[X86][AVX512] add support in reduce/range/scalef/rndScale

[AsafBadouh]

Summary

include intrinsics and encoding tests

[AsafBadouh]

remove scalar version of reduce and rndscale nodes

[delena]

LGTM

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

def int_x86_avx512_mask_cvtuqq2ps_256 :
           [IntrNoMem]>;
 
   def int_x86_avx512_mask_cvtuqq2ps_512 : 
         GCCBuiltin<"__builtin_ia32_cvtuqq2ps512_mask">,
           Intrinsic<[llvm_v8f32_ty],
           [llvm_v8i64_ty, llvm_v8f32_ty,  llvm_i8_ty,  llvm_i32_ty],
           [IntrNoMem]>;
 
-  def int_x86_avx512_mask_rndscale_ps_512: GCCBuiltin<"__builtin_ia32_rndscaleps_mask">,
+  def int_x86_avx512_mask_rndscale_pd_128 : GCCBuiltin<"__builtin_ia32_rndscalepd_128_mask">,
+        Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_i32_ty,
+                                     llvm_v2f64_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_rndscale_pd_256 : GCCBuiltin<"__builtin_ia32_rndscalepd_256_mask">,
+        Intrinsic<[llvm_v4f64_ty], [llvm_v4f64_ty, llvm_i32_ty, 
+                                     llvm_v4f64_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_rndscale_pd_512 : GCCBuiltin<"__builtin_ia32_rndscalepd_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_rndscale_ps_128 : GCCBuiltin<"__builtin_ia32_rndscaleps_128_mask">,
+        Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_i32_ty,
+                                     llvm_v4f32_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_rndscale_ps_256 : GCCBuiltin<"__builtin_ia32_rndscaleps_256_mask">,
+        Intrinsic<[llvm_v8f32_ty], [llvm_v8f32_ty, llvm_i32_ty, 
+                                     llvm_v8f32_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_rndscale_ps_512 : GCCBuiltin<"__builtin_ia32_rndscaleps_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_rndscale_pd_512: GCCBuiltin<"__builtin_ia32_rndscalepd_mask">,
+  def int_x86_avx512_mask_reduce_pd_128 : GCCBuiltin<"__builtin_ia32_reducepd128_mask">,
+        Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_i32_ty, 
+                                     llvm_v2f64_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_reduce_pd_256 : GCCBuiltin<"__builtin_ia32_reducepd256_mask">,
+        Intrinsic<[llvm_v4f64_ty], [llvm_v4f64_ty, llvm_i32_ty,
+                                     llvm_v4f64_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_reduce_pd_512 : GCCBuiltin<"__builtin_ia32_reducepd512_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_reduce_ps_128 : GCCBuiltin<"__builtin_ia32_reduceps128_mask">,
+        Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_i32_ty, 
+                                     llvm_v4f32_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_reduce_ps_256 : GCCBuiltin<"__builtin_ia32_reduceps256_mask">,
+        Intrinsic<[llvm_v8f32_ty], [llvm_v8f32_ty, llvm_i32_ty,
+                                     llvm_v8f32_ty,  llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_reduce_ps_512 : GCCBuiltin<"__builtin_ia32_reduceps512_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_range_pd_128 : GCCBuiltin<"__builtin_ia32_rangepd128_mask">,
+        Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, llvm_i32_ty, 
+                                    llvm_v2f64_ty,  llvm_i8_ty], [IntrNoMem]>;
+def int_x86_avx512_mask_range_pd_256 : GCCBuiltin<"__builtin_ia32_rangepd256_mask">,
+        Intrinsic<[llvm_v4f64_ty], [llvm_v4f64_ty, llvm_v4f64_ty, llvm_i32_ty,
+                                    llvm_v4f64_ty,  llvm_i8_ty], [IntrNoMem]>;
+def int_x86_avx512_mask_range_pd_512 : GCCBuiltin<"__builtin_ia32_rangepd512_mask">,
+        Intrinsic<[llvm_v8f64_ty], [llvm_v8f64_ty, llvm_v8f64_ty, llvm_i32_ty,
+                                    llvm_v8f64_ty,  llvm_i8_ty,  llvm_i32_ty], [IntrNoMem]>;
+def int_x86_avx512_mask_range_ps_128 : GCCBuiltin<"__builtin_ia32_rangeps128_mask">,
+        Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_v4f32_ty, llvm_i32_ty,
+                                    llvm_v4f32_ty,  llvm_i8_ty], [IntrNoMem]>;
+def int_x86_avx512_mask_range_ps_256 : GCCBuiltin<"__builtin_ia32_rangeps256_mask">,
+        Intrinsic<[llvm_v8f32_ty], [llvm_v8f32_ty, llvm_v8f32_ty, llvm_i32_ty, 
+                                    llvm_v8f32_ty,  llvm_i8_ty], [IntrNoMem]>;
+def int_x86_avx512_mask_range_ps_512 : GCCBuiltin<"__builtin_ia32_rangeps512_mask">,
+        Intrinsic<[llvm_v16f32_ty], [llvm_v16f32_ty, llvm_v16f32_ty, llvm_i32_ty,
+                                     llvm_v16f32_ty,  llvm_i16_ty,  llvm_i32_ty], [IntrNoMem]>;
 }
 
 // Vector load with broadcast
 let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_avx512_vbroadcast_ss_512 :
         GCCBuiltin<"__builtin_ia32_vbroadcastss512">,
         Intrinsic<[llvm_v16f32_ty], [llvm_ptr_ty], [IntrReadArgMem]>;
   def int_x86_avx512_vbroadcast_ss_ps_512 :
let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_avx512_mask_rndscale_ss : GCCBuiltin<"__builtin_ia32_rndscaless_mask">,
           Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty,
                                      llvm_i8_ty, llvm_i32_ty, llvm_i32_ty],
                                      [IntrNoMem]>;
   def int_x86_avx512_mask_rndscale_sd : GCCBuiltin<"__builtin_ia32_rndscalesd_mask">,
           Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty,
                                       llvm_i8_ty, llvm_i32_ty, llvm_i32_ty],
                                      [IntrNoMem]>;
-
+  def int_x86_avx512_mask_range_ss : GCCBuiltin<"__builtin_ia32_rangess_mask">,
+          Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty,
+                                     llvm_i8_ty, llvm_i32_ty, llvm_i32_ty],
+                                     [IntrNoMem]>;
+  def int_x86_avx512_mask_range_sd : GCCBuiltin<"__builtin_ia32_rangesd_mask">,
+          Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty,
+                                      llvm_i8_ty, llvm_i32_ty, llvm_i32_ty],
+                                     [IntrNoMem]>;
+  def int_x86_avx512_mask_reduce_ss : GCCBuiltin<"__builtin_ia32_reducess_mask">,
+          Intrinsic<[llvm_v4f32_ty], [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty,
+                                     llvm_i8_ty, llvm_i32_ty, llvm_i32_ty],
+                                     [IntrNoMem]>;
+  def int_x86_avx512_mask_reduce_sd : GCCBuiltin<"__builtin_ia32_reducesd_mask">,
+          Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty,
+                                      llvm_i8_ty, llvm_i32_ty, llvm_i32_ty],
+                                     [IntrNoMem]>;
+  def int_x86_avx512_mask_scalef_sd : GCCBuiltin<"__builtin_ia32_scalefsd_round">,
+          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_scalef_ss : GCCBuiltin<"__builtin_ia32_scalefss_round">,
+          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_scalef_pd_128 : GCCBuiltin<"__builtin_ia32_scalefpd128_mask">,
           Intrinsic<[llvm_v2f64_ty], [llvm_v2f64_ty, llvm_v2f64_ty, 
                     llvm_v2f64_ty, llvm_i8_ty], [IntrNoMem]>;
   def int_x86_avx512_mask_scalef_pd_256 : GCCBuiltin<"__builtin_ia32_scalefpd256_mask">,
           Intrinsic<[llvm_v4f64_ty], [llvm_v4f64_ty, llvm_v4f64_ty, 
                     llvm_v4f64_ty, llvm_i8_ty],[IntrNoMem]>;
   def int_x86_avx512_mask_scalef_pd_512 : GCCBuiltin<"__builtin_ia32_scalefpd512_mask">,
           Intrinsic<[llvm_v8f64_ty], [llvm_v8f64_ty, llvm_v8f64_ty, 

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

enum NodeType : unsigned {
       VPERMV3,
       VPERMIV3,
       VPERMI,
       VPERM2X128,
       //Fix Up Special Packed Float32/64 values
       VFIXUPIMM,
       //Range Restriction Calculation For Packed Pairs of Float32/64 values
       VRANGE,
+      // Reduce - Perform Reduction Transformation on scalar\packed FP
+      VREDUCE,
+      // RndScale - Round FP Values To Include A Given Number Of Fraction Bits
+      VRNDSCALE,
       // Broadcast scalar to vector
       VBROADCAST,
       // Broadcast subvector to vector
       SUBV_BROADCAST,
       // Insert/Extract vector element
       VINSERT,
       VEXTRACT,
 
enum NodeType : unsigned {
       FMSUBADD,
       // FMA with rounding mode
       FMADD_RND,
       FNMADD_RND,
       FMSUB_RND,
       FNMSUB_RND,
       FMADDSUB_RND,
       FMSUBADD_RND,
-      RNDSCALE,
 
       // Compress and expand
       COMPRESS,
       EXPAND,
 
       //Convert Unsigned/Integer to Scalar Floating-Point Value
       //with rounding mode
       SINT_TO_FP_RND,

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_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),
+      // if not, we set rounding mode to "current".
+      SDValue Rnd;
+      if (Op.getNumOperands() == 7)
+        Rnd = Op.getOperand(6);
+      else
+        Rnd = DAG.getConstant(X86::STATIC_ROUNDING::CUR_DIRECTION, dl, MVT::i32);
+      return getVectorMaskingNode(DAG.getNode(IntrData->Opc0, dl, VT,
+        Src1, Src2, Imm, Rnd),
+        Mask, PassThru, Subtarget, DAG);
+    }
     case INTR_TYPE_3OP_MASK: {
       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);
       // We specify 2 possible opcodes for intrinsics with rounding modes.
       // First, we check if the intrinsic may have non-default rounding mode,
const char *X86TargetLowering::getTargetNodeName(unsigned Opcode) const {
   case X86ISD::FMADDSUB:           return "X86ISD::FMADDSUB";
   case X86ISD::FMSUBADD:           return "X86ISD::FMSUBADD";
   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::RNDSCALE:           return "X86ISD::RNDSCALE";
+  case X86ISD::VRNDSCALE:          return "X86ISD::VRNDSCALE";
+  case X86ISD::VREDUCE:            return "X86ISD::VREDUCE";
   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

def rm : I< opc, MRMSrcMem, (outs _.FRC:$dst),
                          (ins _.FRC:$src1, _.ScalarMemOp:$src2), 
                          OpcodeStr#"\t{$src2, $src1, $dst|$dst, $src1, $src2}",
                          [(set _.FRC:$dst, (OpNode _.FRC:$src1,
                          (_.ScalarLdFrag addr:$src2)))], itins.rr>;
   }
 }
 
 multiclass avx512_fp_scalar_round<bits<8> opc, string OpcodeStr,X86VectorVTInfo _,
-                         SDNode VecNode, OpndItins itins, bit IsCommutable> {
+                         SDNode VecNode, OpndItins itins, bit IsCommutable = 0> {
 
   defm rrb : AVX512_maskable_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
                           (ins _.RC:$src1, _.RC:$src2, AVX512RC:$rc), OpcodeStr,
                           "$rc, $src2, $src1", "$src1, $src2, $rc",
                           (VecNode (_.VT _.RC:$src1), (_.VT _.RC:$src2),
                           (i32 imm:$rc)), itins.rr, IsCommutable>,
                           EVEX_B, EVEX_RC;
 }
defm rmb: AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
                      "${src2}"##_.BroadcastStr##", $src1",
                      "$src1, ${src2}"##_.BroadcastStr,
                      (OpNode  _.RC:$src1, (_.VT (X86VBroadcast
                                                 (_.ScalarLdFrag addr:$src2))), (i32 FROUND_CURRENT))>,
                      EVEX_4V, EVEX_B;
   }//let mayLoad = 1
 }
 
-multiclass avx512_fp_scalef_all<bits<8> opc, string OpcodeStr, SDNode OpNode> {
+multiclass avx512_fp_scalef_scalar<bits<8> opc, string OpcodeStr, SDNode OpNode,
+                            X86VectorVTInfo _> {
+  defm rr: AVX512_maskable_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
+                  (ins _.RC:$src1, _.RC:$src2), OpcodeStr##_.Suffix,
+                  "$src2, $src1", "$src1, $src2",
+                  (_.VT (OpNode _.RC:$src1, _.RC:$src2, (i32 FROUND_CURRENT)))>;
+  let mayLoad = 1 in {
+    defm rm: AVX512_maskable_scalar<opc, MRMSrcMem, _, (outs _.RC:$dst),
+                    (ins _.RC:$src1, _.MemOp:$src2), OpcodeStr##_.Suffix,
+                    "$src2, $src1", "$src1, $src2",
+                    (OpNode _.RC:$src1, (_.LdFrag addr:$src2), (i32 FROUND_CURRENT))>;
+  }//let mayLoad = 1
+}
+
+multiclass avx512_fp_scalef_all<bits<8> opc, bits<8> opcScaler, string OpcodeStr, SDNode OpNode> {
   defm PSZ : avx512_fp_scalef_p<opc, OpcodeStr, OpNode, v16f32_info>, 
              avx512_fp_round_packed<opc, OpcodeStr, OpNode, v16f32_info>,
                               EVEX_V512, EVEX_CD8<32, CD8VF>;
   defm PDZ : avx512_fp_scalef_p<opc, OpcodeStr, OpNode, v8f64_info>, 
              avx512_fp_round_packed<opc, OpcodeStr, OpNode, v8f64_info>,
                               EVEX_V512, VEX_W, EVEX_CD8<64, CD8VF>;
+  defm SSZ128 : avx512_fp_scalef_scalar<opcScaler, OpcodeStr, OpNode, f32x_info>,
+                avx512_fp_scalar_round<opcScaler, OpcodeStr##"ss", f32x_info, OpNode, SSE_ALU_ITINS_S.s>,
+                              EVEX_4V,EVEX_CD8<32, CD8VT1>;
+  defm SDZ128 : avx512_fp_scalef_scalar<opcScaler, OpcodeStr, OpNode, f64x_info>,
+                avx512_fp_scalar_round<opcScaler, OpcodeStr##"sd", f64x_info, OpNode, SSE_ALU_ITINS_S.d>,
+                              EVEX_4V, EVEX_CD8<64, CD8VT1>, VEX_W;
+
   // Define only if AVX512VL feature is present.
   let Predicates = [HasVLX] in {
     defm PSZ128 : avx512_fp_scalef_p<opc, OpcodeStr, OpNode, v4f32x_info>,
                                    EVEX_V128, EVEX_CD8<32, CD8VF>;
     defm PSZ256 : avx512_fp_scalef_p<opc, OpcodeStr, OpNode, v8f32x_info>,
                                    EVEX_V256, EVEX_CD8<32, CD8VF>;
     defm PDZ128 : avx512_fp_scalef_p<opc, OpcodeStr, OpNode, v2f64x_info>,
                                    EVEX_V128, VEX_W, EVEX_CD8<64, CD8VF>;
     defm PDZ256 : avx512_fp_scalef_p<opc, OpcodeStr, OpNode, v4f64x_info>,
                                    EVEX_V256, VEX_W, EVEX_CD8<64, CD8VF>;
   }
 }
-defm VSCALEF : avx512_fp_scalef_all<0x2C, "vscalef", X86scalef>, T8PD;
+defm VSCALEF : avx512_fp_scalef_all<0x2C, 0x2D, "vscalef", X86scalef>, T8PD;
 
 //===----------------------------------------------------------------------===//
 // AVX-512  VPTESTM instructions
 //===----------------------------------------------------------------------===//
 
 multiclass avx512_vptest<bits<8> opc, string OpcodeStr, SDNode OpNode,
                             X86VectorVTInfo _> {
   defm rr : AVX512_maskable_cmp<opc, MRMSrcReg, _, (outs _.KRC:$dst),
let Predicates = [HasAVX512] in {
   def : Pat<(int_x86_sse2_sqrt_sd VR128X:$src),
             (COPY_TO_REGCLASS (VSQRTSDZr (f64 (IMPLICIT_DEF)),
                                         (COPY_TO_REGCLASS VR128X:$src, FR64)),
                               VR128X)>;
   def : Pat<(int_x86_sse2_sqrt_sd sse_load_f64:$src),
             (VSQRTSDZm_Int (v2f64 (IMPLICIT_DEF)), sse_load_f64:$src)>;
 }
 
-
-multiclass avx512_rndscale<bits<8> opc, string OpcodeStr,
-                            X86MemOperand x86memop, RegisterClass RC,
-                            PatFrag mem_frag, Domain d> {
-let ExeDomain = d in {
-  // Intrinsic operation, reg.
-  // Vector intrinsic operation, reg
-  def r : AVX512AIi8<opc, MRMSrcReg,
-                    (outs RC:$dst), (ins RC:$src1, i32u8imm:$src2),
-                    !strconcat(OpcodeStr,
-                    "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
-                    []>, EVEX;
-
-  // Vector intrinsic operation, mem
-  def m : AVX512AIi8<opc, MRMSrcMem,
-                    (outs RC:$dst), (ins x86memop:$src1, i32u8imm:$src2),
-                    !strconcat(OpcodeStr,
-                    "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
-                    []>, EVEX;
-} // ExeDomain
-}
-
-defm VRNDSCALEPSZ : avx512_rndscale<0x08, "vrndscaleps", f512mem, VR512,
-                                loadv16f32, SSEPackedSingle>, EVEX_V512,
-                                EVEX_CD8<32, CD8VF>;
-
-def : Pat<(v16f32 (int_x86_avx512_mask_rndscale_ps_512 (v16f32 VR512:$src1),
-                   imm:$src2, (v16f32 VR512:$src1), (i16 -1),
-                   FROUND_CURRENT)),
-                   (VRNDSCALEPSZr VR512:$src1, imm:$src2)>;
-
-
-defm VRNDSCALEPDZ : avx512_rndscale<0x09, "vrndscalepd", f512mem, VR512,
-                                loadv8f64, SSEPackedDouble>, EVEX_V512,
-                                VEX_W, EVEX_CD8<64, CD8VF>;
-
-def : Pat<(v8f64 (int_x86_avx512_mask_rndscale_pd_512 (v8f64 VR512:$src1),
-                  imm:$src2, (v8f64 VR512:$src1), (i8 -1),
-                  FROUND_CURRENT)),
-                   (VRNDSCALEPDZr VR512:$src1, imm:$src2)>;
-
 multiclass
 avx512_rndscale_scalar<bits<8> opc, string OpcodeStr, X86VectorVTInfo _> {
 
   let ExeDomain = _.ExeDomain in {
   defm r : AVX512_maskable_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
                            (ins _.RC:$src1, _.RC:$src2, i32u8imm:$src3), OpcodeStr,
                            "$src3, $src2, $src1", "$src1, $src2, $src3",
-                           (_.VT (X86RndScale (_.VT _.RC:$src1), (_.VT _.RC:$src2),
+                           (_.VT (X86RndScales (_.VT _.RC:$src1), (_.VT _.RC:$src2),
                             (i32 imm:$src3), (i32 FROUND_CURRENT)))>;
 
   defm rb : AVX512_maskable_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
                          (ins _.RC:$src1, _.RC:$src2, i32u8imm:$src3), OpcodeStr,
-                         "{sae}, $src3, $src2, $src1", "$src1, $src2, $src3, {sae}",
-                         (_.VT (X86RndScale (_.VT _.RC:$src1), (_.VT _.RC:$src2),
+                         "$src3, {sae}, $src2, $src1", "$src1, $src2, {sae}, $src3",
+                         (_.VT (X86RndScales (_.VT _.RC:$src1), (_.VT _.RC:$src2),
                          (i32 imm:$src3), (i32 FROUND_NO_EXC)))>, EVEX_B;
 
   let mayLoad = 1 in
   defm m : AVX512_maskable_scalar<opc, MRMSrcMem, _, (outs _.RC:$dst),
                          (ins _.RC:$src1, _.MemOp:$src2, i32u8imm:$src3), OpcodeStr,
                          "$src3, $src2, $src1", "$src1, $src2, $src3",
-                         (_.VT (X86RndScale (_.VT _.RC:$src1),
+                         (_.VT (X86RndScales (_.VT _.RC:$src1),
                           (_.VT (scalar_to_vector (_.ScalarLdFrag addr:$src2))),
                           (i32 imm:$src3), (i32 FROUND_CURRENT)))>;
   }
   let Predicates = [HasAVX512] in {
   def : Pat<(ffloor _.FRC:$src), (COPY_TO_REGCLASS
              (_.VT (!cast<Instruction>(NAME##r) (_.VT (IMPLICIT_DEF)),
              (_.VT (COPY_TO_REGCLASS _.FRC:$src, _.RC)), (i32 0x1))), _.FRC)>;
   def : Pat<(fceil _.FRC:$src), (COPY_TO_REGCLASS

 }
 
 defm VRNDSCALESS : avx512_rndscale_scalar<0x0A, "vrndscaless", f32x_info>,
                                 AVX512AIi8Base, EVEX_4V, EVEX_CD8<32, CD8VT1>;
 
 defm VRNDSCALESD : avx512_rndscale_scalar<0x0B, "vrndscalesd", f64x_info>, VEX_W,
                                 AVX512AIi8Base, EVEX_4V, EVEX_CD8<64, CD8VT1>;
 
-let Predicates = [HasAVX512] in {
-def : Pat<(v16f32 (ffloor VR512:$src)),
-          (VRNDSCALEPSZr VR512:$src, (i32 0x1))>;
-def : Pat<(v16f32 (fnearbyint VR512:$src)),
-          (VRNDSCALEPSZr VR512:$src, (i32 0xC))>;
-def : Pat<(v16f32 (fceil VR512:$src)),
-          (VRNDSCALEPSZr VR512:$src, (i32 0x2))>;
-def : Pat<(v16f32 (frint VR512:$src)),
-          (VRNDSCALEPSZr VR512:$src, (i32 0x4))>;
-def : Pat<(v16f32 (ftrunc VR512:$src)),
-          (VRNDSCALEPSZr VR512:$src, (i32 0x3))>;
-
-def : Pat<(v8f64 (ffloor VR512:$src)),
-          (VRNDSCALEPDZr VR512:$src, (i32 0x1))>;
-def : Pat<(v8f64 (fnearbyint VR512:$src)),
-          (VRNDSCALEPDZr VR512:$src, (i32 0xC))>;
-def : Pat<(v8f64 (fceil VR512:$src)),
-          (VRNDSCALEPDZr VR512:$src, (i32 0x2))>;
-def : Pat<(v8f64 (frint VR512:$src)),
-          (VRNDSCALEPDZr VR512:$src, (i32 0x4))>;
-def : Pat<(v8f64 (ftrunc VR512:$src)),
-          (VRNDSCALEPDZr VR512:$src, (i32 0x3))>;
-}
 //-------------------------------------------------
 // Integer truncate and extend operations
 //-------------------------------------------------
 
 multiclass avx512_trunc_sat<bits<8> opc, string OpcodeStr,
                           RegisterClass dstRC, RegisterClass srcRC,
                           RegisterClass KRC, X86MemOperand x86memop> {
   def rr : AVX512XS8I<opc, MRMDestReg, (outs dstRC:$dst),
defm VPEXPANDD : expand_by_elt_width <0x89, "vpexpandd", avx512vl_i32_info>,
                                          EVEX;
 defm VPEXPANDQ : expand_by_elt_width <0x89, "vpexpandq", avx512vl_i64_info>,
                                          EVEX, VEX_W;
 defm VEXPANDPS : expand_by_elt_width <0x88, "vexpandps", avx512vl_f32_info>,
                                          EVEX;
 defm VEXPANDPD : expand_by_elt_width <0x88, "vexpandpd", avx512vl_f64_info>,
                                          EVEX, VEX_W;
 
+//handle instruction  reg_vec1 = op(reg_vec,imm)
+//                               op(mem_vec,imm)
+//                               op(broadcast(eltVt),imm)
+//all instruction created with FROUND_CURRENT
+multiclass avx512_unary_fp_packed_imm<bits<8> opc, string OpcodeStr, SDNode OpNode,
+                                                            X86VectorVTInfo _>{
+  defm rri : AVX512_maskable<opc, MRMSrcReg, _, (outs _.RC:$dst),
+                      (ins _.RC:$src1, i32u8imm:$src2),
+                      OpcodeStr##_.Suffix, "$src2, $src1", "$src2, $src2",
+                      (OpNode (_.VT _.RC:$src1),
+                              (i32 imm:$src2),
+                              (i32 FROUND_CURRENT))>;
+  let mayLoad = 1 in {
+    defm rmi : AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
+                      (ins _.MemOp:$src1, i32u8imm:$src2),
+                      OpcodeStr##_.Suffix, "$src2, $src1", "$src1, $src2",
+                      (OpNode (_.VT (bitconvert (_.LdFrag addr:$src1))),
+                              (i32 imm:$src2),
+                              (i32 FROUND_CURRENT))>;
+    defm rmbi : AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
+                      (ins _.ScalarMemOp:$src1, i32u8imm:$src2),
+                      OpcodeStr##_.Suffix, "$src2, ${src1}"##_.BroadcastStr,
+                      "${src1}"##_.BroadcastStr##", $src2",
+                      (OpNode (_.VT (X86VBroadcast(_.ScalarLdFrag addr:$src1))),
+                              (i32 imm:$src2),
+                              (i32 FROUND_CURRENT))>, EVEX_B;
+  }
+}
+
+//handle instruction  reg_vec1 = op(reg_vec2,reg_vec3,imm),{sae}
+multiclass avx512_unary_fp_sae_packed_imm<bits<8> opc, string OpcodeStr,
+                                             SDNode OpNode, X86VectorVTInfo _>{
+  defm rrib : AVX512_maskable<opc, MRMSrcReg, _, (outs _.RC:$dst),
+                      (ins _.RC:$src1, i32u8imm:$src2),
+                      OpcodeStr##_.Suffix, "$src2,{sae}, $src1",
+                      "$src1, {sae}, $src2",
+                      (OpNode (_.VT _.RC:$src1),
+                              (i32 imm:$src2),
+                              (i32 FROUND_NO_EXC))>, EVEX_B;
+}
+
+multiclass avx512_common_unary_fp_sae_packed_imm<string OpcodeStr,
+            AVX512VLVectorVTInfo _, bits<8> opc, SDNode OpNode, Predicate prd>{
+  let Predicates = [prd] in {
+    defm Z    : avx512_unary_fp_packed_imm<opc, OpcodeStr, OpNode, _.info512>,
+                avx512_unary_fp_sae_packed_imm<opc, OpcodeStr, OpNode, _.info512>,
+                                  EVEX_V512;
+  }
+  let Predicates = [prd, HasVLX] in {
+    defm Z128 : avx512_unary_fp_packed_imm<opc, OpcodeStr, OpNode, _.info128>,
+                                  EVEX_V128;
+    defm Z256 : avx512_unary_fp_packed_imm<opc, OpcodeStr, OpNode, _.info256>,
+                                  EVEX_V256;
+  }
+}
+
 //handle instruction  reg_vec1 = op(reg_vec2,reg_vec3,imm)
 //                               op(reg_vec2,mem_vec,imm)
 //                               op(reg_vec2,broadcast(eltVt),imm)
 //all instruction created with FROUND_CURRENT
 multiclass avx512_fp_packed_imm<bits<8> opc, string OpcodeStr, SDNode OpNode,
                                                             X86VectorVTInfo _>{
   defm rri : AVX512_maskable<opc, MRMSrcReg, _, (outs _.RC:$dst),
-                      (ins _.RC:$src1, _.RC:$src2, u8imm:$src3),
+                      (ins _.RC:$src1, _.RC:$src2, i32u8imm:$src3),
                       OpcodeStr, "$src3, $src2, $src1", "$src1, $src2, $src3",
                       (OpNode (_.VT _.RC:$src1),
                               (_.VT _.RC:$src2),
-                              (i8 imm:$src3),
+                              (i32 imm:$src3),
                               (i32 FROUND_CURRENT))>;
   let mayLoad = 1 in {
     defm rmi : AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
-                      (ins _.RC:$src1, _.MemOp:$src2, u8imm:$src3),
+                      (ins _.RC:$src1, _.MemOp:$src2, i32u8imm:$src3),
                       OpcodeStr, "$src3, $src2, $src1", "$src1, $src2, $src3",
                       (OpNode (_.VT _.RC:$src1),
                               (_.VT (bitconvert (_.LdFrag addr:$src2))),
-                              (i8 imm:$src3),
+                              (i32 imm:$src3),
                               (i32 FROUND_CURRENT))>;
     defm rmbi : AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
-                      (ins _.RC:$src1, _.ScalarMemOp:$src2, u8imm:$src3),
+                      (ins _.RC:$src1, _.ScalarMemOp:$src2, i32u8imm:$src3),
                       OpcodeStr, "$src3, ${src2}"##_.BroadcastStr##", $src1",
                       "$src1, ${src2}"##_.BroadcastStr##", $src3",
                       (OpNode (_.VT _.RC:$src1),
                               (_.VT (X86VBroadcast(_.ScalarLdFrag addr:$src2))),
-                              (i8 imm:$src3),
+                              (i32 imm:$src3),
                               (i32 FROUND_CURRENT))>, EVEX_B;
   }
 }
 
 //handle instruction  reg_vec1 = op(reg_vec2,reg_vec3,imm)
 //                               op(reg_vec2,mem_vec,imm)
 //                               op(reg_vec2,broadcast(eltVt),imm)
 multiclass avx512_3Op_imm8<bits<8> opc, string OpcodeStr, SDNode OpNode,

 
 //handle scalar instruction  reg_vec1 = op(reg_vec2,reg_vec3,imm)
 //                                      op(reg_vec2,mem_scalar,imm)
 //all instruction created with FROUND_CURRENT
 multiclass avx512_fp_scalar_imm<bits<8> opc, string OpcodeStr, SDNode OpNode,
                                                            X86VectorVTInfo _> {
 
   defm rri : AVX512_maskable_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
-                      (ins _.RC:$src1, _.RC:$src2, u8imm:$src3),
+                      (ins _.RC:$src1, _.RC:$src2, i32u8imm:$src3),
                       OpcodeStr, "$src3, $src2, $src1", "$src1, $src2, $src3",
                       (OpNode (_.VT _.RC:$src1),
                               (_.VT _.RC:$src2),
-                              (i8 imm:$src3),
+                              (i32 imm:$src3),
                               (i32 FROUND_CURRENT))>;
   let mayLoad = 1 in {
     defm rmi : AVX512_maskable_scalar<opc, MRMSrcMem, _, (outs _.RC:$dst),
-                      (ins _.RC:$src1, _.MemOp:$src2, u8imm:$src3),
+                      (ins _.RC:$src1, _.MemOp:$src2, i32u8imm:$src3),
                       OpcodeStr, "$src3, $src2, $src1", "$src1, $src2, $src3",
                       (OpNode (_.VT _.RC:$src1),
                               (_.VT (scalar_to_vector
                                         (_.ScalarLdFrag addr:$src2))),
-                              (i8 imm:$src3),
+                              (i32 imm:$src3),
                               (i32 FROUND_CURRENT))>;
 
     let isAsmParserOnly = 1 in {
       defm rmi_alt :AVX512_maskable_in_asm<opc, MRMSrcMem, _, (outs _.FRC:$dst),
                       (ins _.FRC:$src1, _.ScalarMemOp:$src2, u8imm:$src3),
                       OpcodeStr, "$src3, $src2, $src1", "$src1, $src2, $src3",
                       []>;
     }
   }
 }
 
 //handle instruction  reg_vec1 = op(reg_vec2,reg_vec3,imm),{sae}
 multiclass avx512_fp_sae_packed_imm<bits<8> opc, string OpcodeStr,
                                              SDNode OpNode, X86VectorVTInfo _>{
   defm rrib : AVX512_maskable<opc, MRMSrcReg, _, (outs _.RC:$dst),
-                      (ins _.RC:$src1, _.RC:$src2, u8imm:$src3),
+                      (ins _.RC:$src1, _.RC:$src2, i32u8imm:$src3),
                       OpcodeStr, "$src3,{sae}, $src2, $src1",
                       "$src1, $src2,{sae}, $src3",
                       (OpNode (_.VT _.RC:$src1),
                               (_.VT _.RC:$src2),
-                              (i8 imm:$src3),
+                              (i32 imm:$src3),
                               (i32 FROUND_NO_EXC))>, EVEX_B;
 }
 //handle scalar instruction  reg_vec1 = op(reg_vec2,reg_vec3,imm),{sae}
 multiclass avx512_fp_sae_scalar_imm<bits<8> opc, string OpcodeStr,
                                              SDNode OpNode, X86VectorVTInfo _> {
-  defm NAME: avx512_fp_sae_packed_imm<opc, OpcodeStr, OpNode, _>;
+  defm NAME#rrib : AVX512_maskable_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
+                      (ins _.RC:$src1, _.RC:$src2, i32u8imm:$src3),
+                      OpcodeStr, "$src3,{sae}, $src2, $src1",
+                      "$src1, $src2,{sae}, $src3",
+                      (OpNode (_.VT _.RC:$src1),
+                              (_.VT _.RC:$src2),
+                              (i32 imm:$src3),
+                              (i32 FROUND_NO_EXC))>, EVEX_B;
 }
 
 multiclass avx512_common_fp_sae_packed_imm<string OpcodeStr,
             AVX512VLVectorVTInfo _, bits<8> opc, SDNode OpNode, Predicate prd>{
   let Predicates = [prd] in {
     defm Z    : avx512_fp_packed_imm<opc, OpcodeStr, OpNode, _.info512>,
                 avx512_fp_sae_packed_imm<opc, OpcodeStr, OpNode, _.info512>,
                                   EVEX_V512;

 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;
+}
+
 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 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>;
 
 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;
   }
 }
+let Predicates = [HasAVX512] in {
+def : Pat<(v16f32 (ffloor VR512:$src)),
+          (VRNDSCALEPSZrri VR512:$src, (i32 0x1))>;
+def : Pat<(v16f32 (fnearbyint VR512:$src)),
+          (VRNDSCALEPSZrri VR512:$src, (i32 0xC))>;
+def : Pat<(v16f32 (fceil VR512:$src)),
+          (VRNDSCALEPSZrri VR512:$src, (i32 0x2))>;
+def : Pat<(v16f32 (frint VR512:$src)),
+          (VRNDSCALEPSZrri VR512:$src, (i32 0x4))>;
+def : Pat<(v16f32 (ftrunc VR512:$src)),
+          (VRNDSCALEPSZrri VR512:$src, (i32 0x3))>;
+
+def : Pat<(v8f64 (ffloor VR512:$src)),
+          (VRNDSCALEPDZrri VR512:$src, (i32 0x1))>;
+def : Pat<(v8f64 (fnearbyint VR512:$src)),
+          (VRNDSCALEPDZrri VR512:$src, (i32 0xC))>;
+def : Pat<(v8f64 (fceil VR512:$src)),
+          (VRNDSCALEPDZrri VR512:$src, (i32 0x2))>;
+def : Pat<(v8f64 (frint VR512:$src)),
+          (VRNDSCALEPDZrri VR512:$src, (i32 0x4))>;
+def : Pat<(v8f64 (ftrunc VR512:$src)),
+          (VRNDSCALEPDZrri VR512:$src, (i32 0x3))>;
+}
 
 defm VSHUFF32X4 : avx512_shuff_packed_128<"vshuff32x4",avx512vl_f32_info, 0x23>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<32, CD8VF>;
 defm VSHUFF64X2 : avx512_shuff_packed_128<"vshuff64x2",avx512vl_f64_info, 0x23>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<64, CD8VF>, VEX_W;
 defm VSHUFI32X4 : avx512_shuff_packed_128<"vshufi32x4",avx512vl_i32_info, 0x43>,
       AVX512AIi8Base, EVEX_4V, EVEX_CD8<32, CD8VF>;
 defm VSHUFI64X2 : avx512_shuff_packed_128<"vshufi64x2",avx512vl_i64_info, 0x43>,

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

 def SDTShuff2OpM : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>,
                                         SDTCisVec<2>]>;
 def SDTShuff2OpI : SDTypeProfile<1, 2, [SDTCisVec<0>,
                                  SDTCisSameAs<0,1>, SDTCisInt<2>]>;
 def SDTShuff3OpI : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
                                  SDTCisSameAs<0,2>, SDTCisInt<3>]>;
 def SDTFPBinOpImmRound: SDTypeProfile<1, 4, [SDTCisVec<0>, SDTCisSameAs<0,1>,
                              SDTCisSameAs<0,2>, SDTCisInt<3>, SDTCisInt<4>]>;
+def SDTFPUnaryOpImmRound: SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
+                              SDTCisInt<2>, SDTCisInt<3>]>;
 
 def SDTVBroadcast  : SDTypeProfile<1, 1, [SDTCisVec<0>]>;
 def SDTVBroadcastm : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>]>;
 
 def SDTBlend : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
                              SDTCisSameAs<1,2>, SDTCisVT<3, i8>]>;
 
 def SDTFPBinOpRound : SDTypeProfile<1, 3, [      // fadd_round, fmul_round, etc.

 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 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 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 X86RndScale  : SDNode<"X86ISD::RNDSCALE", STDFp3SrcRm>;
+def X86RndScales : SDNode<"X86ISD::VRNDSCALE", STDFp3SrcRm>;
+def X86Reduces   : SDNode<"X86ISD::VREDUCE", 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

 namespace llvm {
 
 enum IntrinsicType {
   INTR_NO_TYPE,
   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, FMA_OP_MASK, FMA_OP_MASKZ, FMA_OP_MASK3, VPERM_3OP_MASK,
+  INTR_TYPE_3OP_MASK, INTR_TYPE_3OP_MASK_RM, 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,
   EXPAND_FROM_MEM, BLEND
 };
 
 struct IntrinsicData {
 
   unsigned      Id;
static const IntrinsicData  IntrinsicsWithoutChain[] = {
   X86_INTRINSIC_DATA(avx512_mask_psubus_w_256, INTR_TYPE_2OP_MASK, X86ISD::SUBUS, 0),
   X86_INTRINSIC_DATA(avx512_mask_psubus_w_512, INTR_TYPE_2OP_MASK, X86ISD::SUBUS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pxor_d_128, INTR_TYPE_2OP_MASK, ISD::XOR, 0),
   X86_INTRINSIC_DATA(avx512_mask_pxor_d_256, INTR_TYPE_2OP_MASK, ISD::XOR, 0),
   X86_INTRINSIC_DATA(avx512_mask_pxor_d_512, INTR_TYPE_2OP_MASK, ISD::XOR, 0),
   X86_INTRINSIC_DATA(avx512_mask_pxor_q_128, INTR_TYPE_2OP_MASK, ISD::XOR, 0),
   X86_INTRINSIC_DATA(avx512_mask_pxor_q_256, INTR_TYPE_2OP_MASK, ISD::XOR, 0),
   X86_INTRINSIC_DATA(avx512_mask_pxor_q_512, INTR_TYPE_2OP_MASK, ISD::XOR, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_pd_128, INTR_TYPE_3OP_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_pd_256, INTR_TYPE_3OP_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_pd_512, INTR_TYPE_3OP_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_ps_128, INTR_TYPE_3OP_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_ps_256, INTR_TYPE_3OP_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_ps_512, INTR_TYPE_3OP_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_sd, INTR_TYPE_SCALAR_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_range_ss, INTR_TYPE_SCALAR_MASK_RM, X86ISD::VRANGE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_pd_128, INTR_TYPE_2OP_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_pd_256, INTR_TYPE_2OP_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_pd_512, INTR_TYPE_2OP_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_ps_128, INTR_TYPE_2OP_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_ps_256, INTR_TYPE_2OP_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_ps_512, INTR_TYPE_2OP_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_sd, INTR_TYPE_SCALAR_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_reduce_ss, INTR_TYPE_SCALAR_MASK_RM, X86ISD::VREDUCE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_rndscale_pd_128, INTR_TYPE_2OP_MASK_RM, X86ISD::VRNDSCALE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_rndscale_pd_256, INTR_TYPE_2OP_MASK_RM, X86ISD::VRNDSCALE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_rndscale_pd_512, INTR_TYPE_2OP_MASK_RM, X86ISD::VRNDSCALE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_rndscale_ps_128, INTR_TYPE_2OP_MASK_RM, X86ISD::VRNDSCALE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_rndscale_ps_256, INTR_TYPE_2OP_MASK_RM, X86ISD::VRNDSCALE, 0),
+  X86_INTRINSIC_DATA(avx512_mask_rndscale_ps_512, INTR_TYPE_2OP_MASK_RM, X86ISD::VRNDSCALE, 0),
   X86_INTRINSIC_DATA(avx512_mask_rndscale_sd,   INTR_TYPE_SCALAR_MASK_RM,
-                     X86ISD::RNDSCALE, 0),
+                     X86ISD::VRNDSCALE, 0),
   X86_INTRINSIC_DATA(avx512_mask_rndscale_ss,   INTR_TYPE_SCALAR_MASK_RM,
-                     X86ISD::RNDSCALE, 0),
+                     X86ISD::VRNDSCALE, 0),
   X86_INTRINSIC_DATA(avx512_mask_scalef_pd_128, INTR_TYPE_2OP_MASK_RM,
                      X86ISD::SCALEF, 0),
   X86_INTRINSIC_DATA(avx512_mask_scalef_pd_256, INTR_TYPE_2OP_MASK_RM,
                      X86ISD::SCALEF, 0),
   X86_INTRINSIC_DATA(avx512_mask_scalef_pd_512, INTR_TYPE_2OP_MASK_RM,
                      X86ISD::SCALEF, 0),
   X86_INTRINSIC_DATA(avx512_mask_scalef_ps_128, INTR_TYPE_2OP_MASK_RM,
                      X86ISD::SCALEF, 0),
   X86_INTRINSIC_DATA(avx512_mask_scalef_ps_256, INTR_TYPE_2OP_MASK_RM,
                      X86ISD::SCALEF, 0),
   X86_INTRINSIC_DATA(avx512_mask_scalef_ps_512, INTR_TYPE_2OP_MASK_RM,
                      X86ISD::SCALEF, 0),
+  X86_INTRINSIC_DATA(avx512_mask_scalef_sd, INTR_TYPE_SCALAR_MASK_RM,
+                     X86ISD::SCALEF, 0),
+  X86_INTRINSIC_DATA(avx512_mask_scalef_ss, INTR_TYPE_SCALAR_MASK_RM,
+                     X86ISD::SCALEF, 0),
   X86_INTRINSIC_DATA(avx512_mask_sqrt_pd_128, INTR_TYPE_1OP_MASK, ISD::FSQRT, 0),
   X86_INTRINSIC_DATA(avx512_mask_sqrt_pd_256, INTR_TYPE_1OP_MASK, ISD::FSQRT, 0),
   X86_INTRINSIC_DATA(avx512_mask_sqrt_pd_512, INTR_TYPE_1OP_MASK_RM, ISD::FSQRT,
                      X86ISD::FSQRT_RND),
   X86_INTRINSIC_DATA(avx512_mask_sqrt_ps_128, INTR_TYPE_1OP_MASK, ISD::FSQRT, 0),
   X86_INTRINSIC_DATA(avx512_mask_sqrt_ps_256, INTR_TYPE_1OP_MASK, ISD::FSQRT, 0),
   X86_INTRINSIC_DATA(avx512_mask_sqrt_ps_512, INTR_TYPE_1OP_MASK_RM, ISD::FSQRT,
                      X86ISD::FSQRT_RND),

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

 ; CHECK-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
 ; CHECK-NEXT:    retq
   %res = call <16 x i32> @llvm.x86.avx512.mask.cvttps2udq.512(<16 x float> %x0, <16 x i32> %x1, i16 %x2, i32 4)
   %res1 = call <16 x i32> @llvm.x86.avx512.mask.cvttps2udq.512(<16 x float> %x0, <16 x i32> %x1, i16 -1, i32 8)
   %res2 = add <16 x i32> %res, %res1
   ret <16 x i32> %res2
 }
 
+
+declare <4 x float> @llvm.x86.avx512.mask.scalef.ss(<4 x float>, <4 x float>,<4 x float>, i8, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_scalef_ss
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vscalefss {{.*}}{%k1} 
+; CHECK: vscalefss  {rn-sae}
+define <4 x float>@test_int_x86_avx512_mask_scalef_ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4) {
+  %res = call <4 x float> @llvm.x86.avx512.mask.scalef.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4, i32 4)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.scalef.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 -1, i32 8)
+  %res2 = fadd <4 x float> %res, %res1
+  ret <4 x float> %res2
+}
+
+declare <2 x double> @llvm.x86.avx512.mask.scalef.sd(<2 x double>, <2 x double>,<2 x double>, i8, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_scalef_sd
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vscalefsd {{.*}}{%k1} 
+; CHECK: vscalefsd  {rn-sae}
+define <2 x double>@test_int_x86_avx512_mask_scalef_sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
+  %res = call <2 x double> @llvm.x86.avx512.mask.scalef.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4, i32 4)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.scalef.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 -1, i32 8)
+  %res2 = fadd <2 x double> %res, %res1
+  ret <2 x double> %res2
+}

llvm/trunk/test/CodeGen/X86/avx512dq-intrinsics.ll

+
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=knl -mattr=+avx512dq | FileCheck %s
 
 declare <8 x i64> @llvm.x86.avx512.mask.cvtpd2qq.512(<8 x double>, <8 x i64>, i8, i32)
 
 define <8 x i64>@test_int_x86_avx512_mask_cvt_pd2qq_512(<8 x double> %x0, <8 x i64> %x1, i8 %x2) {
 ; CHECK-LABEL: test_int_x86_avx512_mask_cvt_pd2qq_512:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovb %edi, %k1

 ; CHECK-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %res = call <8 x float> @llvm.x86.avx512.mask.cvtuqq2ps.512(<8 x i64> %x0, <8 x float> %x1, i8 %x2, i32 4)
   %res1 = call <8 x float> @llvm.x86.avx512.mask.cvtuqq2ps.512(<8 x i64> %x0, <8 x float> %x1, i8 -1, i32 0)
   %res2 = fadd <8 x float> %res, %res1
   ret <8 x float> %res2
 }
 
+declare <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double>, i32, <8 x double>, i8, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_pd_512
+; CHECK-NOT: call
+; CHECK: kmov
+; CHECK: vreducepd {{.*}}{%k1}
+; CHECK: vreducepd
+; CHECK: {sae}
+define <8 x double>@test_int_x86_avx512_mask_reduce_pd_512(<8 x double> %x0, <8 x double> %x2, i8 %x3) {
+  %res = call <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double> %x0, i32 8, <8 x double> %x2, i8 %x3, i32 4)
+  %res1 = call <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double> %x0, i32 4, <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.reduce.ps.512(<16 x float>, i32, <16 x float>, i16, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_ps_512
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreduceps
+; CHECK: {sae}
+; CKECK: {%k1}
+; CHECK: vreduceps
+define <16 x float>@test_int_x86_avx512_mask_reduce_ps_512(<16 x float> %x0, <16 x float> %x2, i16 %x3) {
+  %res = call <16 x float> @llvm.x86.avx512.mask.reduce.ps.512(<16 x float> %x0, i32 44, <16 x float> %x2, i16 %x3, i32 8)
+  %res1 = call <16 x float> @llvm.x86.avx512.mask.reduce.ps.512(<16 x float> %x0, i32 11, <16 x float> %x2, i16 -1, i32 4)
+  %res2 = fadd <16 x float> %res, %res1
+  ret <16 x float> %res2
+}
+
+declare <8 x double> @llvm.x86.avx512.mask.range.pd.512(<8 x double>, <8 x double>, i32, <8 x double>, i8, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_pd_512
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangepd
+; CKECK: {%k1}
+; CHECK: vrangepd
+; CHECK: {sae}
+define <8 x double>@test_int_x86_avx512_mask_range_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x3, i8 %x4) {
+  %res = call <8 x double> @llvm.x86.avx512.mask.range.pd.512(<8 x double> %x0, <8 x double> %x1, i32 8, <8 x double> %x3, i8 %x4, i32 4)
+  %res1 = call <8 x double> @llvm.x86.avx512.mask.range.pd.512(<8 x double> %x0, <8 x double> %x1, i32 4, <8 x double> %x3, i8 -1, i32 8)
+  %res2 = fadd <8 x double> %res, %res1
+  ret <8 x double> %res2
+}
+
+declare <16 x float> @llvm.x86.avx512.mask.range.ps.512(<16 x float>, <16 x float>, i32, <16 x float>, i16, i32)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_ps_512
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangeps
+; CKECK: {%k1}
+; CHECK: vrangeps
+; CHECK: {sae}
+define <16 x float>@test_int_x86_avx512_mask_range_ps_512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x3, i16 %x4) {
+  %res = call <16 x float> @llvm.x86.avx512.mask.range.ps.512(<16 x float> %x0, <16 x float> %x1, i32 88, <16 x float> %x3, i16 %x4, i32 4)
+  %res1 = call <16 x float> @llvm.x86.avx512.mask.range.ps.512(<16 x float> %x0, <16 x float> %x1, i32 4, <16 x float> %x3, i16 -1, i32 8)
+  %res2 = fadd <16 x float> %res, %res1
+  ret <16 x float> %res2
+}
+
+declare <4 x float> @llvm.x86.avx512.mask.reduce.ss(<4 x float>, <4 x float>,<4 x float>, i8, i32, i32)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_ss
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreducess
+; CKECK: {%k1}
+; CHECK: vreducess
+; CHECK: {sae}
+define <4 x float>@test_int_x86_avx512_mask_reduce_ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4) {
+  %res = call <4 x float> @llvm.x86.avx512.mask.reduce.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4, i32 4, i32 4)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.reduce.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 -1, i32 4, i32 8)
+  %res2 = fadd <4 x float> %res, %res1
+  ret <4 x float> %res2
+}
+
+declare <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float>, <4 x float>,<4 x float>, i8, i32, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_ss
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangess
+; CHECK: {sae}
+; CKECK: {%k1}
+; CHECK: vrangess
+; CHECK: {sae} 
+define <4 x float>@test_int_x86_avx512_mask_range_ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4) {
+  %res = call <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4, i32 4, i32 8)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 -1, i32 4, i32 8)
+  %res2 = fadd <4 x float> %res, %res1
+  ret <4 x float> %res2
+}
+
+declare <2 x double> @llvm.x86.avx512.mask.reduce.sd(<2 x double>, <2 x double>,<2 x double>, i8, i32, i32)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_sd
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreducesd
+; CKECK: {%k1}
+; CHECK: vreducesd
+; CHECK: {sae} 
+define <2 x double>@test_int_x86_avx512_mask_reduce_sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
+  %res = call <2 x double> @llvm.x86.avx512.mask.reduce.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4, i32 4, i32 4)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.reduce.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 -1, i32 4, i32 8)
+  %res2 = fadd <2 x double> %res, %res1
+  ret <2 x double> %res2
+}
+
+declare <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double>, <2 x double>,<2 x double>, i8, i32, i32)
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_sd
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangesd
+; CKECK: {%k1}
+; CHECK: vrangesd
+; CHECK: {sae} 
+define <2 x double>@test_int_x86_avx512_mask_range_sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
+  %res = call <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4, i32 4, i32 4)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 -1, i32 4, i32 8)
+  %res2 = fadd <2 x double> %res, %res1
+  ret <2 x double> %res2
+}

llvm/trunk/test/CodeGen/X86/avx512dqvl-intrinsics.ll

 ; CHECK-NEXT:    vcvttps2uqq %xmm0, %ymm0
 ; CHECK-NEXT:    vpaddq %ymm0, %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %res = call <4 x i64> @llvm.x86.avx512.mask.cvttps2uqq.256(<4 x float> %x0, <4 x i64> %x1, i8 %x2)
   %res1 = call <4 x i64> @llvm.x86.avx512.mask.cvttps2uqq.256(<4 x float> %x0, <4 x i64> %x1, i8 -1)
   %res2 = add <4 x i64> %res, %res1
   ret <4 x i64> %res2
 }
+
+declare <2 x double> @llvm.x86.avx512.mask.reduce.pd.128(<2 x double>, i32, <2 x double>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_pd_128
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreducepd {{.*}}{%k1} 
+; CHECK: vreducepd
+define <2 x double>@test_int_x86_avx512_mask_reduce_pd_128(<2 x double> %x0, <2 x double> %x2, i8 %x3) {
+  %res = call <2 x double> @llvm.x86.avx512.mask.reduce.pd.128(<2 x double> %x0, i32 4, <2 x double> %x2, i8 %x3)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.reduce.pd.128(<2 x double> %x0, i32 8, <2 x double> %x2, i8 -1)
+  %res2 = fadd <2 x double> %res, %res1
+  ret <2 x double> %res2
+}
+
+declare <4 x double> @llvm.x86.avx512.mask.reduce.pd.256(<4 x double>, i32, <4 x double>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_pd_256
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreducepd {{.*}}{%k1} 
+; CHECK: vreducepd
+define <4 x double>@test_int_x86_avx512_mask_reduce_pd_256(<4 x double> %x0, <4 x double> %x2, i8 %x3) {
+  %res = call <4 x double> @llvm.x86.avx512.mask.reduce.pd.256(<4 x double> %x0, i32 4, <4 x double> %x2, i8 %x3)
+  %res1 = call <4 x double> @llvm.x86.avx512.mask.reduce.pd.256(<4 x double> %x0, i32 0, <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.reduce.ps.128(<4 x float>, i32, <4 x float>, i8)
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_ps_128
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreduceps {{.*}}{%k1} 
+; CHECK: vreduceps
+define <4 x float>@test_int_x86_avx512_mask_reduce_ps_128(<4 x float> %x0, <4 x float> %x2, i8 %x3) {
+  %res = call <4 x float> @llvm.x86.avx512.mask.reduce.ps.128(<4 x float> %x0, i32 4, <4 x float> %x2, i8 %x3)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.reduce.ps.128(<4 x float> %x0, i32 88, <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.reduce.ps.256(<8 x float>, i32, <8 x float>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_reduce_ps_256
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vreduceps {{.*}}{%k1} 
+; CHECK: vreduceps
+define <8 x float>@test_int_x86_avx512_mask_reduce_ps_256(<8 x float> %x0, <8 x float> %x2, i8 %x3) {
+  %res = call <8 x float> @llvm.x86.avx512.mask.reduce.ps.256(<8 x float> %x0, i32 11, <8 x float> %x2, i8 %x3)
+  %res1 = call <8 x float> @llvm.x86.avx512.mask.reduce.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.range.pd.128(<2 x double>, <2 x double>, i32, <2 x double>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_pd_128
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangepd {{.*}}{%k1} 
+; CHECK: vrangepd
+define <2 x double>@test_int_x86_avx512_mask_range_pd_128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
+  %res = call <2 x double> @llvm.x86.avx512.mask.range.pd.128(<2 x double> %x0, <2 x double> %x1, i32 4, <2 x double> %x3, i8 %x4)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.range.pd.128(<2 x double> %x0, <2 x double> %x1, i32 8, <2 x double> %x3, i8 -1)
+  %res2 = fadd <2 x double> %res, %res1
+  ret <2 x double> %res2
+}
+
+declare <4 x double> @llvm.x86.avx512.mask.range.pd.256(<4 x double>, <4 x double>, i32, <4 x double>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_pd_256
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangepd {{.*}}{%k1} 
+; CHECK: vrangepd
+define <4 x double>@test_int_x86_avx512_mask_range_pd_256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x3, i8 %x4) {
+  %res = call <4 x double> @llvm.x86.avx512.mask.range.pd.256(<4 x double> %x0, <4 x double> %x1, i32 4, <4 x double> %x3, i8 %x4)
+  %res1 = call <4 x double> @llvm.x86.avx512.mask.range.pd.256(<4 x double> %x0, <4 x double> %x1, i32 88, <4 x double> %x3, i8 -1)
+  %res2 = fadd <4 x double> %res, %res1
+  ret <4 x double> %res2
+}
+
+declare <4 x float> @llvm.x86.avx512.mask.range.ps.128(<4 x float>, <4 x float>, i32, <4 x float>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_ps_128
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangeps {{.*}}{%k1} 
+; CHECK: vrangeps
+define <4 x float>@test_int_x86_avx512_mask_range_ps_128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4) {
+  %res = call <4 x float> @llvm.x86.avx512.mask.range.ps.128(<4 x float> %x0, <4 x float> %x1, i32 4, <4 x float> %x3, i8 %x4)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.range.ps.128(<4 x float> %x0, <4 x float> %x1, i32 88, <4 x float> %x3, i8 -1)
+  %res2 = fadd <4 x float> %res, %res1
+  ret <4 x float> %res2
+}
+
+declare <8 x float> @llvm.x86.avx512.mask.range.ps.256(<8 x float>, <8 x float>, i32, <8 x float>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_range_ps_256
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrangeps {{.*}}{%k1} 
+; CHECK: vrangeps
+define <8 x float>@test_int_x86_avx512_mask_range_ps_256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x3, i8 %x4) {
+  %res = call <8 x float> @llvm.x86.avx512.mask.range.ps.256(<8 x float> %x0, <8 x float> %x1, i32 4, <8 x float> %x3, i8 %x4)
+  %res1 = call <8 x float> @llvm.x86.avx512.mask.range.ps.256(<8 x float> %x0, <8 x float> %x1, i32 88, <8 x float> %x3, i8 -1)
+  %res2 = fadd <8 x float> %res, %res1
+  ret <8 x float> %res2
+}

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

 ; CHECK-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; CHECK-NEXT:    retq
   %res = call <8 x float> @llvm.x86.avx512.mask.cvtudq2ps.256(<8 x i32> %x0, <8 x float> %x1, i8 %x2)
   %res1 = call <8 x float> @llvm.x86.avx512.mask.cvtudq2ps.256(<8 x i32> %x0, <8 x float> %x1, i8 -1)
   %res2 = fadd <8 x float> %res, %res1
   ret <8 x float> %res2
 }
 
+declare <2 x double> @llvm.x86.avx512.mask.rndscale.pd.128(<2 x double>, i32, <2 x double>, i8)
+; CHECK-LABEL: @test_int_x86_avx512_mask_rndscale_pd_128
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrndscalepd {{.*}}{%k1} 
+; CHECK: vrndscalepd
+define <2 x double>@test_int_x86_avx512_mask_rndscale_pd_128(<2 x double> %x0, <2 x double> %x2, i8 %x3) {
+  %res = call <2 x double> @llvm.x86.avx512.mask.rndscale.pd.128(<2 x double> %x0, i32 4, <2 x double> %x2, i8 %x3)
+  %res1 = call <2 x double> @llvm.x86.avx512.mask.rndscale.pd.128(<2 x double> %x0, i32 88, <2 x double> %x2, i8 -1)
+  %res2 = fadd <2 x double> %res, %res1
+  ret <2 x double> %res2
+}
+
+declare <4 x double> @llvm.x86.avx512.mask.rndscale.pd.256(<4 x double>, i32, <4 x double>, i8)
+; CHECK-LABEL: @test_int_x86_avx512_mask_rndscale_pd_256
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrndscalepd {{.*}}{%k1} 
+; CHECK: vrndscalepd
+define <4 x double>@test_int_x86_avx512_mask_rndscale_pd_256(<4 x double> %x0, <4 x double> %x2, i8 %x3) {
+  %res = call <4 x double> @llvm.x86.avx512.mask.rndscale.pd.256(<4 x double> %x0, i32 4, <4 x double> %x2, i8 %x3)
+  %res1 = call <4 x double> @llvm.x86.avx512.mask.rndscale.pd.256(<4 x double> %x0, i32 88, <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.rndscale.ps.128(<4 x float>, i32, <4 x float>, i8)
+; CHECK-LABEL: @test_int_x86_avx512_mask_rndscale_ps_128
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrndscaleps {{.*}}{%k1} 
+; CHECK: vrndscaleps
+define <4 x float>@test_int_x86_avx512_mask_rndscale_ps_128(<4 x float> %x0, <4 x float> %x2, i8 %x3) {
+  %res = call <4 x float> @llvm.x86.avx512.mask.rndscale.ps.128(<4 x float> %x0, i32 88, <4 x float> %x2, i8 %x3)
+  %res1 = call <4 x float> @llvm.x86.avx512.mask.rndscale.ps.128(<4 x float> %x0, i32 4, <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.rndscale.ps.256(<8 x float>, i32, <8 x float>, i8)
+
+; CHECK-LABEL: @test_int_x86_avx512_mask_rndscale_ps_256
+; CHECK-NOT: call 
+; CHECK: kmov 
+; CHECK: vrndscaleps {{.*}}{%k1} 
+; 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
+}

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

 // CHECK: vcvtpd2ps -1024(%rdx){1to8}, %ymm5
 // CHECK:  encoding: [0x62,0xf1,0xfd,0x58,0x5a,0x6a,0x80]
           vcvtpd2ps -1024(%rdx){1to8}, %ymm5
 
 // CHECK: vcvtpd2ps -1032(%rdx){1to8}, %ymm5
 // CHECK:  encoding: [0x62,0xf1,0xfd,0x58,0x5a,0xaa,0xf8,0xfb,0xff,0xff]
           vcvtpd2ps -1032(%rdx){1to8}, %ymm5
 
+// CHECK: vscalefsd %xmm21, %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x00,0x2d,0xed]
+          vscalefsd %xmm21, %xmm22, %xmm21
+
+// CHECK: vscalefsd %xmm21, %xmm22, %xmm21 {%k2}
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x02,0x2d,0xed]
+          vscalefsd %xmm21, %xmm22, %xmm21 {%k2}
+
+// CHECK: vscalefsd %xmm21, %xmm22, %xmm21 {%k2} {z}
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x82,0x2d,0xed]
+          vscalefsd %xmm21, %xmm22, %xmm21 {%k2} {z}
+
+// CHECK: vscalefsd {rn-sae}, %xmm21, %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x10,0x2d,0xed]
+          vscalefsd {rn-sae}, %xmm21, %xmm22, %xmm21
+
+// CHECK: vscalefsd {ru-sae}, %xmm21, %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x50,0x2d,0xed]
+          vscalefsd {ru-sae}, %xmm21, %xmm22, %xmm21
+
+// CHECK: vscalefsd {rd-sae}, %xmm21, %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x30,0x2d,0xed]
+          vscalefsd {rd-sae}, %xmm21, %xmm22, %xmm21
+
+// CHECK: vscalefsd {rz-sae}, %xmm21, %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x70,0x2d,0xed]
+          vscalefsd {rz-sae}, %xmm21, %xmm22, %xmm21
+
+// CHECK: vscalefsd (%rcx), %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xe2,0xcd,0x00,0x2d,0x29]
+          vscalefsd (%rcx), %xmm22, %xmm21
+
+// CHECK: vscalefsd 291(%rax,%r14,8), %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xa2,0xcd,0x00,0x2d,0xac,0xf0,0x23,0x01,0x00,0x00]
+          vscalefsd 291(%rax,%r14,8), %xmm22, %xmm21
+
+// CHECK: vscalefsd 1016(%rdx), %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xe2,0xcd,0x00,0x2d,0x6a,0x7f]
+          vscalefsd 1016(%rdx), %xmm22, %xmm21
+
+// CHECK: vscalefsd 1024(%rdx), %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xe2,0xcd,0x00,0x2d,0xaa,0x00,0x04,0x00,0x00]
+          vscalefsd 1024(%rdx), %xmm22, %xmm21
+
+// CHECK: vscalefsd -1024(%rdx), %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xe2,0xcd,0x00,0x2d,0x6a,0x80]
+          vscalefsd -1024(%rdx), %xmm22, %xmm21
+
+// CHECK: vscalefsd -1032(%rdx), %xmm22, %xmm21
+// CHECK:  encoding: [0x62,0xe2,0xcd,0x00,0x2d,0xaa,0xf8,0xfb,0xff,0xff]
+          vscalefsd -1032(%rdx), %xmm22, %xmm21
+
+// CHECK: vscalefss %xmm23, %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x32,0x05,0x08,0x2d,0xef]
+          vscalefss %xmm23, %xmm15, %xmm13
+
+// CHECK: vscalefss %xmm23, %xmm15, %xmm13 {%k3}
+// CHECK:  encoding: [0x62,0x32,0x05,0x0b,0x2d,0xef]
+          vscalefss %xmm23, %xmm15, %xmm13 {%k3}
+
+// CHECK: vscalefss %xmm23, %xmm15, %xmm13 {%k3} {z}
+// CHECK:  encoding: [0x62,0x32,0x05,0x8b,0x2d,0xef]
+          vscalefss %xmm23, %xmm15, %xmm13 {%k3} {z}
+
+// CHECK: vscalefss {rn-sae}, %xmm23, %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x32,0x05,0x18,0x2d,0xef]
+          vscalefss {rn-sae}, %xmm23, %xmm15, %xmm13
+
+// CHECK: vscalefss {ru-sae}, %xmm23, %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x32,0x05,0x58,0x2d,0xef]
+          vscalefss {ru-sae}, %xmm23, %xmm15, %xmm13
+
+// CHECK: vscalefss {rd-sae}, %xmm23, %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x32,0x05,0x38,0x2d,0xef]
+          vscalefss {rd-sae}, %xmm23, %xmm15, %xmm13
+
+// CHECK: vscalefss {rz-sae}, %xmm23, %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x32,0x05,0x78,0x2d,0xef]
+          vscalefss {rz-sae}, %xmm23, %xmm15, %xmm13
+
+// CHECK: vscalefss (%rcx), %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x72,0x05,0x08,0x2d,0x29]
+          vscalefss (%rcx), %xmm15, %xmm13
+
+// CHECK: vscalefss 291(%rax,%r14,8), %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x32,0x05,0x08,0x2d,0xac,0xf0,0x23,0x01,0x00,0x00]
+          vscalefss 291(%rax,%r14,8), %xmm15, %xmm13
+
+// CHECK: vscalefss 508(%rdx), %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x72,0x05,0x08,0x2d,0x6a,0x7f]
+          vscalefss 508(%rdx), %xmm15, %xmm13
+
+// CHECK: vscalefss 512(%rdx), %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x72,0x05,0x08,0x2d,0xaa,0x00,0x02,0x00,0x00]
+          vscalefss 512(%rdx), %xmm15, %xmm13
+
+// CHECK: vscalefss -512(%rdx), %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x72,0x05,0x08,0x2d,0x6a,0x80]
+          vscalefss -512(%rdx), %xmm15, %xmm13
+
+// CHECK: vscalefss -516(%rdx), %xmm15, %xmm13
+// CHECK:  encoding: [0x62,0x72,0x05,0x08,0x2d,0xaa,0xfc,0xfd,0xff,0xff]
+          vscalefss -516(%rdx), %xmm15, %xmm13
+
+// CHECK: vrndscalepd $171, %zmm7, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0xf7,0xab]
+          vrndscalepd $0xab, %zmm7, %zmm22
+
+// CHECK: vrndscalepd $171, %zmm7, %zmm22 {%k1}
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x49,0x09,0xf7,0xab]
+          vrndscalepd $0xab, %zmm7, %zmm22 {%k1}
+
+// CHECK: vrndscalepd $171, %zmm7, %zmm22 {%k1} {z}
+// CHECK:  encoding: [0x62,0xe3,0xfd,0xc9,0x09,0xf7,0xab]
+          vrndscalepd $0xab, %zmm7, %zmm22 {%k1} {z}
+
+// CHECK: vrndscalepd $171,{sae}, %zmm7, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x09,0xf7,0xab]
+          vrndscalepd $0xab,{sae}, %zmm7, %zmm22
+
+// CHECK: vrndscalepd $123, %zmm7, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0xf7,0x7b]
+          vrndscalepd $0x7b, %zmm7, %zmm22
+
+// CHECK: vrndscalepd $123,{sae}, %zmm7, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x09,0xf7,0x7b]
+          vrndscalepd $0x7b,{sae}, %zmm7, %zmm22
+
+// CHECK: vrndscalepd $123, (%rcx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0x31,0x7b]
+          vrndscalepd $0x7b, (%rcx), %zmm22
+
+// CHECK: vrndscalepd $123, 291(%rax,%r14,8), %zmm22
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x48,0x09,0xb4,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 291(%rax,%r14,8), %zmm22
+
+// CHECK: vrndscalepd $123, (%rcx){1to8}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x09,0x31,0x7b]
+          vrndscalepd $0x7b, (%rcx){1to8}, %zmm22
+
+// CHECK: vrndscalepd $123, 8128(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0x72,0x7f,0x7b]
+          vrndscalepd $0x7b, 8128(%rdx), %zmm22
+
+// CHECK: vrndscalepd $123, 8192(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0xb2,0x00,0x20,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 8192(%rdx), %zmm22
+
+// CHECK: vrndscalepd $123, -8192(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0x72,0x80,0x7b]
+          vrndscalepd $0x7b, -8192(%rdx), %zmm22
+
+// CHECK: vrndscalepd $123, -8256(%rdx), %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x09,0xb2,0xc0,0xdf,0xff,0xff,0x7b]
+          vrndscalepd $0x7b, -8256(%rdx), %zmm22
+
+// CHECK: vrndscalepd $123, 1016(%rdx){1to8}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x09,0x72,0x7f,0x7b]
+          vrndscalepd $0x7b, 1016(%rdx){1to8}, %zmm22
+
+// CHECK: vrndscalepd $123, 1024(%rdx){1to8}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x09,0xb2,0x00,0x04,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 1024(%rdx){1to8}, %zmm22
+
+// CHECK: vrndscalepd $123, -1024(%rdx){1to8}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x09,0x72,0x80,0x7b]
+          vrndscalepd $0x7b, -1024(%rdx){1to8}, %zmm22
+
+// CHECK: vrndscalepd $123, -1032(%rdx){1to8}, %zmm22
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x09,0xb2,0xf8,0xfb,0xff,0xff,0x7b]
+          vrndscalepd $0x7b, -1032(%rdx){1to8}, %zmm22
+
+// CHECK: vrndscaleps $171, %zmm7, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0xef,0xab]
+          vrndscaleps $0xab, %zmm7, %zmm13
+
+// CHECK: vrndscaleps $171, %zmm7, %zmm13 {%k1}
+// CHECK:  encoding: [0x62,0x73,0x7d,0x49,0x08,0xef,0xab]
+          vrndscaleps $0xab, %zmm7, %zmm13 {%k1}
+
+// CHECK: vrndscaleps $171, %zmm7, %zmm13 {%k1} {z}
+// CHECK:  encoding: [0x62,0x73,0x7d,0xc9,0x08,0xef,0xab]
+          vrndscaleps $0xab, %zmm7, %zmm13 {%k1} {z}
+
+// CHECK: vrndscaleps $171,{sae}, %zmm7, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x18,0x08,0xef,0xab]
+          vrndscaleps $0xab,{sae}, %zmm7, %zmm13
+
+// CHECK: vrndscaleps $123, %zmm7, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0xef,0x7b]
+          vrndscaleps $0x7b, %zmm7, %zmm13
+
+// CHECK: vrndscaleps $123,{sae}, %zmm7, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x18,0x08,0xef,0x7b]
+          vrndscaleps $0x7b,{sae}, %zmm7, %zmm13
+
+// CHECK: vrndscaleps $123, (%rcx), %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0x29,0x7b]
+          vrndscaleps $0x7b, (%rcx), %zmm13
+
+// CHECK: vrndscaleps $123, 291(%rax,%r14,8), %zmm13
+// CHECK:  encoding: [0x62,0x33,0x7d,0x48,0x08,0xac,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 291(%rax,%r14,8), %zmm13
+
+// CHECK: vrndscaleps $123, (%rcx){1to16}, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x58,0x08,0x29,0x7b]
+          vrndscaleps $0x7b, (%rcx){1to16}, %zmm13
+
+// CHECK: vrndscaleps $123, 8128(%rdx), %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0x6a,0x7f,0x7b]
+          vrndscaleps $0x7b, 8128(%rdx), %zmm13
+
+// CHECK: vrndscaleps $123, 8192(%rdx), %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0xaa,0x00,0x20,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 8192(%rdx), %zmm13
+
+// CHECK: vrndscaleps $123, -8192(%rdx), %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0x6a,0x80,0x7b]
+          vrndscaleps $0x7b, -8192(%rdx), %zmm13
+
+// CHECK: vrndscaleps $123, -8256(%rdx), %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x48,0x08,0xaa,0xc0,0xdf,0xff,0xff,0x7b]
+          vrndscaleps $0x7b, -8256(%rdx), %zmm13
+
+// CHECK: vrndscaleps $123, 508(%rdx){1to16}, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x58,0x08,0x6a,0x7f,0x7b]
+          vrndscaleps $0x7b, 508(%rdx){1to16}, %zmm13
+
+// CHECK: vrndscaleps $123, 512(%rdx){1to16}, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x58,0x08,0xaa,0x00,0x02,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 512(%rdx){1to16}, %zmm13
+
+// CHECK: vrndscaleps $123, -512(%rdx){1to16}, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x58,0x08,0x6a,0x80,0x7b]
+          vrndscaleps $0x7b, -512(%rdx){1to16}, %zmm13
+
+// CHECK: vrndscaleps $123, -516(%rdx){1to16}, %zmm13
+// CHECK:  encoding: [0x62,0x73,0x7d,0x58,0x08,0xaa,0xfc,0xfd,0xff,0xff,0x7b]
+          vrndscaleps $0x7b, -516(%rdx){1to16}, %zmm13
+
+// CHECK: vrndscalesd $171, %xmm15, %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x43,0x9d,0x08,0x0b,0xcf,0xab]
+          vrndscalesd $0xab, %xmm15, %xmm12, %xmm25
+
+// CHECK: vrndscalesd $171, %xmm15, %xmm12, %xmm25 {%k6}
+// CHECK:  encoding: [0x62,0x43,0x9d,0x0e,0x0b,0xcf,0xab]
+          vrndscalesd $0xab, %xmm15, %xmm12, %xmm25 {%k6}
+
+// CHECK: vrndscalesd $171, %xmm15, %xmm12, %xmm25 {%k6} {z}
+// CHECK:  encoding: [0x62,0x43,0x9d,0x8e,0x0b,0xcf,0xab]
+          vrndscalesd $0xab, %xmm15, %xmm12, %xmm25 {%k6} {z}
+
+// CHECK: vrndscalesd $171, {sae}, %xmm15, %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x43,0x9d,0x18,0x0b,0xcf,0xab]
+          vrndscalesd $0xab,{sae}, %xmm15, %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, %xmm15, %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x43,0x9d,0x08,0x0b,0xcf,0x7b]
+          vrndscalesd $0x7b, %xmm15, %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, {sae}, %xmm15, %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x43,0x9d,0x18,0x0b,0xcf,0x7b]
+          vrndscalesd $0x7b,{sae}, %xmm15, %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, (%rcx), %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x63,0x9d,0x08,0x0b,0x09,0x7b]
+          vrndscalesd $0x7b, (%rcx), %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, 291(%rax,%r14,8), %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x23,0x9d,0x08,0x0b,0x8c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscalesd $0x7b, 291(%rax,%r14,8), %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, 1016(%rdx), %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x63,0x9d,0x08,0x0b,0x4a,0x7f,0x7b]
+          vrndscalesd $0x7b, 1016(%rdx), %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, 1024(%rdx), %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x63,0x9d,0x08,0x0b,0x8a,0x00,0x04,0x00,0x00,0x7b]
+          vrndscalesd $0x7b, 1024(%rdx), %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, -1024(%rdx), %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x63,0x9d,0x08,0x0b,0x4a,0x80,0x7b]
+          vrndscalesd $0x7b, -1024(%rdx), %xmm12, %xmm25
+
+// CHECK: vrndscalesd $123, -1032(%rdx), %xmm12, %xmm25
+// CHECK:  encoding: [0x62,0x63,0x9d,0x08,0x0b,0x8a,0xf8,0xfb,0xff,0xff,0x7b]
+          vrndscalesd $0x7b, -1032(%rdx), %xmm12, %xmm25
+
+// CHECK: vrndscaless $171, %xmm17, %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x33,0x25,0x08,0x0a,0xd9,0xab]
+          vrndscaless $0xab, %xmm17, %xmm11, %xmm11
+
+// CHECK: vrndscaless $171, %xmm17, %xmm11, %xmm11 {%k3}
+// CHECK:  encoding: [0x62,0x33,0x25,0x0b,0x0a,0xd9,0xab]
+          vrndscaless $0xab, %xmm17, %xmm11, %xmm11 {%k3}
+
+// CHECK: vrndscaless $171, %xmm17, %xmm11, %xmm11 {%k3} {z}
+// CHECK:  encoding: [0x62,0x33,0x25,0x8b,0x0a,0xd9,0xab]
+          vrndscaless $0xab, %xmm17, %xmm11, %xmm11 {%k3} {z}
+
+// CHECK: vrndscaless $171, {sae}, %xmm17, %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x33,0x25,0x18,0x0a,0xd9,0xab]
+          vrndscaless $0xab,{sae}, %xmm17, %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, %xmm17, %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x33,0x25,0x08,0x0a,0xd9,0x7b]
+          vrndscaless $0x7b, %xmm17, %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, {sae}, %xmm17, %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x33,0x25,0x18,0x0a,0xd9,0x7b]
+          vrndscaless $0x7b,{sae}, %xmm17, %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, (%rcx), %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x73,0x25,0x08,0x0a,0x19,0x7b]
+          vrndscaless $0x7b, (%rcx), %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, 291(%rax,%r14,8), %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x33,0x25,0x08,0x0a,0x9c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscaless $0x7b, 291(%rax,%r14,8), %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, 508(%rdx), %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x73,0x25,0x08,0x0a,0x5a,0x7f,0x7b]
+          vrndscaless $0x7b, 508(%rdx), %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, 512(%rdx), %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x73,0x25,0x08,0x0a,0x9a,0x00,0x02,0x00,0x00,0x7b]
+          vrndscaless $0x7b, 512(%rdx), %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, -512(%rdx), %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x73,0x25,0x08,0x0a,0x5a,0x80,0x7b]
+          vrndscaless $0x7b, -512(%rdx), %xmm11, %xmm11
+
+// CHECK: vrndscaless $123, -516(%rdx), %xmm11, %xmm11
+// CHECK:  encoding: [0x62,0x73,0x25,0x08,0x0a,0x9a,0xfc,0xfd,0xff,0xff,0x7b]
+          vrndscaless $0x7b, -516(%rdx), %xmm11, %xmm11
+
 // CHECK: vfmadd132ss %xmm22, %xmm17, %xmm30
 // CHECK:  encoding: [0x62,0x22,0x75,0x00,0x99,0xf6]
           vfmadd132ss %xmm22, %xmm17, %xmm30
 
 // CHECK: vfmadd132ss %xmm22, %xmm17, %xmm30 {%k3}
 // CHECK:  encoding: [0x62,0x22,0x75,0x03,0x99,0xf6]
           vfmadd132ss %xmm22, %xmm17, %xmm30 {%k3}
 

llvm/trunk/test/MC/X86/x86-64-avx512dq.s

 // CHECK: vrangess $123, -512(%rdx), %xmm24, %xmm25
 // CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x51,0x4a,0x80,0x7b]
           vrangess $0x7b,-512(%rdx), %xmm24, %xmm25
 
 // CHECK: vrangess $123, -516(%rdx), %xmm24, %xmm25
 // CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x51,0x8a,0xfc,0xfd,0xff,0xff,0x7b]
           vrangess $0x7b,-516(%rdx), %xmm24, %xmm25
 
+// CHECK: vreducepd $171, %zmm19, %zmm19
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x48,0x56,0xdb,0xab]
+          vreducepd $0xab, %zmm19, %zmm19
+
+// CHECK: vreducepd $171, %zmm19, %zmm19 {%k6}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x4e,0x56,0xdb,0xab]
+          vreducepd $0xab, %zmm19, %zmm19 {%k6}
+
+// CHECK: vreducepd $171, %zmm19, %zmm19 {%k6} {z}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0xce,0x56,0xdb,0xab]
+          vreducepd $0xab, %zmm19, %zmm19 {%k6} {z}
+
+// CHECK: vreducepd $171,{sae}, %zmm19, %zmm19
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x18,0x56,0xdb,0xab]
+          vreducepd $0xab,{sae}, %zmm19, %zmm19
+
+// CHECK: vreducepd $123, %zmm19, %zmm19
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x48,0x56,0xdb,0x7b]
+          vreducepd $0x7b, %zmm19, %zmm19
+
+// CHECK: vreducepd $123,{sae}, %zmm19, %zmm19
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x18,0x56,0xdb,0x7b]
+          vreducepd $0x7b,{sae}, %zmm19, %zmm19
+
+// CHECK: vreducepd $123, (%rcx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x19,0x7b]
+          vreducepd $0x7b,(%rcx), %zmm19
+
+// CHECK: vreducepd $123, 291(%rax,%r14,8), %zmm19
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x48,0x56,0x9c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreducepd $0x7b,291(%rax,%r14,8), %zmm19
+
+// CHECK: vreducepd $123, (%rcx){1to8}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x19,0x7b]
+          vreducepd $0x7b,(%rcx){1to8}, %zmm19
+
+// CHECK: vreducepd $123, 8128(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x5a,0x7f,0x7b]
+          vreducepd $0x7b,8128(%rdx), %zmm19
+
+// CHECK: vreducepd $123, 8192(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x9a,0x00,0x20,0x00,0x00,0x7b]
+          vreducepd $0x7b,8192(%rdx), %zmm19
+
+// CHECK: vreducepd $123, -8192(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x5a,0x80,0x7b]
+          vreducepd $0x7b,-8192(%rdx), %zmm19
+
+// CHECK: vreducepd $123, -8256(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x9a,0xc0,0xdf,0xff,0xff,0x7b]
+          vreducepd $0x7b,-8256(%rdx), %zmm19
+
+// CHECK: vreducepd $123, 1016(%rdx){1to8}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x5a,0x7f,0x7b]
+          vreducepd $0x7b,1016(%rdx){1to8}, %zmm19
+
+// CHECK: vreducepd $123, 1024(%rdx){1to8}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x9a,0x00,0x04,0x00,0x00,0x7b]
+          vreducepd $0x7b,1024(%rdx){1to8}, %zmm19
+
+// CHECK: vreducepd $123, -1024(%rdx){1to8}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x5a,0x80,0x7b]
+          vreducepd $0x7b,-1024(%rdx){1to8}, %zmm19
+
+// CHECK: vreducepd $123, -1032(%rdx){1to8}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x9a,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducepd $0x7b,-1032(%rdx){1to8}, %zmm19
+
+// CHECK: vreduceps $171, %zmm29, %zmm19
+// CHECK:  encoding: [0x62,0x83,0x7d,0x48,0x56,0xdd,0xab]
+          vreduceps $0xab, %zmm29, %zmm19
+
+// CHECK: vreduceps $171, %zmm29, %zmm19 {%k3}
+// CHECK:  encoding: [0x62,0x83,0x7d,0x4b,0x56,0xdd,0xab]
+          vreduceps $0xab, %zmm29, %zmm19 {%k3}
+
+// CHECK: vreduceps $171, %zmm29, %zmm19 {%k3} {z}
+// CHECK:  encoding: [0x62,0x83,0x7d,0xcb,0x56,0xdd,0xab]
+          vreduceps $0xab, %zmm29, %zmm19 {%k3} {z}
+
+// CHECK: vreduceps $171,{sae}, %zmm29, %zmm19
+// CHECK:  encoding: [0x62,0x83,0x7d,0x18,0x56,0xdd,0xab]
+          vreduceps $0xab,{sae}, %zmm29, %zmm19
+
+// CHECK: vreduceps $123, %zmm29, %zmm19
+// CHECK:  encoding: [0x62,0x83,0x7d,0x48,0x56,0xdd,0x7b]
+          vreduceps $0x7b, %zmm29, %zmm19
+
+// CHECK: vreduceps $123,{sae}, %zmm29, %zmm19
+// CHECK:  encoding: [0x62,0x83,0x7d,0x18,0x56,0xdd,0x7b]
+          vreduceps $0x7b,{sae}, %zmm29, %zmm19
+
+// CHECK: vreduceps $123, (%rcx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x56,0x19,0x7b]
+          vreduceps $0x7b,(%rcx), %zmm19
+
+// CHECK: vreduceps $123, 291(%rax,%r14,8), %zmm19
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x48,0x56,0x9c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreduceps $0x7b,291(%rax,%r14,8), %zmm19
+
+// CHECK: vreduceps $123, (%rcx){1to16}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x56,0x19,0x7b]
+          vreduceps $0x7b,(%rcx){1to16}, %zmm19
+
+// CHECK: vreduceps $123, 8128(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x56,0x5a,0x7f,0x7b]
+          vreduceps $0x7b,8128(%rdx), %zmm19
+
+// CHECK: vreduceps $123, 8192(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x56,0x9a,0x00,0x20,0x00,0x00,0x7b]
+          vreduceps $0x7b,8192(%rdx), %zmm19
+
+// CHECK: vreduceps $123, -8192(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x56,0x5a,0x80,0x7b]
+          vreduceps $0x7b,-8192(%rdx), %zmm19
+
+// CHECK: vreduceps $123, -8256(%rdx), %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x48,0x56,0x9a,0xc0,0xdf,0xff,0xff,0x7b]
+          vreduceps $0x7b,-8256(%rdx), %zmm19
+
+// CHECK: vreduceps $123, 508(%rdx){1to16}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x56,0x5a,0x7f,0x7b]
+          vreduceps $0x7b,508(%rdx){1to16}, %zmm19
+
+// CHECK: vreduceps $123, 512(%rdx){1to16}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x56,0x9a,0x00,0x02,0x00,0x00,0x7b]
+          vreduceps $0x7b,512(%rdx){1to16}, %zmm19
+
+// CHECK: vreduceps $123, -512(%rdx){1to16}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x56,0x5a,0x80,0x7b]
+          vreduceps $0x7b,-512(%rdx){1to16}, %zmm19
+
+// CHECK: vreduceps $123, -516(%rdx){1to16}, %zmm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x58,0x56,0x9a,0xfc,0xfd,0xff,0xff,0x7b]
+          vreduceps $0x7b,-516(%rdx){1to16}, %zmm19
+
+// CHECK: vreducesd $171, %xmm25, %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0x83,0xf5,0x00,0x57,0xc9,0xab]
+          vreducesd $0xab, %xmm25, %xmm17, %xmm17
+
+// CHECK: vreducesd $171, %xmm25, %xmm17, %xmm17 {%k6}
+// CHECK:  encoding: [0x62,0x83,0xf5,0x06,0x57,0xc9,0xab]
+          vreducesd $0xab, %xmm25, %xmm17, %xmm17 {%k6}
+
+// CHECK: vreducesd $171, %xmm25, %xmm17, %xmm17 {%k6} {z}
+// CHECK:  encoding: [0x62,0x83,0xf5,0x86,0x57,0xc9,0xab]
+          vreducesd $0xab, %xmm25, %xmm17, %xmm17 {%k6} {z}
+
+// CHECK: vreducesd $171,{sae}, %xmm25, %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0x83,0xf5,0x10,0x57,0xc9,0xab]
+          vreducesd $0xab,{sae}, %xmm25, %xmm17, %xmm17
+
+// CHECK: vreducesd $123, %xmm25, %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0x83,0xf5,0x00,0x57,0xc9,0x7b]
+          vreducesd $0x7b, %xmm25, %xmm17, %xmm17
+
+// CHECK: vreducesd $123,{sae}, %xmm25, %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0x83,0xf5,0x10,0x57,0xc9,0x7b]
+          vreducesd $0x7b,{sae}, %xmm25, %xmm17, %xmm17
+
+// CHECK: vreducesd $123, (%rcx), %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0xe3,0xf5,0x00,0x57,0x09,0x7b]
+          vreducesd $0x7b,(%rcx), %xmm17, %xmm17
+
+// CHECK: vreducesd $123, 291(%rax,%r14,8), %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0xa3,0xf5,0x00,0x57,0x8c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreducesd $0x7b,291(%rax,%r14,8), %xmm17, %xmm17
+
+// CHECK: vreducesd $123, 1016(%rdx), %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0xe3,0xf5,0x00,0x57,0x4a,0x7f,0x7b]
+          vreducesd $0x7b,1016(%rdx), %xmm17, %xmm17
+
+// CHECK: vreducesd $123, 1024(%rdx), %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0xe3,0xf5,0x00,0x57,0x8a,0x00,0x04,0x00,0x00,0x7b]
+          vreducesd $0x7b,1024(%rdx), %xmm17, %xmm17
+
+// CHECK: vreducesd $123, -1024(%rdx), %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0xe3,0xf5,0x00,0x57,0x4a,0x80,0x7b]
+          vreducesd $0x7b,-1024(%rdx), %xmm17, %xmm17
+
+// CHECK: vreducesd $123, -1032(%rdx), %xmm17, %xmm17
+// CHECK:  encoding: [0x62,0xe3,0xf5,0x00,0x57,0x8a,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducesd $0x7b,-1032(%rdx), %xmm17, %xmm17
+
+// CHECK: vreducess $171, %xmm21, %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x15,0x00,0x57,0xf5,0xab]
+          vreducess $0xab, %xmm21, %xmm29, %xmm30
+
+// CHECK: vreducess $171, %xmm21, %xmm29, %xmm30 {%k1}
+// CHECK:  encoding: [0x62,0x23,0x15,0x01,0x57,0xf5,0xab]
+          vreducess $0xab, %xmm21, %xmm29, %xmm30 {%k1}
+
+// CHECK: vreducess $171, %xmm21, %xmm29, %xmm30 {%k1} {z}
+// CHECK:  encoding: [0x62,0x23,0x15,0x81,0x57,0xf5,0xab]
+          vreducess $0xab, %xmm21, %xmm29, %xmm30 {%k1} {z}
+
+// CHECK: vreducess $171,{sae}, %xmm21, %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x15,0x10,0x57,0xf5,0xab]
+          vreducess $0xab,{sae}, %xmm21, %xmm29, %xmm30
+
+// CHECK: vreducess $123, %xmm21, %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x15,0x00,0x57,0xf5,0x7b]
+          vreducess $0x7b, %xmm21, %xmm29, %xmm30
+
+// CHECK: vreducess $123,{sae}, %xmm21, %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x15,0x10,0x57,0xf5,0x7b]
+          vreducess $0x7b,{sae}, %xmm21, %xmm29, %xmm30
+
+// CHECK: vreducess $123, (%rcx), %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x15,0x00,0x57,0x31,0x7b]
+          vreducess $0x7b,(%rcx), %xmm29, %xmm30
+
+// CHECK: vreducess $123, 291(%rax,%r14,8), %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x15,0x00,0x57,0xb4,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreducess $0x7b,291(%rax,%r14,8), %xmm29, %xmm30
+
+// CHECK: vreducess $123, 508(%rdx), %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x15,0x00,0x57,0x72,0x7f,0x7b]
+          vreducess $0x7b,508(%rdx), %xmm29, %xmm30
+
+// CHECK: vreducess $123, 512(%rdx), %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x15,0x00,0x57,0xb2,0x00,0x02,0x00,0x00,0x7b]
+          vreducess $0x7b,512(%rdx), %xmm29, %xmm30
+
+// CHECK: vreducess $123, -512(%rdx), %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x15,0x00,0x57,0x72,0x80,0x7b]
+          vreducess $0x7b,-512(%rdx), %xmm29, %xmm30
+
+// CHECK: vreducess $123, -516(%rdx), %xmm29, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x15,0x00,0x57,0xb2,0xfc,0xfd,0xff,0xff,0x7b]
+          vreducess $0x7b,-516(%rdx), %xmm29, %xmm30
+
+// CHECK: vreducepd $171, %zmm28, %zmm18
+// CHECK:  encoding: [0x62,0x83,0xfd,0x48,0x56,0xd4,0xab]
+          vreducepd $0xab, %zmm28, %zmm18
+
+// CHECK: vreducepd $171, %zmm28, %zmm18 {%k5}
+// CHECK:  encoding: [0x62,0x83,0xfd,0x4d,0x56,0xd4,0xab]
+          vreducepd $0xab, %zmm28, %zmm18 {%k5}
+
+// CHECK: vreducepd $171, %zmm28, %zmm18 {%k5} {z}
+// CHECK:  encoding: [0x62,0x83,0xfd,0xcd,0x56,0xd4,0xab]
+          vreducepd $0xab, %zmm28, %zmm18 {%k5} {z}
+
+// CHECK: vreducepd $171,{sae}, %zmm28, %zmm18
+// CHECK:  encoding: [0x62,0x83,0xfd,0x18,0x56,0xd4,0xab]
+          vreducepd $0xab,{sae}, %zmm28, %zmm18
+
+// CHECK: vreducepd $123, %zmm28, %zmm18
+// CHECK:  encoding: [0x62,0x83,0xfd,0x48,0x56,0xd4,0x7b]
+          vreducepd $0x7b, %zmm28, %zmm18
+
+// CHECK: vreducepd $123,{sae}, %zmm28, %zmm18
+// CHECK:  encoding: [0x62,0x83,0xfd,0x18,0x56,0xd4,0x7b]
+          vreducepd $0x7b,{sae}, %zmm28, %zmm18
+
+// CHECK: vreducepd $123, (%rcx), %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x11,0x7b]
+          vreducepd $0x7b,(%rcx), %zmm18
+
+// CHECK: vreducepd $123, 4660(%rax,%r14,8), %zmm18
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x48,0x56,0x94,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreducepd $0x7b,4660(%rax,%r14,8), %zmm18
+
+// CHECK: vreducepd $123, (%rcx){1to8}, %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x11,0x7b]
+          vreducepd $0x7b,(%rcx){1to8}, %zmm18
+
+// CHECK: vreducepd $123, 8128(%rdx), %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x52,0x7f,0x7b]
+          vreducepd $0x7b,8128(%rdx), %zmm18
+
+// CHECK: vreducepd $123, 8192(%rdx), %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x92,0x00,0x20,0x00,0x00,0x7b]
+          vreducepd $0x7b,8192(%rdx), %zmm18
+
+// CHECK: vreducepd $123, -8192(%rdx), %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x52,0x80,0x7b]
+          vreducepd $0x7b,-8192(%rdx), %zmm18
+
+// CHECK: vreducepd $123, -8256(%rdx), %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x48,0x56,0x92,0xc0,0xdf,0xff,0xff,0x7b]
+          vreducepd $0x7b,-8256(%rdx), %zmm18
+
+// CHECK: vreducepd $123, 1016(%rdx){1to8}, %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x52,0x7f,0x7b]
+          vreducepd $0x7b,1016(%rdx){1to8}, %zmm18
+
+// CHECK: vreducepd $123, 1024(%rdx){1to8}, %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x92,0x00,0x04,0x00,0x00,0x7b]
+          vreducepd $0x7b,1024(%rdx){1to8}, %zmm18
+
+// CHECK: vreducepd $123, -1024(%rdx){1to8}, %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x52,0x80,0x7b]
+          vreducepd $0x7b,-1024(%rdx){1to8}, %zmm18
+
+// CHECK: vreducepd $123, -1032(%rdx){1to8}, %zmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x58,0x56,0x92,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducepd $0x7b,-1032(%rdx){1to8}, %zmm18
+
+// CHECK: vreduceps $171, %zmm25, %zmm26
+// CHECK:  encoding: [0x62,0x03,0x7d,0x48,0x56,0xd1,0xab]
+          vreduceps $0xab, %zmm25, %zmm26
+
+// CHECK: vreduceps $171, %zmm25, %zmm26 {%k3}
+// CHECK:  encoding: [0x62,0x03,0x7d,0x4b,0x56,0xd1,0xab]
+          vreduceps $0xab, %zmm25, %zmm26 {%k3}
+
+// CHECK: vreduceps $171, %zmm25, %zmm26 {%k3} {z}
+// CHECK:  encoding: [0x62,0x03,0x7d,0xcb,0x56,0xd1,0xab]
+          vreduceps $0xab, %zmm25, %zmm26 {%k3} {z}
+
+// CHECK: vreduceps $171,{sae}, %zmm25, %zmm26
+// CHECK:  encoding: [0x62,0x03,0x7d,0x18,0x56,0xd1,0xab]
+          vreduceps $0xab,{sae}, %zmm25, %zmm26
+
+// CHECK: vreduceps $123, %zmm25, %zmm26
+// CHECK:  encoding: [0x62,0x03,0x7d,0x48,0x56,0xd1,0x7b]
+          vreduceps $0x7b, %zmm25, %zmm26
+
+// CHECK: vreduceps $123,{sae}, %zmm25, %zmm26
+// CHECK:  encoding: [0x62,0x03,0x7d,0x18,0x56,0xd1,0x7b]
+          vreduceps $0x7b,{sae}, %zmm25, %zmm26
+
+// CHECK: vreduceps $123, (%rcx), %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x48,0x56,0x11,0x7b]
+          vreduceps $0x7b,(%rcx), %zmm26
+
+// CHECK: vreduceps $123, 4660(%rax,%r14,8), %zmm26
+// CHECK:  encoding: [0x62,0x23,0x7d,0x48,0x56,0x94,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreduceps $0x7b,4660(%rax,%r14,8), %zmm26
+
+// CHECK: vreduceps $123, (%rcx){1to16}, %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x58,0x56,0x11,0x7b]
+          vreduceps $0x7b,(%rcx){1to16}, %zmm26
+
+// CHECK: vreduceps $123, 8128(%rdx), %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x48,0x56,0x52,0x7f,0x7b]
+          vreduceps $0x7b,8128(%rdx), %zmm26
+
+// CHECK: vreduceps $123, 8192(%rdx), %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x48,0x56,0x92,0x00,0x20,0x00,0x00,0x7b]
+          vreduceps $0x7b,8192(%rdx), %zmm26
+
+// CHECK: vreduceps $123, -8192(%rdx), %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x48,0x56,0x52,0x80,0x7b]
+          vreduceps $0x7b,-8192(%rdx), %zmm26
+
+// CHECK: vreduceps $123, -8256(%rdx), %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x48,0x56,0x92,0xc0,0xdf,0xff,0xff,0x7b]
+          vreduceps $0x7b,-8256(%rdx), %zmm26
+
+// CHECK: vreduceps $123, 508(%rdx){1to16}, %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x58,0x56,0x52,0x7f,0x7b]
+          vreduceps $0x7b,508(%rdx){1to16}, %zmm26
+
+// CHECK: vreduceps $123, 512(%rdx){1to16}, %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x58,0x56,0x92,0x00,0x02,0x00,0x00,0x7b]
+          vreduceps $0x7b,512(%rdx){1to16}, %zmm26
+
+// CHECK: vreduceps $123, -512(%rdx){1to16}, %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x58,0x56,0x52,0x80,0x7b]
+          vreduceps $0x7b,-512(%rdx){1to16}, %zmm26
+
+// CHECK: vreduceps $123, -516(%rdx){1to16}, %zmm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x58,0x56,0x92,0xfc,0xfd,0xff,0xff,0x7b]
+          vreduceps $0x7b,-516(%rdx){1to16}, %zmm26
+
+// CHECK: vreducesd $171, %xmm24, %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x03,0xe5,0x00,0x57,0xc8,0xab]
+          vreducesd $0xab, %xmm24, %xmm19, %xmm25
+
+// CHECK: vreducesd $171, %xmm24, %xmm19, %xmm25 {%k3}
+// CHECK:  encoding: [0x62,0x03,0xe5,0x03,0x57,0xc8,0xab]
+          vreducesd $0xab, %xmm24, %xmm19, %xmm25 {%k3}
+
+// CHECK: vreducesd $171, %xmm24, %xmm19, %xmm25 {%k3} {z}
+// CHECK:  encoding: [0x62,0x03,0xe5,0x83,0x57,0xc8,0xab]
+          vreducesd $0xab, %xmm24, %xmm19, %xmm25 {%k3} {z}
+
+// CHECK: vreducesd $171,{sae}, %xmm24, %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x03,0xe5,0x10,0x57,0xc8,0xab]
+          vreducesd $0xab,{sae}, %xmm24, %xmm19, %xmm25
+
+// CHECK: vreducesd $123, %xmm24, %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x03,0xe5,0x00,0x57,0xc8,0x7b]
+          vreducesd $0x7b, %xmm24, %xmm19, %xmm25
+
+// CHECK: vreducesd $123,{sae}, %xmm24, %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x03,0xe5,0x10,0x57,0xc8,0x7b]
+          vreducesd $0x7b,{sae}, %xmm24, %xmm19, %xmm25
+
+// CHECK: vreducesd $123, (%rcx), %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xe5,0x00,0x57,0x09,0x7b]
+          vreducesd $0x7b,(%rcx), %xmm19, %xmm25
+
+// CHECK: vreducesd $123, 4660(%rax,%r14,8), %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x23,0xe5,0x00,0x57,0x8c,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreducesd $0x7b,4660(%rax,%r14,8), %xmm19, %xmm25
+
+// CHECK: vreducesd $123, 1016(%rdx), %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xe5,0x00,0x57,0x4a,0x7f,0x7b]
+          vreducesd $0x7b,1016(%rdx), %xmm19, %xmm25
+
+// CHECK: vreducesd $123, 1024(%rdx), %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xe5,0x00,0x57,0x8a,0x00,0x04,0x00,0x00,0x7b]
+          vreducesd $0x7b,1024(%rdx), %xmm19, %xmm25
+
+// CHECK: vreducesd $123, -1024(%rdx), %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xe5,0x00,0x57,0x4a,0x80,0x7b]
+          vreducesd $0x7b,-1024(%rdx), %xmm19, %xmm25
+
+// CHECK: vreducesd $123, -1032(%rdx), %xmm19, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xe5,0x00,0x57,0x8a,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducesd $0x7b,-1032(%rdx), %xmm19, %xmm25
+
+// CHECK: vreducess $171, %xmm21, %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x3d,0x00,0x57,0xf5,0xab]
+          vreducess $0xab, %xmm21, %xmm24, %xmm30
+
+// CHECK: vreducess $171, %xmm21, %xmm24, %xmm30 {%k2}
+// CHECK:  encoding: [0x62,0x23,0x3d,0x02,0x57,0xf5,0xab]
+          vreducess $0xab, %xmm21, %xmm24, %xmm30 {%k2}
+
+// CHECK: vreducess $171, %xmm21, %xmm24, %xmm30 {%k2} {z}
+// CHECK:  encoding: [0x62,0x23,0x3d,0x82,0x57,0xf5,0xab]
+          vreducess $0xab, %xmm21, %xmm24, %xmm30 {%k2} {z}
+
+// CHECK: vreducess $171,{sae}, %xmm21, %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x3d,0x10,0x57,0xf5,0xab]
+          vreducess $0xab,{sae}, %xmm21, %xmm24, %xmm30
+
+// CHECK: vreducess $123, %xmm21, %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x3d,0x00,0x57,0xf5,0x7b]
+          vreducess $0x7b, %xmm21, %xmm24, %xmm30
+
+// CHECK: vreducess $123,{sae}, %xmm21, %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x3d,0x10,0x57,0xf5,0x7b]
+          vreducess $0x7b,{sae}, %xmm21, %xmm24, %xmm30
+
+// CHECK: vreducess $123, (%rcx), %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x57,0x31,0x7b]
+          vreducess $0x7b,(%rcx), %xmm24, %xmm30
+
+// CHECK: vreducess $123, 4660(%rax,%r14,8), %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x23,0x3d,0x00,0x57,0xb4,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreducess $0x7b,4660(%rax,%r14,8), %xmm24, %xmm30
+
+// CHECK: vreducess $123, 508(%rdx), %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x57,0x72,0x7f,0x7b]
+          vreducess $0x7b,508(%rdx), %xmm24, %xmm30
+
+// CHECK: vreducess $123, 512(%rdx), %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x57,0xb2,0x00,0x02,0x00,0x00,0x7b]
+          vreducess $0x7b,512(%rdx), %xmm24, %xmm30
+
+// CHECK: vreducess $123, -512(%rdx), %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x57,0x72,0x80,0x7b]
+          vreducess $0x7b,-512(%rdx), %xmm24, %xmm30
+
+// CHECK: vreducess $123, -516(%rdx), %xmm24, %xmm30
+// CHECK:  encoding: [0x62,0x63,0x3d,0x00,0x57,0xb2,0xfc,0xfd,0xff,0xff,0x7b]
+          vreducess $0x7b,-516(%rdx), %xmm24, %xmm30
+
 // CHECK: vcvtpd2qq %zmm29, %zmm18
 // CHECK:  encoding: [0x62,0x81,0xfd,0x48,0x7b,0xd5]
           vcvtpd2qq %zmm29, %zmm18
 
 // CHECK: vcvtpd2qq %zmm29, %zmm18 {%k6}
 // CHECK:  encoding: [0x62,0x81,0xfd,0x4e,0x7b,0xd5]
           vcvtpd2qq %zmm29, %zmm18 {%k6}
 

llvm/trunk/test/MC/X86/x86-64-avx512dq_vl.s

 // CHECK: vrangeps $123, -512(%rdx){1to8}, %ymm23, %ymm24
 // CHECK:  encoding: [0x62,0x63,0x45,0x30,0x50,0x42,0x80,0x7b]
           vrangeps $0x7b,-512(%rdx){1to8}, %ymm23, %ymm24
 
 // CHECK: vrangeps $123, -516(%rdx){1to8}, %ymm23, %ymm24
 // CHECK:  encoding: [0x62,0x63,0x45,0x30,0x50,0x82,0xfc,0xfd,0xff,0xff,0x7b]
           vrangeps $0x7b,-516(%rdx){1to8}, %ymm23, %ymm24
 
+// CHECK: vreducepd $171, %xmm17, %xmm18
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x08,0x56,0xd1,0xab]
+          vreducepd $0xab, %xmm17, %xmm18
+
+// CHECK: vreducepd $171, %xmm17, %xmm18 {%k3}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x0b,0x56,0xd1,0xab]
+          vreducepd $0xab, %xmm17, %xmm18 {%k3}
+
+// CHECK: vreducepd $171, %xmm17, %xmm18 {%k3} {z}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x8b,0x56,0xd1,0xab]
+          vreducepd $0xab, %xmm17, %xmm18 {%k3} {z}
+
+// CHECK: vreducepd $123, %xmm17, %xmm18
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x08,0x56,0xd1,0x7b]
+          vreducepd $0x7b, %xmm17, %xmm18
+
+// CHECK: vreducepd $123, (%rcx), %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x08,0x56,0x11,0x7b]
+          vreducepd $0x7b,(%rcx), %xmm18
+
+// CHECK: vreducepd $123, 291(%rax,%r14,8), %xmm18
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x08,0x56,0x94,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreducepd $0x7b,291(%rax,%r14,8), %xmm18
+
+// CHECK: vreducepd $171, %xmm28, %xmm25
+// CHECK:  encoding: [0x62,0x03,0xfd,0x08,0x56,0xcc,0xab]
+          vreducepd $0xab, %xmm28, %xmm25
+
+// CHECK: vreducepd $171, %xmm28, %xmm25 {%k4}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x0c,0x56,0xcc,0xab]
+          vreducepd $0xab, %xmm28, %xmm25 {%k4}
+
+// CHECK: vreducepd $171, %xmm28, %xmm25 {%k4} {z}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x8c,0x56,0xcc,0xab]
+          vreducepd $0xab, %xmm28, %xmm25 {%k4} {z}
+
+// CHECK: vreducepd $123, %xmm28, %xmm25
+// CHECK:  encoding: [0x62,0x03,0xfd,0x08,0x56,0xcc,0x7b]
+          vreducepd $0x7b, %xmm28, %xmm25
+
+// CHECK: vreducepd $123, (%rcx), %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x56,0x09,0x7b]
+          vreducepd $0x7b,(%rcx), %xmm25
+
+// CHECK: vreducepd $123, 4660(%rax,%r14,8), %xmm25
+// CHECK:  encoding: [0x62,0x23,0xfd,0x08,0x56,0x8c,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreducepd $0x7b,4660(%rax,%r14,8), %xmm25
+
+// CHECK: vreducepd $123, (%rcx){1to2}, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x56,0x09,0x7b]
+          vreducepd $0x7b,(%rcx){1to2}, %xmm25
+
+// CHECK: vreducepd $123, 2032(%rdx), %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x56,0x4a,0x7f,0x7b]
+          vreducepd $0x7b,2032(%rdx), %xmm25
+
+// CHECK: vreducepd $123, 2048(%rdx), %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x56,0x8a,0x00,0x08,0x00,0x00,0x7b]
+          vreducepd $0x7b,2048(%rdx), %xmm25
+
+// CHECK: vreducepd $123, -2048(%rdx), %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x56,0x4a,0x80,0x7b]
+          vreducepd $0x7b,-2048(%rdx), %xmm25
+
+// CHECK: vreducepd $123, -2064(%rdx), %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x56,0x8a,0xf0,0xf7,0xff,0xff,0x7b]
+          vreducepd $0x7b,-2064(%rdx), %xmm25
+
+// CHECK: vreducepd $123, 1016(%rdx){1to2}, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x56,0x4a,0x7f,0x7b]
+          vreducepd $0x7b,1016(%rdx){1to2}, %xmm25
+
+// CHECK: vreducepd $123, 1024(%rdx){1to2}, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x56,0x8a,0x00,0x04,0x00,0x00,0x7b]
+          vreducepd $0x7b,1024(%rdx){1to2}, %xmm25
+
+// CHECK: vreducepd $123, -1024(%rdx){1to2}, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x56,0x4a,0x80,0x7b]
+          vreducepd $0x7b,-1024(%rdx){1to2}, %xmm25
+
+// CHECK: vreducepd $123, -1032(%rdx){1to2}, %xmm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x56,0x8a,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducepd $0x7b,-1032(%rdx){1to2}, %xmm25
+
+// CHECK: vreducepd $171, %ymm17, %ymm28
+// CHECK:  encoding: [0x62,0x23,0xfd,0x28,0x56,0xe1,0xab]
+          vreducepd $0xab, %ymm17, %ymm28
+
+// CHECK: vreducepd $171, %ymm17, %ymm28 {%k4}
+// CHECK:  encoding: [0x62,0x23,0xfd,0x2c,0x56,0xe1,0xab]
+          vreducepd $0xab, %ymm17, %ymm28 {%k4}
+
+// CHECK: vreducepd $171, %ymm17, %ymm28 {%k4} {z}
+// CHECK:  encoding: [0x62,0x23,0xfd,0xac,0x56,0xe1,0xab]
+          vreducepd $0xab, %ymm17, %ymm28 {%k4} {z}
+
+// CHECK: vreducepd $123, %ymm17, %ymm28
+// CHECK:  encoding: [0x62,0x23,0xfd,0x28,0x56,0xe1,0x7b]
+          vreducepd $0x7b, %ymm17, %ymm28
+
+// CHECK: vreducepd $123, (%rcx), %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x21,0x7b]
+          vreducepd $0x7b,(%rcx), %ymm28
+
+// CHECK: vreducepd $123, 4660(%rax,%r14,8), %ymm28
+// CHECK:  encoding: [0x62,0x23,0xfd,0x28,0x56,0xa4,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreducepd $0x7b,4660(%rax,%r14,8), %ymm28
+
+// CHECK: vreducepd $123, (%rcx){1to4}, %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x21,0x7b]
+          vreducepd $0x7b,(%rcx){1to4}, %ymm28
+
+// CHECK: vreducepd $123, 4064(%rdx), %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x62,0x7f,0x7b]
+          vreducepd $0x7b,4064(%rdx), %ymm28
+
+// CHECK: vreducepd $123, 4096(%rdx), %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0xa2,0x00,0x10,0x00,0x00,0x7b]
+          vreducepd $0x7b,4096(%rdx), %ymm28
+
+// CHECK: vreducepd $123, -4096(%rdx), %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x62,0x80,0x7b]
+          vreducepd $0x7b,-4096(%rdx), %ymm28
+
+// CHECK: vreducepd $123, -4128(%rdx), %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0xa2,0xe0,0xef,0xff,0xff,0x7b]
+          vreducepd $0x7b,-4128(%rdx), %ymm28
+
+// CHECK: vreducepd $123, 1016(%rdx){1to4}, %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x62,0x7f,0x7b]
+          vreducepd $0x7b,1016(%rdx){1to4}, %ymm28
+
+// CHECK: vreducepd $123, 1024(%rdx){1to4}, %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0xa2,0x00,0x04,0x00,0x00,0x7b]
+          vreducepd $0x7b,1024(%rdx){1to4}, %ymm28
+
+// CHECK: vreducepd $123, -1024(%rdx){1to4}, %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x62,0x80,0x7b]
+          vreducepd $0x7b,-1024(%rdx){1to4}, %ymm28
+
+// CHECK: vreducepd $123, -1032(%rdx){1to4}, %ymm28
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0xa2,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducepd $0x7b,-1032(%rdx){1to4}, %ymm28
+
+// CHECK: vreduceps $171, %xmm21, %xmm29
+// CHECK:  encoding: [0x62,0x23,0x7d,0x08,0x56,0xed,0xab]
+          vreduceps $0xab, %xmm21, %xmm29
+
+// CHECK: vreduceps $171, %xmm21, %xmm29 {%k7}
+// CHECK:  encoding: [0x62,0x23,0x7d,0x0f,0x56,0xed,0xab]
+          vreduceps $0xab, %xmm21, %xmm29 {%k7}
+
+// CHECK: vreduceps $171, %xmm21, %xmm29 {%k7} {z}
+// CHECK:  encoding: [0x62,0x23,0x7d,0x8f,0x56,0xed,0xab]
+          vreduceps $0xab, %xmm21, %xmm29 {%k7} {z}
+
+// CHECK: vreduceps $123, %xmm21, %xmm29
+// CHECK:  encoding: [0x62,0x23,0x7d,0x08,0x56,0xed,0x7b]
+          vreduceps $0x7b, %xmm21, %xmm29
+
+// CHECK: vreduceps $123, (%rcx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x56,0x29,0x7b]
+          vreduceps $0x7b,(%rcx), %xmm29
+
+// CHECK: vreduceps $123, 4660(%rax,%r14,8), %xmm29
+// CHECK:  encoding: [0x62,0x23,0x7d,0x08,0x56,0xac,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreduceps $0x7b,4660(%rax,%r14,8), %xmm29
+
+// CHECK: vreduceps $123, (%rcx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x56,0x29,0x7b]
+          vreduceps $0x7b,(%rcx){1to4}, %xmm29
+
+// CHECK: vreduceps $123, 2032(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x56,0x6a,0x7f,0x7b]
+          vreduceps $0x7b,2032(%rdx), %xmm29
+
+// CHECK: vreduceps $123, 2048(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x56,0xaa,0x00,0x08,0x00,0x00,0x7b]
+          vreduceps $0x7b,2048(%rdx), %xmm29
+
+// CHECK: vreduceps $123, -2048(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x56,0x6a,0x80,0x7b]
+          vreduceps $0x7b,-2048(%rdx), %xmm29
+
+// CHECK: vreduceps $123, -2064(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x08,0x56,0xaa,0xf0,0xf7,0xff,0xff,0x7b]
+          vreduceps $0x7b,-2064(%rdx), %xmm29
+
+// CHECK: vreduceps $123, 508(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x56,0x6a,0x7f,0x7b]
+          vreduceps $0x7b,508(%rdx){1to4}, %xmm29
+
+// CHECK: vreduceps $123, 512(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x56,0xaa,0x00,0x02,0x00,0x00,0x7b]
+          vreduceps $0x7b,512(%rdx){1to4}, %xmm29
+
+// CHECK: vreduceps $123, -512(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x56,0x6a,0x80,0x7b]
+          vreduceps $0x7b,-512(%rdx){1to4}, %xmm29
+
+// CHECK: vreduceps $123, -516(%rdx){1to4}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0x7d,0x18,0x56,0xaa,0xfc,0xfd,0xff,0xff,0x7b]
+          vreduceps $0x7b,-516(%rdx){1to4}, %xmm29
+
+// CHECK: vreduceps $171, %ymm23, %ymm25
+// CHECK:  encoding: [0x62,0x23,0x7d,0x28,0x56,0xcf,0xab]
+          vreduceps $0xab, %ymm23, %ymm25
+
+// CHECK: vreduceps $171, %ymm23, %ymm25 {%k3}
+// CHECK:  encoding: [0x62,0x23,0x7d,0x2b,0x56,0xcf,0xab]
+          vreduceps $0xab, %ymm23, %ymm25 {%k3}
+
+// CHECK: vreduceps $171, %ymm23, %ymm25 {%k3} {z}
+// CHECK:  encoding: [0x62,0x23,0x7d,0xab,0x56,0xcf,0xab]
+          vreduceps $0xab, %ymm23, %ymm25 {%k3} {z}
+
+// CHECK: vreduceps $123, %ymm23, %ymm25
+// CHECK:  encoding: [0x62,0x23,0x7d,0x28,0x56,0xcf,0x7b]
+          vreduceps $0x7b, %ymm23, %ymm25
+
+// CHECK: vreduceps $123, (%rcx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x09,0x7b]
+          vreduceps $0x7b,(%rcx), %ymm25
+
+// CHECK: vreduceps $123, 4660(%rax,%r14,8), %ymm25
+// CHECK:  encoding: [0x62,0x23,0x7d,0x28,0x56,0x8c,0xf0,0x34,0x12,0x00,0x00,0x7b]
+          vreduceps $0x7b,4660(%rax,%r14,8), %ymm25
+
+// CHECK: vreduceps $123, (%rcx){1to8}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x09,0x7b]
+          vreduceps $0x7b,(%rcx){1to8}, %ymm25
+
+// CHECK: vreduceps $123, 4064(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x4a,0x7f,0x7b]
+          vreduceps $0x7b,4064(%rdx), %ymm25
+
+// CHECK: vreduceps $123, 4096(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x8a,0x00,0x10,0x00,0x00,0x7b]
+          vreduceps $0x7b,4096(%rdx), %ymm25
+
+// CHECK: vreduceps $123, -4096(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x4a,0x80,0x7b]
+          vreduceps $0x7b,-4096(%rdx), %ymm25
+
+// CHECK: vreduceps $123, -4128(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x8a,0xe0,0xef,0xff,0xff,0x7b]
+          vreduceps $0x7b,-4128(%rdx), %ymm25
+
+// CHECK: vreduceps $123, 508(%rdx){1to8}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x4a,0x7f,0x7b]
+          vreduceps $0x7b,508(%rdx){1to8}, %ymm25
+
+// CHECK: vreduceps $123, 512(%rdx){1to8}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x8a,0x00,0x02,0x00,0x00,0x7b]
+          vreduceps $0x7b,512(%rdx){1to8}, %ymm25
+
+// CHECK: vreduceps $123, -512(%rdx){1to8}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x4a,0x80,0x7b]
+          vreduceps $0x7b,-512(%rdx){1to8}, %ymm25
+
+// CHECK: vreduceps $123, -516(%rdx){1to8}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x8a,0xfc,0xfd,0xff,0xff,0x7b]
+          vreduceps $0x7b,-516(%rdx){1to8}, %ymm25
+
+// CHECK: vreducepd $123, (%rcx){1to2}, %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x56,0x11,0x7b]
+          vreducepd $0x7b,(%rcx){1to2}, %xmm18
+
+// CHECK: vreducepd $123, 2032(%rdx), %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x08,0x56,0x52,0x7f,0x7b]
+          vreducepd $0x7b,2032(%rdx), %xmm18
+
+// CHECK: vreducepd $123, 2048(%rdx), %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x08,0x56,0x92,0x00,0x08,0x00,0x00,0x7b]
+          vreducepd $0x7b,2048(%rdx), %xmm18
+
+// CHECK: vreducepd $123, -2048(%rdx), %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x08,0x56,0x52,0x80,0x7b]
+          vreducepd $0x7b,-2048(%rdx), %xmm18
+
+// CHECK: vreducepd $123, -2064(%rdx), %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x08,0x56,0x92,0xf0,0xf7,0xff,0xff,0x7b]
+          vreducepd $0x7b,-2064(%rdx), %xmm18
+
+// CHECK: vreducepd $123, 1016(%rdx){1to2}, %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x56,0x52,0x7f,0x7b]
+          vreducepd $0x7b,1016(%rdx){1to2}, %xmm18
+
+// CHECK: vreducepd $123, 1024(%rdx){1to2}, %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x56,0x92,0x00,0x04,0x00,0x00,0x7b]
+          vreducepd $0x7b,1024(%rdx){1to2}, %xmm18
+
+// CHECK: vreducepd $123, -1024(%rdx){1to2}, %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x56,0x52,0x80,0x7b]
+          vreducepd $0x7b,-1024(%rdx){1to2}, %xmm18
+
+// CHECK: vreducepd $123, -1032(%rdx){1to2}, %xmm18
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x18,0x56,0x92,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducepd $0x7b,-1032(%rdx){1to2}, %xmm18
+
+// CHECK: vreducepd $171, %ymm29, %ymm25
+// CHECK:  encoding: [0x62,0x03,0xfd,0x28,0x56,0xcd,0xab]
+          vreducepd $0xab, %ymm29, %ymm25
+
+// CHECK: vreducepd $171, %ymm29, %ymm25 {%k1}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x29,0x56,0xcd,0xab]
+          vreducepd $0xab, %ymm29, %ymm25 {%k1}
+
+// CHECK: vreducepd $171, %ymm29, %ymm25 {%k1} {z}
+// CHECK:  encoding: [0x62,0x03,0xfd,0xa9,0x56,0xcd,0xab]
+          vreducepd $0xab, %ymm29, %ymm25 {%k1} {z}
+
+// CHECK: vreducepd $123, %ymm29, %ymm25
+// CHECK:  encoding: [0x62,0x03,0xfd,0x28,0x56,0xcd,0x7b]
+          vreducepd $0x7b, %ymm29, %ymm25
+
+// CHECK: vreducepd $123, (%rcx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x09,0x7b]
+          vreducepd $0x7b,(%rcx), %ymm25
+
+// CHECK: vreducepd $123, 291(%rax,%r14,8), %ymm25
+// CHECK:  encoding: [0x62,0x23,0xfd,0x28,0x56,0x8c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreducepd $0x7b,291(%rax,%r14,8), %ymm25
+
+// CHECK: vreducepd $123, (%rcx){1to4}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x09,0x7b]
+          vreducepd $0x7b,(%rcx){1to4}, %ymm25
+
+// CHECK: vreducepd $123, 4064(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x4a,0x7f,0x7b]
+          vreducepd $0x7b,4064(%rdx), %ymm25
+
+// CHECK: vreducepd $123, 4096(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x8a,0x00,0x10,0x00,0x00,0x7b]
+          vreducepd $0x7b,4096(%rdx), %ymm25
+
+// CHECK: vreducepd $123, -4096(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x4a,0x80,0x7b]
+          vreducepd $0x7b,-4096(%rdx), %ymm25
+
+// CHECK: vreducepd $123, -4128(%rdx), %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x28,0x56,0x8a,0xe0,0xef,0xff,0xff,0x7b]
+          vreducepd $0x7b,-4128(%rdx), %ymm25
+
+// CHECK: vreducepd $123, 1016(%rdx){1to4}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x4a,0x7f,0x7b]
+          vreducepd $0x7b,1016(%rdx){1to4}, %ymm25
+
+// CHECK: vreducepd $123, 1024(%rdx){1to4}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x8a,0x00,0x04,0x00,0x00,0x7b]
+          vreducepd $0x7b,1024(%rdx){1to4}, %ymm25
+
+// CHECK: vreducepd $123, -1024(%rdx){1to4}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x4a,0x80,0x7b]
+          vreducepd $0x7b,-1024(%rdx){1to4}, %ymm25
+
+// CHECK: vreducepd $123, -1032(%rdx){1to4}, %ymm25
+// CHECK:  encoding: [0x62,0x63,0xfd,0x38,0x56,0x8a,0xf8,0xfb,0xff,0xff,0x7b]
+          vreducepd $0x7b,-1032(%rdx){1to4}, %ymm25
+
+// CHECK: vreduceps $171, %xmm23, %xmm20
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x08,0x56,0xe7,0xab]
+          vreduceps $0xab, %xmm23, %xmm20
+
+// CHECK: vreduceps $171, %xmm23, %xmm20 {%k7}
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x0f,0x56,0xe7,0xab]
+          vreduceps $0xab, %xmm23, %xmm20 {%k7}
+
+// CHECK: vreduceps $171, %xmm23, %xmm20 {%k7} {z}
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x8f,0x56,0xe7,0xab]
+          vreduceps $0xab, %xmm23, %xmm20 {%k7} {z}
+
+// CHECK: vreduceps $123, %xmm23, %xmm20
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x08,0x56,0xe7,0x7b]
+          vreduceps $0x7b, %xmm23, %xmm20
+
+// CHECK: vreduceps $123, (%rcx), %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x56,0x21,0x7b]
+          vreduceps $0x7b,(%rcx), %xmm20
+
+// CHECK: vreduceps $123, 291(%rax,%r14,8), %xmm20
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x08,0x56,0xa4,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreduceps $0x7b,291(%rax,%r14,8), %xmm20
+
+// CHECK: vreduceps $123, (%rcx){1to4}, %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x56,0x21,0x7b]
+          vreduceps $0x7b,(%rcx){1to4}, %xmm20
+
+// CHECK: vreduceps $123, 2032(%rdx), %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x56,0x62,0x7f,0x7b]
+          vreduceps $0x7b,2032(%rdx), %xmm20
+
+// CHECK: vreduceps $123, 2048(%rdx), %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x56,0xa2,0x00,0x08,0x00,0x00,0x7b]
+          vreduceps $0x7b,2048(%rdx), %xmm20
+
+// CHECK: vreduceps $123, -2048(%rdx), %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x56,0x62,0x80,0x7b]
+          vreduceps $0x7b,-2048(%rdx), %xmm20
+
+// CHECK: vreduceps $123, -2064(%rdx), %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x56,0xa2,0xf0,0xf7,0xff,0xff,0x7b]
+          vreduceps $0x7b,-2064(%rdx), %xmm20
+
+// CHECK: vreduceps $123, 508(%rdx){1to4}, %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x56,0x62,0x7f,0x7b]
+          vreduceps $0x7b,508(%rdx){1to4}, %xmm20
+
+// CHECK: vreduceps $123, 512(%rdx){1to4}, %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x56,0xa2,0x00,0x02,0x00,0x00,0x7b]
+          vreduceps $0x7b,512(%rdx){1to4}, %xmm20
+
+// CHECK: vreduceps $123, -512(%rdx){1to4}, %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x56,0x62,0x80,0x7b]
+          vreduceps $0x7b,-512(%rdx){1to4}, %xmm20
+
+// CHECK: vreduceps $123, -516(%rdx){1to4}, %xmm20
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x56,0xa2,0xfc,0xfd,0xff,0xff,0x7b]
+          vreduceps $0x7b,-516(%rdx){1to4}, %xmm20
+
+// CHECK: vreduceps $171, %ymm22, %ymm26
+// CHECK:  encoding: [0x62,0x23,0x7d,0x28,0x56,0xd6,0xab]
+          vreduceps $0xab, %ymm22, %ymm26
+
+// CHECK: vreduceps $171, %ymm22, %ymm26 {%k6}
+// CHECK:  encoding: [0x62,0x23,0x7d,0x2e,0x56,0xd6,0xab]
+          vreduceps $0xab, %ymm22, %ymm26 {%k6}
+
+// CHECK: vreduceps $171, %ymm22, %ymm26 {%k6} {z}
+// CHECK:  encoding: [0x62,0x23,0x7d,0xae,0x56,0xd6,0xab]
+          vreduceps $0xab, %ymm22, %ymm26 {%k6} {z}
+
+// CHECK: vreduceps $123, %ymm22, %ymm26
+// CHECK:  encoding: [0x62,0x23,0x7d,0x28,0x56,0xd6,0x7b]
+          vreduceps $0x7b, %ymm22, %ymm26
+
+// CHECK: vreduceps $123, (%rcx), %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x11,0x7b]
+          vreduceps $0x7b,(%rcx), %ymm26
+
+// CHECK: vreduceps $123, 291(%rax,%r14,8), %ymm26
+// CHECK:  encoding: [0x62,0x23,0x7d,0x28,0x56,0x94,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vreduceps $0x7b,291(%rax,%r14,8), %ymm26
+
+// CHECK: vreduceps $123, (%rcx){1to8}, %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x11,0x7b]
+          vreduceps $0x7b,(%rcx){1to8}, %ymm26
+
+// CHECK: vreduceps $123, 4064(%rdx), %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x52,0x7f,0x7b]
+          vreduceps $0x7b,4064(%rdx), %ymm26
+
+// CHECK: vreduceps $123, 4096(%rdx), %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x92,0x00,0x10,0x00,0x00,0x7b]
+          vreduceps $0x7b,4096(%rdx), %ymm26
+
+// CHECK: vreduceps $123, -4096(%rdx), %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x52,0x80,0x7b]
+          vreduceps $0x7b,-4096(%rdx), %ymm26
+
+// CHECK: vreduceps $123, -4128(%rdx), %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x28,0x56,0x92,0xe0,0xef,0xff,0xff,0x7b]
+          vreduceps $0x7b,-4128(%rdx), %ymm26
+
+// CHECK: vreduceps $123, 508(%rdx){1to8}, %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x52,0x7f,0x7b]
+          vreduceps $0x7b,508(%rdx){1to8}, %ymm26
+
+// CHECK: vreduceps $123, 512(%rdx){1to8}, %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x92,0x00,0x02,0x00,0x00,0x7b]
+          vreduceps $0x7b,512(%rdx){1to8}, %ymm26
+
+// CHECK: vreduceps $123, -512(%rdx){1to8}, %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x52,0x80,0x7b]
+          vreduceps $0x7b,-512(%rdx){1to8}, %ymm26
+
+// CHECK: vreduceps $123, -516(%rdx){1to8}, %ymm26
+// CHECK:  encoding: [0x62,0x63,0x7d,0x38,0x56,0x92,0xfc,0xfd,0xff,0xff,0x7b]
+          vreduceps $0x7b,-516(%rdx){1to8}, %ymm26
+
 // CHECK: vcvtpd2qq %xmm22, %xmm24
 // CHECK:  encoding: [0x62,0x21,0xfd,0x08,0x7b,0xc6]
           vcvtpd2qq %xmm22, %xmm24
 
 // CHECK: vcvtpd2qq %xmm22, %xmm24 {%k6}
 // CHECK:  encoding: [0x62,0x21,0xfd,0x0e,0x7b,0xc6]
           vcvtpd2qq %xmm22, %xmm24 {%k6}
 

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

 // CHECK: vscalefps -512(%rdx){1to8}, %ymm22, %ymm25
 // CHECK:  encoding: [0x62,0x62,0x4d,0x30,0x2c,0x4a,0x80]
           vscalefps -512(%rdx){1to8}, %ymm22, %ymm25
 
 // CHECK: vscalefps -516(%rdx){1to8}, %ymm22, %ymm25
 // CHECK:  encoding: [0x62,0x62,0x4d,0x30,0x2c,0x8a,0xfc,0xfd,0xff,0xff]
           vscalefps -516(%rdx){1to8}, %ymm22, %ymm25
 
+// CHECK: vrndscalepd $171, %xmm28, %xmm29
+// CHECK:  encoding: [0x62,0x03,0xfd,0x08,0x09,0xec,0xab]
+          vrndscalepd $0xab, %xmm28, %xmm29
+
+// CHECK: vrndscalepd $171, %xmm28, %xmm29 {%k4}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x0c,0x09,0xec,0xab]
+          vrndscalepd $0xab, %xmm28, %xmm29 {%k4}
+
+// CHECK: vrndscalepd $171, %xmm28, %xmm29 {%k4} {z}
+// CHECK:  encoding: [0x62,0x03,0xfd,0x8c,0x09,0xec,0xab]
+          vrndscalepd $0xab, %xmm28, %xmm29 {%k4} {z}
+
+// CHECK: vrndscalepd $123, %xmm28, %xmm29
+// CHECK:  encoding: [0x62,0x03,0xfd,0x08,0x09,0xec,0x7b]
+          vrndscalepd $0x7b, %xmm28, %xmm29
+
+// CHECK: vrndscalepd $123, (%rcx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x09,0x29,0x7b]
+          vrndscalepd $0x7b, (%rcx), %xmm29
+
+// CHECK: vrndscalepd $123, 291(%rax,%r14,8), %xmm29
+// CHECK:  encoding: [0x62,0x23,0xfd,0x08,0x09,0xac,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 291(%rax,%r14,8), %xmm29
+
+// CHECK: vrndscalepd $123, (%rcx){1to2}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x09,0x29,0x7b]
+          vrndscalepd $0x7b, (%rcx){1to2}, %xmm29
+
+// CHECK: vrndscalepd $123, 2032(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x09,0x6a,0x7f,0x7b]
+          vrndscalepd $0x7b, 2032(%rdx), %xmm29
+
+// CHECK: vrndscalepd $123, 2048(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x09,0xaa,0x00,0x08,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 2048(%rdx), %xmm29
+
+// CHECK: vrndscalepd $123, -2048(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x09,0x6a,0x80,0x7b]
+          vrndscalepd $0x7b, -2048(%rdx), %xmm29
+
+// CHECK: vrndscalepd $123, -2064(%rdx), %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x08,0x09,0xaa,0xf0,0xf7,0xff,0xff,0x7b]
+          vrndscalepd $0x7b, -2064(%rdx), %xmm29
+
+// CHECK: vrndscalepd $123, 1016(%rdx){1to2}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x09,0x6a,0x7f,0x7b]
+          vrndscalepd $0x7b, 1016(%rdx){1to2}, %xmm29
+
+// CHECK: vrndscalepd $123, 1024(%rdx){1to2}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x09,0xaa,0x00,0x04,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 1024(%rdx){1to2}, %xmm29
+
+// CHECK: vrndscalepd $123, -1024(%rdx){1to2}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x09,0x6a,0x80,0x7b]
+          vrndscalepd $0x7b, -1024(%rdx){1to2}, %xmm29
+
+// CHECK: vrndscalepd $123, -1032(%rdx){1to2}, %xmm29
+// CHECK:  encoding: [0x62,0x63,0xfd,0x18,0x09,0xaa,0xf8,0xfb,0xff,0xff,0x7b]
+          vrndscalepd $0x7b, -1032(%rdx){1to2}, %xmm29
+
+// CHECK: vrndscalepd $171, %ymm22, %ymm17
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x28,0x09,0xce,0xab]
+          vrndscalepd $0xab, %ymm22, %ymm17
+
+// CHECK: vrndscalepd $171, %ymm22, %ymm17 {%k7}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x2f,0x09,0xce,0xab]
+          vrndscalepd $0xab, %ymm22, %ymm17 {%k7}
+
+// CHECK: vrndscalepd $171, %ymm22, %ymm17 {%k7} {z}
+// CHECK:  encoding: [0x62,0xa3,0xfd,0xaf,0x09,0xce,0xab]
+          vrndscalepd $0xab, %ymm22, %ymm17 {%k7} {z}
+
+// CHECK: vrndscalepd $123, %ymm22, %ymm17
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x28,0x09,0xce,0x7b]
+          vrndscalepd $0x7b, %ymm22, %ymm17
+
+// CHECK: vrndscalepd $123, (%rcx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x09,0x09,0x7b]
+          vrndscalepd $0x7b, (%rcx), %ymm17
+
+// CHECK: vrndscalepd $123, 291(%rax,%r14,8), %ymm17
+// CHECK:  encoding: [0x62,0xa3,0xfd,0x28,0x09,0x8c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 291(%rax,%r14,8), %ymm17
+
+// CHECK: vrndscalepd $123, (%rcx){1to4}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x09,0x09,0x7b]
+          vrndscalepd $0x7b, (%rcx){1to4}, %ymm17
+
+// CHECK: vrndscalepd $123, 4064(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x09,0x4a,0x7f,0x7b]
+          vrndscalepd $0x7b, 4064(%rdx), %ymm17
+
+// CHECK: vrndscalepd $123, 4096(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x09,0x8a,0x00,0x10,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 4096(%rdx), %ymm17
+
+// CHECK: vrndscalepd $123, -4096(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x09,0x4a,0x80,0x7b]
+          vrndscalepd $0x7b, -4096(%rdx), %ymm17
+
+// CHECK: vrndscalepd $123, -4128(%rdx), %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x28,0x09,0x8a,0xe0,0xef,0xff,0xff,0x7b]
+          vrndscalepd $0x7b, -4128(%rdx), %ymm17
+
+// CHECK: vrndscalepd $123, 1016(%rdx){1to4}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x09,0x4a,0x7f,0x7b]
+          vrndscalepd $0x7b, 1016(%rdx){1to4}, %ymm17
+
+// CHECK: vrndscalepd $123, 1024(%rdx){1to4}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x09,0x8a,0x00,0x04,0x00,0x00,0x7b]
+          vrndscalepd $0x7b, 1024(%rdx){1to4}, %ymm17
+
+// CHECK: vrndscalepd $123, -1024(%rdx){1to4}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x09,0x4a,0x80,0x7b]
+          vrndscalepd $0x7b, -1024(%rdx){1to4}, %ymm17
+
+// CHECK: vrndscalepd $123, -1032(%rdx){1to4}, %ymm17
+// CHECK:  encoding: [0x62,0xe3,0xfd,0x38,0x09,0x8a,0xf8,0xfb,0xff,0xff,0x7b]
+          vrndscalepd $0x7b, -1032(%rdx){1to4}, %ymm17
+
+// CHECK: vrndscaleps $171, %xmm26, %xmm22
+// CHECK:  encoding: [0x62,0x83,0x7d,0x08,0x08,0xf2,0xab]
+          vrndscaleps $0xab, %xmm26, %xmm22
+
+// CHECK: vrndscaleps $171, %xmm26, %xmm22 {%k4}
+// CHECK:  encoding: [0x62,0x83,0x7d,0x0c,0x08,0xf2,0xab]
+          vrndscaleps $0xab, %xmm26, %xmm22 {%k4}
+
+// CHECK: vrndscaleps $171, %xmm26, %xmm22 {%k4} {z}
+// CHECK:  encoding: [0x62,0x83,0x7d,0x8c,0x08,0xf2,0xab]
+          vrndscaleps $0xab, %xmm26, %xmm22 {%k4} {z}
+
+// CHECK: vrndscaleps $123, %xmm26, %xmm22
+// CHECK:  encoding: [0x62,0x83,0x7d,0x08,0x08,0xf2,0x7b]
+          vrndscaleps $0x7b, %xmm26, %xmm22
+
+// CHECK: vrndscaleps $123, (%rcx), %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x08,0x31,0x7b]
+          vrndscaleps $0x7b, (%rcx), %xmm22
+
+// CHECK: vrndscaleps $123, 291(%rax,%r14,8), %xmm22
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x08,0x08,0xb4,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 291(%rax,%r14,8), %xmm22
+
+// CHECK: vrndscaleps $123, (%rcx){1to4}, %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x08,0x31,0x7b]
+          vrndscaleps $0x7b, (%rcx){1to4}, %xmm22
+
+// CHECK: vrndscaleps $123, 2032(%rdx), %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x08,0x72,0x7f,0x7b]
+          vrndscaleps $0x7b, 2032(%rdx), %xmm22
+
+// CHECK: vrndscaleps $123, 2048(%rdx), %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x08,0xb2,0x00,0x08,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 2048(%rdx), %xmm22
+
+// CHECK: vrndscaleps $123, -2048(%rdx), %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x08,0x72,0x80,0x7b]
+          vrndscaleps $0x7b, -2048(%rdx), %xmm22
+
+// CHECK: vrndscaleps $123, -2064(%rdx), %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x08,0x08,0xb2,0xf0,0xf7,0xff,0xff,0x7b]
+          vrndscaleps $0x7b, -2064(%rdx), %xmm22
+
+// CHECK: vrndscaleps $123, 508(%rdx){1to4}, %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x08,0x72,0x7f,0x7b]
+          vrndscaleps $0x7b, 508(%rdx){1to4}, %xmm22
+
+// CHECK: vrndscaleps $123, 512(%rdx){1to4}, %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x08,0xb2,0x00,0x02,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 512(%rdx){1to4}, %xmm22
+
+// CHECK: vrndscaleps $123, -512(%rdx){1to4}, %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x08,0x72,0x80,0x7b]
+          vrndscaleps $0x7b, -512(%rdx){1to4}, %xmm22
+
+// CHECK: vrndscaleps $123, -516(%rdx){1to4}, %xmm22
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x18,0x08,0xb2,0xfc,0xfd,0xff,0xff,0x7b]
+          vrndscaleps $0x7b, -516(%rdx){1to4}, %xmm22
+
+// CHECK: vrndscaleps $171, %ymm17, %ymm19
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x28,0x08,0xd9,0xab]
+          vrndscaleps $0xab, %ymm17, %ymm19
+
+// CHECK: vrndscaleps $171, %ymm17, %ymm19 {%k7}
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x2f,0x08,0xd9,0xab]
+          vrndscaleps $0xab, %ymm17, %ymm19 {%k7}
+
+// CHECK: vrndscaleps $171, %ymm17, %ymm19 {%k7} {z}
+// CHECK:  encoding: [0x62,0xa3,0x7d,0xaf,0x08,0xd9,0xab]
+          vrndscaleps $0xab, %ymm17, %ymm19 {%k7} {z}
+
+// CHECK: vrndscaleps $123, %ymm17, %ymm19
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x28,0x08,0xd9,0x7b]
+          vrndscaleps $0x7b, %ymm17, %ymm19
+
+// CHECK: vrndscaleps $123, (%rcx), %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x08,0x19,0x7b]
+          vrndscaleps $0x7b, (%rcx), %ymm19
+
+// CHECK: vrndscaleps $123, 291(%rax,%r14,8), %ymm19
+// CHECK:  encoding: [0x62,0xa3,0x7d,0x28,0x08,0x9c,0xf0,0x23,0x01,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 291(%rax,%r14,8), %ymm19
+
+// CHECK: vrndscaleps $123, (%rcx){1to8}, %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x08,0x19,0x7b]
+          vrndscaleps $0x7b, (%rcx){1to8}, %ymm19
+
+// CHECK: vrndscaleps $123, 4064(%rdx), %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x08,0x5a,0x7f,0x7b]
+          vrndscaleps $0x7b, 4064(%rdx), %ymm19
+
+// CHECK: vrndscaleps $123, 4096(%rdx), %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x08,0x9a,0x00,0x10,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 4096(%rdx), %ymm19
+
+// CHECK: vrndscaleps $123, -4096(%rdx), %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x08,0x5a,0x80,0x7b]
+          vrndscaleps $0x7b, -4096(%rdx), %ymm19
+
+// CHECK: vrndscaleps $123, -4128(%rdx), %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x28,0x08,0x9a,0xe0,0xef,0xff,0xff,0x7b]
+          vrndscaleps $0x7b, -4128(%rdx), %ymm19
+
+// CHECK: vrndscaleps $123, 508(%rdx){1to8}, %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x08,0x5a,0x7f,0x7b]
+          vrndscaleps $0x7b, 508(%rdx){1to8}, %ymm19
+
+// CHECK: vrndscaleps $123, 512(%rdx){1to8}, %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x08,0x9a,0x00,0x02,0x00,0x00,0x7b]
+          vrndscaleps $0x7b, 512(%rdx){1to8}, %ymm19
+
+// CHECK: vrndscaleps $123, -512(%rdx){1to8}, %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x08,0x5a,0x80,0x7b]
+          vrndscaleps $0x7b, -512(%rdx){1to8}, %ymm19
+
+// CHECK: vrndscaleps $123, -516(%rdx){1to8}, %ymm19
+// CHECK:  encoding: [0x62,0xe3,0x7d,0x38,0x08,0x9a,0xfc,0xfd,0xff,0xff,0x7b]
+          vrndscaleps $0x7b, -516(%rdx){1to8}, %ymm19
+
 // CHECK: vcvtps2pd %xmm27, %xmm20
 // CHECK:  encoding: [0x62,0x81,0x7c,0x08,0x5a,0xe3]
           vcvtps2pd %xmm27, %xmm20
 
 // CHECK: vcvtps2pd %xmm27, %xmm20 {%k3}
 // CHECK:  encoding: [0x62,0x81,0x7c,0x0b,0x5a,0xe3]
           vcvtps2pd %xmm27, %xmm20 {%k3}