 defm VREDUCESD: avx512_common_fp_sae_scalar_imm<"vreducesd", f64x_info,
                                                  0x57, X86Reduces, HasDQI>,
       AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<64, CD8VT1>, VEX_W;
 defm VREDUCESS: avx512_common_fp_sae_scalar_imm<"vreducess", f32x_info,
                                                  0x57, X86Reduces, HasDQI>,
       AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<32, CD8VT1>;
 
+defm VGETMANTSD: avx512_common_fp_sae_scalar_imm<"vgetmantsd", f64x_info,
+                                                 0x27, X86GetMants, HasAVX512>,
+      AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<64, CD8VT1>, VEX_W;
+defm VGETMANTSS: avx512_common_fp_sae_scalar_imm<"vgetmantss", f32x_info,
+                                                 0x27, X86GetMants, HasAVX512>,
+      AVX512AIi8Base, VEX_LIG, EVEX_4V, EVEX_CD8<32, CD8VT1>;
+
 multiclass avx512_shuff_packed_128<string OpcodeStr, AVX512VLVectorVTInfo _,
                                        bits<8> opc, SDNode OpNode = X86Shuf128>{
   let Predicates = [HasAVX512] in {
     defm Z    : avx512_3Op_imm8<opc, OpcodeStr, OpNode, _.info512>, EVEX_V512;
 
   }
   let Predicates = [HasAVX512, HasVLX] in {
      defm Z256 : avx512_3Op_imm8<opc, OpcodeStr, OpNode, _.info256>, EVEX_V256;

llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td

 def X86VPermilpi  : SDNode<"X86ISD::VPERMILPI", SDTShuff2OpI>;
 def X86VPermv     : SDNode<"X86ISD::VPERMV",    SDTShuff2Op>;
 def X86VPermi     : SDNode<"X86ISD::VPERMI",    SDTShuff2OpI>;
 def X86VPermv3    : SDNode<"X86ISD::VPERMV3",   SDTShuff3Op>;
 def X86VPermiv3   : SDNode<"X86ISD::VPERMIV3",  SDTShuff3Op>;
 
 def X86VPerm2x128 : SDNode<"X86ISD::VPERM2X128", SDTShuff3OpI>;
 
-def X86VFixupimm       : SDNode<"X86ISD::VFIXUPIMM", SDTFPBinOpImmRound>;
-def X86VRange          : SDNode<"X86ISD::VRANGE", SDTFPBinOpImmRound>;
-def X86VReduce          : SDNode<"X86ISD::VREDUCE", SDTFPUnaryOpImmRound>;
-def X86VRndScale       : SDNode<"X86ISD::VRNDSCALE", SDTFPUnaryOpImmRound>;
+def X86VFixupimm   : SDNode<"X86ISD::VFIXUPIMM", SDTFPBinOpImmRound>;
+def X86VRange      : SDNode<"X86ISD::VRANGE",    SDTFPBinOpImmRound>;
+def X86VReduce     : SDNode<"X86ISD::VREDUCE",   SDTFPUnaryOpImmRound>;
+def X86VRndScale   : SDNode<"X86ISD::VRNDSCALE", SDTFPUnaryOpImmRound>;
+def X86VGetMant    : SDNode<"X86ISD::VGETMANT",  SDTFPUnaryOpImmRound>;
 
 def X86SubVBroadcast : SDNode<"X86ISD::SUBV_BROADCAST",
                     SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVec<1>,
                                          SDTCisSubVecOfVec<1, 0>]>, []>;
 def X86VBroadcast : SDNode<"X86ISD::VBROADCAST", SDTVBroadcast>;
 def X86Vinsert   : SDNode<"X86ISD::VINSERT",  SDTypeProfile<1, 3,
                               [SDTCisSameAs<0, 1>, SDTCisPtrTy<3>]>, []>;
 def X86Vextract   : SDNode<"X86ISD::VEXTRACT",  SDTypeProfile<1, 2,

 def X86FnmsubRnd    : SDNode<"X86ISD::FNMSUB_RND",    SDTFmaRound>;
 def X86FmaddsubRnd  : SDNode<"X86ISD::FMADDSUB_RND",  SDTFmaRound>;
 def X86FmsubaddRnd  : SDNode<"X86ISD::FMSUBADD_RND",  SDTFmaRound>;
 
 def X86rsqrt28   : SDNode<"X86ISD::RSQRT28",  STDFp1SrcRm>;
 def X86rcp28     : SDNode<"X86ISD::RCP28",    STDFp1SrcRm>;
 def X86exp2      : SDNode<"X86ISD::EXP2",     STDFp1SrcRm>;
 
-def X86rsqrt28s  : SDNode<"X86ISD::RSQRT28",  STDFp2SrcRm>;
-def X86rcp28s    : SDNode<"X86ISD::RCP28",    STDFp2SrcRm>;
+def X86rsqrt28s  : SDNode<"X86ISD::RSQRT28",   STDFp2SrcRm>;
+def X86rcp28s    : SDNode<"X86ISD::RCP28",     STDFp2SrcRm>;
 def X86RndScales : SDNode<"X86ISD::VRNDSCALE", STDFp3SrcRm>;
-def X86Reduces   : SDNode<"X86ISD::VREDUCE", STDFp3SrcRm>;
+def X86Reduces   : SDNode<"X86ISD::VREDUCE",   STDFp3SrcRm>;
+def X86GetMants  : SDNode<"X86ISD::VGETMANT",  STDFp3SrcRm>;
 
 def SDT_PCMPISTRI : SDTypeProfile<2, 3, [SDTCisVT<0, i32>, SDTCisVT<1, i32>,
                                          SDTCisVT<2, v16i8>, SDTCisVT<3, v16i8>,
                                          SDTCisVT<4, i8>]>;
 def SDT_PCMPESTRI : SDTypeProfile<2, 5, [SDTCisVT<0, i32>, SDTCisVT<1, i32>,
                                          SDTCisVT<2, v16i8>, SDTCisVT<3, i32>,
                                          SDTCisVT<4, v16i8>, SDTCisVT<5, i32>,
                                          SDTCisVT<6, i8>]>;

llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h

enum IntrinsicType {
   GATHER, SCATTER, PREFETCH, RDSEED, RDRAND, RDPMC, RDTSC, XTEST, ADX,
   INTR_TYPE_1OP, INTR_TYPE_2OP, INTR_TYPE_3OP, INTR_TYPE_4OP,
   CMP_MASK, CMP_MASK_CC, VSHIFT, VSHIFT_MASK, COMI,
   INTR_TYPE_1OP_MASK, INTR_TYPE_1OP_MASK_RM,
   INTR_TYPE_2OP_MASK, INTR_TYPE_2OP_MASK_RM,
   INTR_TYPE_3OP_MASK, INTR_TYPE_3OP_MASK_RM, INTR_TYPE_3OP_IMM8_MASK,
   FMA_OP_MASK, FMA_OP_MASKZ, FMA_OP_MASK3, VPERM_3OP_MASK,
   VPERM_3OP_MASKZ,
-  INTR_TYPE_SCALAR_MASK_RM, COMPRESS_EXPAND_IN_REG, COMPRESS_TO_MEM,
+  INTR_TYPE_SCALAR_MASK_RM, INTR_TYPE_3OP_SCALAR_MASK_RM,
+  COMPRESS_EXPAND_IN_REG, COMPRESS_TO_MEM,
   TRUNCATE_TO_MEM_VI8, TRUNCATE_TO_MEM_VI16, TRUNCATE_TO_MEM_VI32,
   EXPAND_FROM_MEM, BLEND
 };
 
 struct IntrinsicData {
 
   unsigned      Id;
   IntrinsicType Type;
X86_INTRINSIC_DATA(avx512_mask_getexp_ps_128, INTR_TYPE_1OP_MASK_RM,
                      X86ISD::FGETEXP_RND, 0),
   X86_INTRINSIC_DATA(avx512_mask_getexp_ps_256, INTR_TYPE_1OP_MASK_RM,
                      X86ISD::FGETEXP_RND, 0),
   X86_INTRINSIC_DATA(avx512_mask_getexp_ps_512, INTR_TYPE_1OP_MASK_RM,
                      X86ISD::FGETEXP_RND, 0),
   X86_INTRINSIC_DATA(avx512_mask_getexp_sd, INTR_TYPE_SCALAR_MASK_RM,
                      X86ISD::FGETEXP_RND, 0),
   X86_INTRINSIC_DATA(avx512_mask_getexp_ss, INTR_TYPE_SCALAR_MASK_RM,
                      X86ISD::FGETEXP_RND, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_pd_128, INTR_TYPE_2OP_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_pd_256, INTR_TYPE_2OP_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_pd_512, INTR_TYPE_2OP_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_ps_128, INTR_TYPE_2OP_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_ps_256, INTR_TYPE_2OP_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_ps_512, INTR_TYPE_2OP_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_sd, INTR_TYPE_3OP_SCALAR_MASK_RM,
+                     X86ISD::VGETMANT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_getmant_ss, INTR_TYPE_3OP_SCALAR_MASK_RM,
+                     X86ISD::VGETMANT, 0),
   X86_INTRINSIC_DATA(avx512_mask_max_pd_128, INTR_TYPE_2OP_MASK, X86ISD::FMAX, 0),
   X86_INTRINSIC_DATA(avx512_mask_max_pd_256, INTR_TYPE_2OP_MASK, X86ISD::FMAX, 0),
   X86_INTRINSIC_DATA(avx512_mask_max_pd_512, INTR_TYPE_2OP_MASK, X86ISD::FMAX,
                      X86ISD::FMAX_RND),
   X86_INTRINSIC_DATA(avx512_mask_max_ps_128, INTR_TYPE_2OP_MASK, X86ISD::FMAX, 0),
   X86_INTRINSIC_DATA(avx512_mask_max_ps_256, INTR_TYPE_2OP_MASK, X86ISD::FMAX, 0),
   X86_INTRINSIC_DATA(avx512_mask_max_ps_512, INTR_TYPE_2OP_MASK, X86ISD::FMAX,
                      X86ISD::FMAX_RND),

llvm/trunk/test/CodeGen/X86/avx512-intrinsics.ll

; CHECK-NEXT:    retq
   %res3 = call <2 x double> @llvm.x86.avx512.mask.getexp.sd(<2 x double>%a0, <2 x double> %a1, <2 x double> zeroinitializer, i8 -1, i32 4)
 
   %res.1 = fadd <2 x double> %res0, %res1
   %res.2 = fadd <2 x double> %res2, %res3
   %res   = fadd <2 x double> %res.1, %res.2
   ret <2 x double> %res
 }
 
+declare <8 x double> @llvm.x86.avx512.mask.getmant.pd.512(<8 x double>, i32, <8 x double>, i8, i32)
+
+define <8 x double>@test_int_x86_avx512_mask_getmant_pd_512(<8 x double> %x0, <8 x double> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_pd_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vgetmantpd $11, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vgetmantpd $11,{sae}, %zmm0, %zmm0
+; CHECK-NEXT:    vaddpd %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x double> @llvm.x86.avx512.mask.getmant.pd.512(<8 x double> %x0, i32 11, <8 x double> %x2, i8 %x3, i32 4)
+  %res1 = call <8 x double> @llvm.x86.avx512.mask.getmant.pd.512(<8 x double> %x0, i32 11, <8 x double> %x2, i8 -1, i32 8)
+  %res2 = fadd <8 x double> %res, %res1
+  ret <8 x double> %res2
+}
+
+declare <16 x float> @llvm.x86.avx512.mask.getmant.ps.512(<16 x float>, i32, <16 x float>, i16, i32)
+
+define <16 x float>@test_int_x86_avx512_mask_getmant_ps_512(<16 x float> %x0, <16 x float> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_ps_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vgetmantps $11, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vgetmantps $11,{sae}, %zmm0, %zmm0
+; CHECK-NEXT:    vaddps %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <16 x float> @llvm.x86.avx512.mask.getmant.ps.512(<16 x float> %x0, i32 11, <16 x float> %x2, i16 %x3, i32 4)
+  %res1 = call <16 x float> @llvm.x86.avx512.mask.getmant.ps.512(<16 x float> %x0, i32 11, <16 x float> %x2, i16 -1, i32 8)
+  %res2 = fadd <16 x float> %res, %res1
+  ret <16 x float> %res2
+}
+
+declare <2 x double> @llvm.x86.avx512.mask.getmant.sd(<2 x double>, <2 x double>, i32, <2 x double>, i8, i32)
+
+define <2 x double>@test_int_x86_avx512_mask_getmant_sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_sd:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    andl $1, %edi
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vmovaps %zmm2, %zmm3
+; CHECK-NEXT:    vgetmantsd $11, %xmm1, %xmm0, %xmm3 {%k1}
+; CHECK-NEXT:    vgetmantsd $11, %xmm1, %xmm0, %xmm4 {%k1} {z}
+; CHECK-NEXT:    vgetmantsd $11, %xmm1, %xmm0, %xmm5
+; CHECK-NEXT:    vgetmantsd $11,{sae}, %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vaddpd %xmm4, %xmm3, %xmm0
+; CHECK-NEXT:    vaddpd %xmm5, %xmm2, %xmm1
+; CHECK-NEXT:    vaddpd %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    retq
+  %res  = call <2 x double> @llvm.x86.avx512.mask.getmant.sd(<2 x double> %x0, <2 x double> %x1, i32 11, <2 x double> %x2, i8 %x3, i32 4)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.getmant.sd(<2 x double> %x0, <2 x double> %x1, i32 11, <2 x double> zeroinitializer, i8 %x3, i32 4)
+  %res2 = call <2 x double> @llvm.x86.avx512.mask.getmant.sd(<2 x double> %x0, <2 x double> %x1, i32 11, <2 x double> %x2, i8 %x3, i32 8)
+  %res3 = call <2 x double> @llvm.x86.avx512.mask.getmant.sd(<2 x double> %x0, <2 x double> %x1, i32 11, <2 x double> %x2, i8 -1, i32 4)
+  %res11 = fadd <2 x double> %res, %res1
+  %res12 = fadd <2 x double> %res2, %res3
+  %res13 = fadd <2 x double> %res11, %res12
+  ret <2 x double> %res13
+}
+
+declare <4 x float> @llvm.x86.avx512.mask.getmant.ss(<4 x float>, <4 x float>, i32, <4 x float>, i8, i32)
+
+define <4 x float>@test_int_x86_avx512_mask_getmant_ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_ss:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    andl $1, %edi
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vgetmantss $11, %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vgetmantss $11, %xmm1, %xmm0, %xmm3 {%k1} {z}
+; CHECK-NEXT:    vgetmantss $11, %xmm1, %xmm0, %xmm4
+; CHECK-NEXT:    vgetmantss $11,{sae}, %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vaddps %xmm3, %xmm2, %xmm1
+; CHECK-NEXT:    vaddps %xmm4, %xmm0, %xmm0
+; CHECK-NEXT:    vaddps %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res  = call <4 x float> @llvm.x86.avx512.mask.getmant.ss(<4 x float> %x0, <4 x float> %x1, i32 11, <4 x float> %x2, i8 %x3, i32 4)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.getmant.ss(<4 x float> %x0, <4 x float> %x1, i32 11, <4 x float> zeroinitializer, i8 %x3, i32 4)
+  %res2 = call <4 x float> @llvm.x86.avx512.mask.getmant.ss(<4 x float> %x0, <4 x float> %x1, i32 11, <4 x float> %x2, i8 -1, i32 8)
+  %res3 = call <4 x float> @llvm.x86.avx512.mask.getmant.ss(<4 x float> %x0, <4 x float> %x1, i32 11, <4 x float> %x2, i8 -1, i32 4)
+  %res11 = fadd <4 x float> %res, %res1
+  %res12 = fadd <4 x float> %res2, %res3
+  %res13 = fadd <4 x float> %res11, %res12
+  ret <4 x float> %res13
+}
+
 declare <8 x double> @llvm.x86.avx512.mask.shuf.pd.512(<8 x double>, <8 x double>, i32, <8 x double>, i8)
 
 define <8 x double>@test_int_x86_avx512_mask_shuf_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x3, i8 %x4) {
 ; CHECK-LABEL: test_int_x86_avx512_mask_shuf_pd_512:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    movzbl %dil, %eax
 ; CHECK-NEXT:    kmovw %eax, %k1
 ; CHECK-NEXT:    vshufpd $22, %zmm1, %zmm0, %zmm2 {%k1}

llvm/trunk/test/CodeGen/X86/avx512vl-intrinsics.ll

 ; CHECK: vrndscaleps
 define <8 x float>@test_int_x86_avx512_mask_rndscale_ps_256(<8 x float> %x0, <8 x float> %x2, i8 %x3) {
   %res = call <8 x float> @llvm.x86.avx512.mask.rndscale.ps.256(<8 x float> %x0, i32 5, <8 x float> %x2, i8 %x3)
   %res1 = call <8 x float> @llvm.x86.avx512.mask.rndscale.ps.256(<8 x float> %x0, i32 66, <8 x float> %x2, i8 -1)
   %res2 = fadd <8 x float> %res, %res1
   ret <8 x float> %res2
 }
 
+declare <2 x double> @llvm.x86.avx512.mask.getmant.pd.128(<2 x double>, i32, <2 x double>, i8)
+
+define <2 x double>@test_int_x86_avx512_mask_getmant_pd_128(<2 x double> %x0, <2 x double> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_pd_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vgetmantpd $11, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vgetmantpd $11, %xmm0, %xmm2 {%k1} {z}
+; CHECK-NEXT:    vgetmantpd $11, %xmm0, %xmm0
+; CHECK-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    vaddpd %xmm0, %xmm2, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <2 x double> @llvm.x86.avx512.mask.getmant.pd.128(<2 x double> %x0, i32 11, <2 x double> %x2, i8 %x3)
+  %res2 = call <2 x double> @llvm.x86.avx512.mask.getmant.pd.128(<2 x double> %x0, i32 11, <2 x double> zeroinitializer, i8 %x3)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.getmant.pd.128(<2 x double> %x0, i32 11, <2 x double> %x2, i8 -1)
+  %res3 = fadd <2 x double> %res, %res1
+  %res4 = fadd <2 x double> %res2, %res3
+  ret <2 x double> %res4
+}
+
+declare <4 x double> @llvm.x86.avx512.mask.getmant.pd.256(<4 x double>, i32, <4 x double>, i8)
+
+define <4 x double>@test_int_x86_avx512_mask_getmant_pd_256(<4 x double> %x0, <4 x double> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_pd_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vgetmantpd $11, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vgetmantpd $11, %ymm0, %ymm0
+; CHECK-NEXT:    vaddpd %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <4 x double> @llvm.x86.avx512.mask.getmant.pd.256(<4 x double> %x0, i32 11, <4 x double> %x2, i8 %x3)
+  %res1 = call <4 x double> @llvm.x86.avx512.mask.getmant.pd.256(<4 x double> %x0, i32 11, <4 x double> %x2, i8 -1)
+  %res2 = fadd <4 x double> %res, %res1
+  ret <4 x double> %res2
+}
+
+declare <4 x float> @llvm.x86.avx512.mask.getmant.ps.128(<4 x float>, i32, <4 x float>, i8)
+
+define <4 x float>@test_int_x86_avx512_mask_getmant_ps_128(<4 x float> %x0, <4 x float> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_ps_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vgetmantps $11, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vgetmantps $11, %xmm0, %xmm0
+; CHECK-NEXT:    vaddps %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <4 x float> @llvm.x86.avx512.mask.getmant.ps.128(<4 x float> %x0, i32 11, <4 x float> %x2, i8 %x3)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.getmant.ps.128(<4 x float> %x0, i32 11, <4 x float> %x2, i8 -1)
+  %res2 = fadd <4 x float> %res, %res1
+  ret <4 x float> %res2
+}
+
+declare <8 x float> @llvm.x86.avx512.mask.getmant.ps.256(<8 x float>, i32, <8 x float>, i8)
+
+define <8 x float>@test_int_x86_avx512_mask_getmant_ps_256(<8 x float> %x0, <8 x float> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_getmant_ps_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vgetmantps $11, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vgetmantps $11, %ymm0, %ymm0
+; CHECK-NEXT:    vaddps %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <8 x float> @llvm.x86.avx512.mask.getmant.ps.256(<8 x float> %x0, i32 11, <8 x float> %x2, i8 %x3)
+  %res1 = call <8 x float> @llvm.x86.avx512.mask.getmant.ps.256(<8 x float> %x0, i32 11, <8 x float> %x2, i8 -1)
+  %res2 = fadd <8 x float> %res, %res1
+  ret <8 x float> %res2
+}
+
 declare <2 x double> @llvm.x86.avx512.mask.shuf.pd.128(<2 x double>, <2 x double>, i32, <2 x double>, i8)
 
 define <2 x double>@test_int_x86_avx512_mask_shuf_pd_128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
 ; CHECK-LABEL: test_int_x86_avx512_mask_shuf_pd_128:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    movzbl %dil, %eax
 ; CHECK-NEXT:    kmovw %eax, %k1
 ; CHECK-NEXT:    vshufpd $22, %xmm1, %xmm0, %xmm2 {%k1}

llvm/trunk/test/MC/X86/avx512-encodings.s

 // CHECK: vgetexpsd -1024(%rdx), %xmm7, %xmm2
 // CHECK:  encoding: [0x62,0xf2,0xc5,0x08,0x43,0x52,0x80]
           vgetexpsd -1024(%rdx), %xmm7, %xmm2
 
 // CHECK: vgetexpsd -1032(%rdx), %xmm7, %xmm2
 // CHECK:  encoding: [0x62,0xf2,0xc5,0x08,0x43,0x92,0xf8,0xfb,0xff,0xff]
           vgetexpsd -1032(%rdx), %xmm7, %xmm2
 
+// CHECK: vgetmantss $171, %xmm12, %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0x6d,0x08,0x27,0xdc,0xab]
+          vgetmantss $0xab, %xmm12, %xmm2, %xmm3
+
+// CHECK: vgetmantss $171, %xmm12, %xmm2, %xmm3 {%k7}
+// CHECK:  encoding: [0x62,0xd3,0x6d,0x0f,0x27,0xdc,0xab]
+          vgetmantss $0xab, %xmm12, %xmm2, %xmm3 {%k7}
+
+// CHECK: vgetmantss $171, %xmm12, %xmm2, %xmm3 {%k7} {z}
+// CHECK:  encoding: [0x62,0xd3,0x6d,0x8f,0x27,0xdc,0xab]
+          vgetmantss $0xab, %xmm12, %xmm2, %xmm3 {%k7} {z}
+
+// CHECK: vgetmantss $171,{sae}, %xmm12, %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0x6d,0x18,0x27,0xdc,0xab]
+          vgetmantss $0xab,{sae}, %xmm12, %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, %xmm12, %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0x6d,0x08,0x27,0xdc,0x7b]
+          vgetmantss $0x7b, %xmm12, %xmm2, %xmm3
+
+// CHECK: vgetmantss $123,{sae}, %xmm12, %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0x6d,0x18,0x27,0xdc,0x7b]
+          vgetmantss $0x7b,{sae}, %xmm12, %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, (%rcx), %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0x6d,0x08,0x27,0x19,0x7b]
+          vgetmantss $0x7b, (%rcx), %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, 291(%rax,%r14,8), %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xb3,0x6d,0x08,0x27,0x9c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantss $0x7b, 291(%rax,%r14,8), %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, 508(%rdx), %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0x6d,0x08,0x27,0x5a,0x7f,0x7b]
+          vgetmantss $0x7b, 508(%rdx), %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, 512(%rdx), %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0x6d,0x08,0x27,0x9a,0x00,0x02,0x00,0x00,0x7b]
+          vgetmantss $0x7b, 512(%rdx), %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, -512(%rdx), %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0x6d,0x08,0x27,0x5a,0x80,0x7b]
+          vgetmantss $0x7b, -512(%rdx), %xmm2, %xmm3
+
+// CHECK: vgetmantss $123, -516(%rdx), %xmm2, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0x6d,0x08,0x27,0x9a,0xfc,0xfd,0xff,0xff,0x7b]
+          vgetmantss $0x7b, -516(%rdx), %xmm2, %xmm3
+
+// CHECK: vgetmantsd $171, %xmm8, %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0xa5,0x00,0x27,0xd8,0xab]
+          vgetmantsd $0xab, %xmm8, %xmm27, %xmm3
+
+// CHECK: vgetmantsd $171, %xmm8, %xmm27, %xmm3 {%k6}
+// CHECK:  encoding: [0x62,0xd3,0xa5,0x06,0x27,0xd8,0xab]
+          vgetmantsd $0xab, %xmm8, %xmm27, %xmm3 {%k6}
+
+// CHECK: vgetmantsd $171, %xmm8, %xmm27, %xmm3 {%k6} {z}
+// CHECK:  encoding: [0x62,0xd3,0xa5,0x86,0x27,0xd8,0xab]
+          vgetmantsd $0xab, %xmm8, %xmm27, %xmm3 {%k6} {z}
+
+// CHECK: vgetmantsd $171,{sae}, %xmm8, %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0xa5,0x10,0x27,0xd8,0xab]
+          vgetmantsd $0xab,{sae}, %xmm8, %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, %xmm8, %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0xa5,0x00,0x27,0xd8,0x7b]
+          vgetmantsd $0x7b, %xmm8, %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123,{sae}, %xmm8, %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xd3,0xa5,0x10,0x27,0xd8,0x7b]
+          vgetmantsd $0x7b,{sae}, %xmm8, %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, (%rcx), %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0xa5,0x00,0x27,0x19,0x7b]
+          vgetmantsd $0x7b, (%rcx), %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, 291(%rax,%r14,8), %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xb3,0xa5,0x00,0x27,0x9c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantsd $0x7b, 291(%rax,%r14,8), %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, 1016(%rdx), %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0xa5,0x00,0x27,0x5a,0x7f,0x7b]
+          vgetmantsd $0x7b, 1016(%rdx), %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, 1024(%rdx), %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0xa5,0x00,0x27,0x9a,0x00,0x04,0x00,0x00,0x7b]
+          vgetmantsd $0x7b, 1024(%rdx), %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, -1024(%rdx), %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0xa5,0x00,0x27,0x5a,0x80,0x7b]
+          vgetmantsd $0x7b, -1024(%rdx), %xmm27, %xmm3
+
+// CHECK: vgetmantsd $123, -1032(%rdx), %xmm27, %xmm3
+// CHECK:  encoding: [0x62,0xf3,0xa5,0x00,0x27,0x9a,0xf8,0xfb,0xff,0xff,0x7b]
+          vgetmantsd $0x7b, -1032(%rdx), %xmm27, %xmm3
+
+// CHECK: vgetmantps $171, %zmm28, %zmm22
+// CHECK:  encoding: [0x62,0x83,0x7d,0x48,0x26,0xf4,0xab]
+          vgetmantps $0xab, %zmm28, %zmm22
+
+// CHECK: vgetmantps $171, %zmm28, %zmm22 {%k3}
+// CHECK:  encoding: [0x62,0x83,0x7d,0x4b,0x26,0xf4,0xab]
+          vgetmantps $0xab, %zmm28, %zmm22 {%k3}
+
+// CHECK: vgetmantps $171, %zmm28, %zmm22 {%k3} {z}
+// CHECK:  encoding: [0x62,0x83,0x7d,0xcb,0x26,0xf4,0xab]
+          vgetmantps $0xab, %zmm28, %zmm22 {%k3} {z}
+
+// CHECK: vgetmantps $171,{sae}, %zmm28, %zmm22
+// CHECK:  encoding: [0x62,0x83,0x7d,0x18,0x26,0xf4,0xab]
+          vgetmantps $0xab,{sae}, %zmm28, %zmm22
+
+// CHECK: vgetmantps $123, %zmm28, %zmm22
+// CHECK:  encoding: [0x62,0x83,0x7d,0x48,0x26,0xf4,0x7b]
+          vgetmantps $0x7b, %zmm28, %zmm22
+
+// CHECK: vgetmantps $123,{sae}, %zmm28, %zmm22
+// CHECK:  encoding: [0x62,0x83,0x7d,0x18,0x26,0xf4,0x7b]
+          vgetmantps $0x7b,{sae}, %zmm28, %zmm22
+
+// CHECK: vgetmantps $123, (%rcx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x26,0x31,0x7b]
+          vgetmantps $0x7b, (%rcx), %zmm22
+
+// CHECK: vgetmantps $123, 291(%rax,%r14,8), %zmm22
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x48,0x26,0xb4,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 291(%rax,%r14,8), %zmm22
+
+// CHECK: vgetmantps $123, (%rcx){1to16}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x26,0x31,0x7b]
+          vgetmantps $0x7b, (%rcx){1to16}, %zmm22
+
+// CHECK: vgetmantps $123, 8128(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x26,0x72,0x7f,0x7b]
+          vgetmantps $0x7b, 8128(%rdx), %zmm22
+
+// CHECK: vgetmantps $123, 8192(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x26,0xb2,0x00,0x20,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 8192(%rdx), %zmm22
+
+// CHECK: vgetmantps $123, -8192(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x26,0x72,0x80,0x7b]
+          vgetmantps $0x7b, -8192(%rdx), %zmm22
+
+// CHECK: vgetmantps $123, -8256(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x26,0xb2,0xc0,0xdf,0xff,0xff,0x7b]
+          vgetmantps $0x7b, -8256(%rdx), %zmm22
+
+// CHECK: vgetmantps $123, 508(%rdx){1to16}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x26,0x72,0x7f,0x7b]
+          vgetmantps $0x7b, 508(%rdx){1to16}, %zmm22
+
+// CHECK: vgetmantps $123, 512(%rdx){1to16}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x26,0xb2,0x00,0x02,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 512(%rdx){1to16}, %zmm22
+
+// CHECK: vgetmantps $123, -512(%rdx){1to16}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x26,0x72,0x80,0x7b]
+          vgetmantps $0x7b, -512(%rdx){1to16}, %zmm22
+
+// CHECK: vgetmantps $123, -516(%rdx){1to16}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x26,0xb2,0xfc,0xfd,0xff,0xff,0x7b]
+          vgetmantps $0x7b, -516(%rdx){1to16}, %zmm22
+
+// CHECK: vgetmantpd $171, %zmm26, %zmm2
+// CHECK:  encoding: [0x62,0x93,0xfd,0x48,0x26,0xd2,0xab]
+          vgetmantpd $0xab, %zmm26, %zmm2
+
+// CHECK: vgetmantpd $171, %zmm26, %zmm2 {%k7}
+// CHECK:  encoding: [0x62,0x93,0xfd,0x4f,0x26,0xd2,0xab]
+          vgetmantpd $0xab, %zmm26, %zmm2 {%k7}
+
+// CHECK: vgetmantpd $171, %zmm26, %zmm2 {%k7} {z}
+// CHECK:  encoding: [0x62,0x93,0xfd,0xcf,0x26,0xd2,0xab]
+          vgetmantpd $0xab, %zmm26, %zmm2 {%k7} {z}
+
+// CHECK: vgetmantpd $171,{sae}, %zmm26, %zmm2
+// CHECK:  encoding: [0x62,0x93,0xfd,0x18,0x26,0xd2,0xab]
+          vgetmantpd $0xab,{sae}, %zmm26, %zmm2
+
+// CHECK: vgetmantpd $123, %zmm26, %zmm2
+// CHECK:  encoding: [0x62,0x93,0xfd,0x48,0x26,0xd2,0x7b]
+          vgetmantpd $0x7b, %zmm26, %zmm2
+
+// CHECK: vgetmantpd $123,{sae}, %zmm26, %zmm2
+// CHECK:  encoding: [0x62,0x93,0xfd,0x18,0x26,0xd2,0x7b]
+          vgetmantpd $0x7b,{sae}, %zmm26, %zmm2
+
+// CHECK: vgetmantpd $123, (%rcx), %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x48,0x26,0x11,0x7b]
+          vgetmantpd $0x7b, (%rcx), %zmm2
+
+// CHECK: vgetmantpd $123, 291(%rax,%r14,8), %zmm2
+// CHECK:  encoding: [0x62,0xb3,0xfd,0x48,0x26,0x94,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 291(%rax,%r14,8), %zmm2
+
+// CHECK: vgetmantpd $123, (%rcx){1to8}, %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x58,0x26,0x11,0x7b]
+          vgetmantpd $0x7b, (%rcx){1to8}, %zmm2
+
+// CHECK: vgetmantpd $123, 8128(%rdx), %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x48,0x26,0x52,0x7f,0x7b]
+          vgetmantpd $0x7b, 8128(%rdx), %zmm2
+
+// CHECK: vgetmantpd $123, 8192(%rdx), %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x48,0x26,0x92,0x00,0x20,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 8192(%rdx), %zmm2
+
+// CHECK: vgetmantpd $123, -8192(%rdx), %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x48,0x26,0x52,0x80,0x7b]
+          vgetmantpd $0x7b, -8192(%rdx), %zmm2
+
+// CHECK: vgetmantpd $123, -8256(%rdx), %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x48,0x26,0x92,0xc0,0xdf,0xff,0xff,0x7b]
+          vgetmantpd $0x7b, -8256(%rdx), %zmm2
+
+// CHECK: vgetmantpd $123, 1016(%rdx){1to8}, %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x58,0x26,0x52,0x7f,0x7b]
+          vgetmantpd $0x7b, 1016(%rdx){1to8}, %zmm2
+
+// CHECK: vgetmantpd $123, 1024(%rdx){1to8}, %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x58,0x26,0x92,0x00,0x04,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 1024(%rdx){1to8}, %zmm2
+
+// CHECK: vgetmantpd $123, -1024(%rdx){1to8}, %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x58,0x26,0x52,0x80,0x7b]
+          vgetmantpd $0x7b, -1024(%rdx){1to8}, %zmm2
+
+// CHECK: vgetmantpd $123, -1032(%rdx){1to8}, %zmm2
+// CHECK:  encoding: [0x62,0xf3,0xfd,0x58,0x26,0x92,0xf8,0xfb,0xff,0xff,0x7b]
+          vgetmantpd $0x7b, -1032(%rdx){1to8}, %zmm2
+
 // CHECK: vshufps $171, %zmm9, %zmm6, %zmm5
 // CHECK:  encoding: [0x62,0xd1,0x4c,0x48,0xc6,0xe9,0xab]
           vshufps $0xab, %zmm9, %zmm6, %zmm5
 
 // CHECK: vshufps $171, %zmm9, %zmm6, %zmm5 {%k6}
 // CHECK:  encoding: [0x62,0xd1,0x4c,0x4e,0xc6,0xe9,0xab]
           vshufps $0xab, %zmm9, %zmm6, %zmm5 {%k6}
 

llvm/trunk/test/MC/X86/x86-64-avx512f_vl.s

 // CHECK: vpunpckhqdq -1024(%rdx){1to4}, %ymm19, %ymm20
 // CHECK:  encoding: [0x62,0xe1,0xe5,0x30,0x6d,0x62,0x80]
           vpunpckhqdq -1024(%rdx){1to4}, %ymm19, %ymm20
 
 // CHECK: vpunpckhqdq -1032(%rdx){1to4}, %ymm19, %ymm20
 // CHECK:  encoding: [0x62,0xe1,0xe5,0x30,0x6d,0xa2,0xf8,0xfb,0xff,0xff]
           vpunpckhqdq -1032(%rdx){1to4}, %ymm19, %ymm20
 
+// CHECK: vgetmantps $171, %xmm23, %xmm29
+// CHECK:  encoding: [0x62,0x23,0x7d,0x08,0x26,0xef,0xab]
+          vgetmantps $0xab, %xmm23, %xmm29
+
+// CHECK: vgetmantps $171, %xmm23, %xmm29 {%k5}
+// CHECK:  encoding: [0x62,0x23,0x7d,0x0d,0x26,0xef,0xab]
+          vgetmantps $0xab, %xmm23, %xmm29 {%k5}
+
+// CHECK: vgetmantps $171, %xmm23, %xmm29 {%k5} {z}
+// CHECK:  encoding: [0x62,0x23,0x7d,0x8d,0x26,0xef,0xab]
+          vgetmantps $0xab, %xmm23, %xmm29 {%k5} {z}
+
+// CHECK: vgetmantps $123, %xmm23, %xmm29
+// CHECK:  encoding: [0x62,0x23,0x7d,0x08,0x26,0xef,0x7b]
+          vgetmantps $0x7b, %xmm23, %xmm29
+
+// CHECK: vgetmantps $123, (%rcx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x26,0x29,0x7b]
+          vgetmantps $0x7b, (%rcx), %xmm29
+
+// CHECK: vgetmantps $123, 291(%rax,%r14,8), %xmm29
+// CHECK:  encoding: [0x62,0x23,0x7d,0x08,0x26,0xac,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 291(%rax,%r14,8), %xmm29
+
+// CHECK: vgetmantps $123, (%rcx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x26,0x29,0x7b]
+          vgetmantps $0x7b, (%rcx){1to4}, %xmm29
+
+// CHECK: vgetmantps $123, 2032(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x26,0x6a,0x7f,0x7b]
+          vgetmantps $0x7b, 2032(%rdx), %xmm29
+
+// CHECK: vgetmantps $123, 2048(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x26,0xaa,0x00,0x08,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 2048(%rdx), %xmm29
+
+// CHECK: vgetmantps $123, -2048(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x26,0x6a,0x80,0x7b]
+          vgetmantps $0x7b, -2048(%rdx), %xmm29
+
+// CHECK: vgetmantps $123, -2064(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x26,0xaa,0xf0,0xf7,0xff,0xff,0x7b]
+          vgetmantps $0x7b, -2064(%rdx), %xmm29
+
+// CHECK: vgetmantps $123, 508(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x26,0x6a,0x7f,0x7b]
+          vgetmantps $0x7b, 508(%rdx){1to4}, %xmm29
+
+// CHECK: vgetmantps $123, 512(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x26,0xaa,0x00,0x02,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 512(%rdx){1to4}, %xmm29
+
+// CHECK: vgetmantps $123, -512(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x26,0x6a,0x80,0x7b]
+          vgetmantps $0x7b, -512(%rdx){1to4}, %xmm29
+
+// CHECK: vgetmantps $123, -516(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x26,0xaa,0xfc,0xfd,0xff,0xff,0x7b]
+          vgetmantps $0x7b, -516(%rdx){1to4}, %xmm29
+
+// CHECK: vgetmantps $171, %ymm29, %ymm17
+// CHECK:  encoding: [0x62,0x83,0x7d,0x28,0x26,0xcd,0xab]
+          vgetmantps $0xab, %ymm29, %ymm17
+
+// CHECK: vgetmantps $171, %ymm29, %ymm17 {%k1}
+// CHECK:  encoding: [0x62,0x83,0x7d,0x29,0x26,0xcd,0xab]
+          vgetmantps $0xab, %ymm29, %ymm17 {%k1}
+
+// CHECK: vgetmantps $171, %ymm29, %ymm17 {%k1} {z}
+// CHECK:  encoding: [0x62,0x83,0x7d,0xa9,0x26,0xcd,0xab]
+          vgetmantps $0xab, %ymm29, %ymm17 {%k1} {z}
+
+// CHECK: vgetmantps $123, %ymm29, %ymm17
+// CHECK:  encoding: [0x62,0x83,0x7d,0x28,0x26,0xcd,0x7b]
+          vgetmantps $0x7b, %ymm29, %ymm17
+
+// CHECK: vgetmantps $123, (%rcx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x26,0x09,0x7b]
+          vgetmantps $0x7b, (%rcx), %ymm17
+
+// CHECK: vgetmantps $123, 291(%rax,%r14,8), %ymm17
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x28,0x26,0x8c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 291(%rax,%r14,8), %ymm17
+
+// CHECK: vgetmantps $123, (%rcx){1to8}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x26,0x09,0x7b]
+          vgetmantps $0x7b, (%rcx){1to8}, %ymm17
+
+// CHECK: vgetmantps $123, 4064(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x26,0x4a,0x7f,0x7b]
+          vgetmantps $0x7b, 4064(%rdx), %ymm17
+
+// CHECK: vgetmantps $123, 4096(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x26,0x8a,0x00,0x10,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 4096(%rdx), %ymm17
+
+// CHECK: vgetmantps $123, -4096(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x26,0x4a,0x80,0x7b]
+          vgetmantps $0x7b, -4096(%rdx), %ymm17
+
+// CHECK: vgetmantps $123, -4128(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x26,0x8a,0xe0,0xef,0xff,0xff,0x7b]
+          vgetmantps $0x7b, -4128(%rdx), %ymm17
+
+// CHECK: vgetmantps $123, 508(%rdx){1to8}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x26,0x4a,0x7f,0x7b]
+          vgetmantps $0x7b, 508(%rdx){1to8}, %ymm17
+
+// CHECK: vgetmantps $123, 512(%rdx){1to8}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x26,0x8a,0x00,0x02,0x00,0x00,0x7b]
+          vgetmantps $0x7b, 512(%rdx){1to8}, %ymm17
+
+// CHECK: vgetmantps $123, -512(%rdx){1to8}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x26,0x4a,0x80,0x7b]
+          vgetmantps $0x7b, -512(%rdx){1to8}, %ymm17
+
+// CHECK: vgetmantps $123, -516(%rdx){1to8}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x26,0x8a,0xfc,0xfd,0xff,0xff,0x7b]
+          vgetmantps $0x7b, -516(%rdx){1to8}, %ymm17
+
+// CHECK: vgetmantpd $171, %xmm29, %xmm28
+// CHECK:  encoding: [0x62,0x03,0xfd,0x08,0x26,0xe5,0xab]
+          vgetmantpd $0xab, %xmm29, %xmm28
+
+// CHECK: vgetmantpd $171, %xmm29, %xmm28 {%k4}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x0c,0x26,0xe5,0xab]
+          vgetmantpd $0xab, %xmm29, %xmm28 {%k4}
+
+// CHECK: vgetmantpd $171, %xmm29, %xmm28 {%k4} {z}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x8c,0x26,0xe5,0xab]
+          vgetmantpd $0xab, %xmm29, %xmm28 {%k4} {z}
+
+// CHECK: vgetmantpd $123, %xmm29, %xmm28
+// CHECK:  encoding: [0x62,0x03,0xfd,0x08,0x26,0xe5,0x7b]
+          vgetmantpd $0x7b, %xmm29, %xmm28
+
+// CHECK: vgetmantpd $123, (%rcx), %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x26,0x21,0x7b]
+          vgetmantpd $0x7b, (%rcx), %xmm28
+
+// CHECK: vgetmantpd $123, 291(%rax,%r14,8), %xmm28
+// CHECK:  encoding: [0x62,0x23,0xfd,0x08,0x26,0xa4,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 291(%rax,%r14,8), %xmm28
+
+// CHECK: vgetmantpd $123, (%rcx){1to2}, %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x26,0x21,0x7b]
+          vgetmantpd $0x7b, (%rcx){1to2}, %xmm28
+
+// CHECK: vgetmantpd $123, 2032(%rdx), %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x26,0x62,0x7f,0x7b]
+          vgetmantpd $0x7b, 2032(%rdx), %xmm28
+
+// CHECK: vgetmantpd $123, 2048(%rdx), %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x26,0xa2,0x00,0x08,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 2048(%rdx), %xmm28
+
+// CHECK: vgetmantpd $123, -2048(%rdx), %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x26,0x62,0x80,0x7b]
+          vgetmantpd $0x7b, -2048(%rdx), %xmm28
+
+// CHECK: vgetmantpd $123, -2064(%rdx), %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x26,0xa2,0xf0,0xf7,0xff,0xff,0x7b]
+          vgetmantpd $0x7b, -2064(%rdx), %xmm28
+
+// CHECK: vgetmantpd $123, 1016(%rdx){1to2}, %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x26,0x62,0x7f,0x7b]
+          vgetmantpd $0x7b, 1016(%rdx){1to2}, %xmm28
+
+// CHECK: vgetmantpd $123, 1024(%rdx){1to2}, %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x26,0xa2,0x00,0x04,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 1024(%rdx){1to2}, %xmm28
+
+// CHECK: vgetmantpd $123, -1024(%rdx){1to2}, %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x26,0x62,0x80,0x7b]
+          vgetmantpd $0x7b, -1024(%rdx){1to2}, %xmm28
+
+// CHECK: vgetmantpd $123, -1032(%rdx){1to2}, %xmm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x26,0xa2,0xf8,0xfb,0xff,0xff,0x7b]
+          vgetmantpd $0x7b, -1032(%rdx){1to2}, %xmm28
+
+// CHECK: vgetmantpd $171, %ymm23, %ymm23
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x28,0x26,0xff,0xab]
+          vgetmantpd $0xab, %ymm23, %ymm23
+
+// CHECK: vgetmantpd $171, %ymm23, %ymm23 {%k5}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x2d,0x26,0xff,0xab]
+          vgetmantpd $0xab, %ymm23, %ymm23 {%k5}
+
+// CHECK: vgetmantpd $171, %ymm23, %ymm23 {%k5} {z}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0xad,0x26,0xff,0xab]
+          vgetmantpd $0xab, %ymm23, %ymm23 {%k5} {z}
+
+// CHECK: vgetmantpd $123, %ymm23, %ymm23
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x28,0x26,0xff,0x7b]
+          vgetmantpd $0x7b, %ymm23, %ymm23
+
+// CHECK: vgetmantpd $123, (%rcx), %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x26,0x39,0x7b]
+          vgetmantpd $0x7b, (%rcx), %ymm23
+
+// CHECK: vgetmantpd $123, 291(%rax,%r14,8), %ymm23
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x28,0x26,0xbc,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 291(%rax,%r14,8), %ymm23
+
+// CHECK: vgetmantpd $123, (%rcx){1to4}, %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x26,0x39,0x7b]
+          vgetmantpd $0x7b, (%rcx){1to4}, %ymm23
+
+// CHECK: vgetmantpd $123, 4064(%rdx), %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x26,0x7a,0x7f,0x7b]
+          vgetmantpd $0x7b, 4064(%rdx), %ymm23
+
+// CHECK: vgetmantpd $123, 4096(%rdx), %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x26,0xba,0x00,0x10,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 4096(%rdx), %ymm23
+
+// CHECK: vgetmantpd $123, -4096(%rdx), %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x26,0x7a,0x80,0x7b]
+          vgetmantpd $0x7b, -4096(%rdx), %ymm23
+
+// CHECK: vgetmantpd $123, -4128(%rdx), %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x26,0xba,0xe0,0xef,0xff,0xff,0x7b]
+          vgetmantpd $0x7b, -4128(%rdx), %ymm23
+
+// CHECK: vgetmantpd $123, 1016(%rdx){1to4}, %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x26,0x7a,0x7f,0x7b]
+          vgetmantpd $0x7b, 1016(%rdx){1to4}, %ymm23
+
+// CHECK: vgetmantpd $123, 1024(%rdx){1to4}, %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x26,0xba,0x00,0x04,0x00,0x00,0x7b]
+          vgetmantpd $0x7b, 1024(%rdx){1to4}, %ymm23
+
+// CHECK: vgetmantpd $123, -1024(%rdx){1to4}, %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x26,0x7a,0x80,0x7b]
+          vgetmantpd $0x7b, -1024(%rdx){1to4}, %ymm23
+
+// CHECK: vgetmantpd $123, -1032(%rdx){1to4}, %ymm23
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x26,0xba,0xf8,0xfb,0xff,0xff,0x7b]
+          vgetmantpd $0x7b, -1032(%rdx){1to4}, %ymm23
+
 // CHECK: vshufps $171, %xmm21, %xmm23, %xmm17
 // CHECK:  encoding: [0x62,0xa1,0x44,0x00,0xc6,0xcd,0xab]
           vshufps $0xab, %xmm21, %xmm23, %xmm17
 
 // CHECK: vshufps $171, %xmm21, %xmm23, %xmm17 {%k3}
 // CHECK:  encoding: [0x62,0xa1,0x44,0x03,0xc6,0xcd,0xab]
           vshufps $0xab, %xmm21, %xmm23, %xmm17 {%k3}