1//=- AArch64SVEInstrInfo.td - AArch64 SVE Instructions -*- tablegen -*-----=// 2// 3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4// See https://llvm.org/LICENSE.txt for license information. 5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6// 7//===----------------------------------------------------------------------===// 8// 9// AArch64 Scalable Vector Extension (SVE) Instruction definitions. 10// 11//===----------------------------------------------------------------------===// 12 13def SDT_AArch64_GLD1 : SDTypeProfile<1, 4, [ 14 SDTCisVec<0>, SDTCisVec<1>, SDTCisPtrTy<2>, SDTCisVec<3>, SDTCisVT<4, OtherVT>, 15 SDTCVecEltisVT<1,i1>, SDTCisSameNumEltsAs<0,1> 16]>; 17 18def SDT_AArch64_GLD1_IMM : SDTypeProfile<1, 4, [ 19 SDTCisVec<0>, SDTCisVec<1>, SDTCisVec<2>, SDTCisInt<3>, SDTCisVT<4, OtherVT>, 20 SDTCVecEltisVT<1,i1>, SDTCisSameNumEltsAs<0,1> 21]>; 22 23def SDT_AArch64_SST1 : SDTypeProfile<0, 5, [ 24 SDTCisVec<0>, SDTCisVec<1>, SDTCisPtrTy<2>, SDTCisVec<3>, SDTCisVT<4, OtherVT>, 25 SDTCVecEltisVT<1,i1>, SDTCisSameNumEltsAs<0,1> 26]>; 27 28def SDT_AArch64_SST1_IMM : SDTypeProfile<0, 5, [ 29 SDTCisVec<0>, SDTCisVec<1>, SDTCisVec<2>, SDTCisInt<3>, SDTCisVT<4, OtherVT>, 30 SDTCVecEltisVT<1,i1>, SDTCisSameNumEltsAs<0,1> 31]>; 32 33def AArch64st1_scatter : SDNode<"AArch64ISD::SST1", SDT_AArch64_SST1, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 34def AArch64st1_scatter_scaled : SDNode<"AArch64ISD::SST1_SCALED", SDT_AArch64_SST1, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 35def AArch64st1_scatter_uxtw : SDNode<"AArch64ISD::SST1_UXTW", SDT_AArch64_SST1, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 36def AArch64st1_scatter_sxtw : SDNode<"AArch64ISD::SST1_SXTW", SDT_AArch64_SST1, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 37def AArch64st1_scatter_uxtw_scaled : SDNode<"AArch64ISD::SST1_UXTW_SCALED", SDT_AArch64_SST1, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 38def AArch64st1_scatter_sxtw_scaled : SDNode<"AArch64ISD::SST1_SXTW_SCALED", SDT_AArch64_SST1, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 39def AArch64st1_scatter_imm : SDNode<"AArch64ISD::SST1_IMM", SDT_AArch64_SST1_IMM, [SDNPHasChain, SDNPMayStore, SDNPOptInGlue]>; 40 41def AArch64ld1_gather : SDNode<"AArch64ISD::GLD1", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 42def AArch64ld1_gather_scaled : SDNode<"AArch64ISD::GLD1_SCALED", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 43def AArch64ld1_gather_uxtw : SDNode<"AArch64ISD::GLD1_UXTW", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 44def AArch64ld1_gather_sxtw : SDNode<"AArch64ISD::GLD1_SXTW", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 45def AArch64ld1_gather_uxtw_scaled : SDNode<"AArch64ISD::GLD1_UXTW_SCALED", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 46def AArch64ld1_gather_sxtw_scaled : SDNode<"AArch64ISD::GLD1_SXTW_SCALED", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 47def AArch64ld1_gather_imm : SDNode<"AArch64ISD::GLD1_IMM", SDT_AArch64_GLD1_IMM, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 48 49def AArch64ld1s_gather : SDNode<"AArch64ISD::GLD1S", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 50def AArch64ld1s_gather_scaled : SDNode<"AArch64ISD::GLD1S_SCALED", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 51def AArch64ld1s_gather_uxtw : SDNode<"AArch64ISD::GLD1S_UXTW", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 52def AArch64ld1s_gather_sxtw : SDNode<"AArch64ISD::GLD1S_SXTW", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 53def AArch64ld1s_gather_uxtw_scaled : SDNode<"AArch64ISD::GLD1S_UXTW_SCALED", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 54def AArch64ld1s_gather_sxtw_scaled : SDNode<"AArch64ISD::GLD1S_SXTW_SCALED", SDT_AArch64_GLD1, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 55def AArch64ld1s_gather_imm : SDNode<"AArch64ISD::GLD1S_IMM", SDT_AArch64_GLD1_IMM, [SDNPHasChain, SDNPMayLoad, SDNPOptInGlue]>; 56 57def SDT_AArch64Reduce : SDTypeProfile<1, 2, [SDTCisVec<1>, SDTCisVec<2>]>; 58 59def AArch64smaxv_pred : SDNode<"AArch64ISD::SMAXV_PRED", SDT_AArch64Reduce>; 60def AArch64umaxv_pred : SDNode<"AArch64ISD::UMAXV_PRED", SDT_AArch64Reduce>; 61def AArch64sminv_pred : SDNode<"AArch64ISD::SMINV_PRED", SDT_AArch64Reduce>; 62def AArch64uminv_pred : SDNode<"AArch64ISD::UMINV_PRED", SDT_AArch64Reduce>; 63def AArch64orv_pred : SDNode<"AArch64ISD::ORV_PRED", SDT_AArch64Reduce>; 64def AArch64eorv_pred : SDNode<"AArch64ISD::EORV_PRED", SDT_AArch64Reduce>; 65def AArch64andv_pred : SDNode<"AArch64ISD::ANDV_PRED", SDT_AArch64Reduce>; 66def AArch64lasta : SDNode<"AArch64ISD::LASTA", SDT_AArch64Reduce>; 67def AArch64lastb : SDNode<"AArch64ISD::LASTB", SDT_AArch64Reduce>; 68 69def SDT_AArch64ReduceWithInit : SDTypeProfile<1, 3, [SDTCisVec<1>, SDTCisVec<3>]>; 70def AArch64clasta_n : SDNode<"AArch64ISD::CLASTA_N", SDT_AArch64ReduceWithInit>; 71def AArch64clastb_n : SDNode<"AArch64ISD::CLASTB_N", SDT_AArch64ReduceWithInit>; 72 73def SDT_AArch64Rev : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisSameAs<0,1>]>; 74def AArch64rev : SDNode<"AArch64ISD::REV", SDT_AArch64Rev>; 75 76def SDT_AArch64PTest : SDTypeProfile<0, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>]>; 77def AArch64ptest : SDNode<"AArch64ISD::PTEST", SDT_AArch64PTest>; 78 79let Predicates = [HasSVE] in { 80 81 def RDFFR_PPz : sve_int_rdffr_pred<0b0, "rdffr">; 82 def RDFFRS_PPz : sve_int_rdffr_pred<0b1, "rdffrs">; 83 def RDFFR_P : sve_int_rdffr_unpred<"rdffr">; 84 def SETFFR : sve_int_setffr<"setffr">; 85 def WRFFR : sve_int_wrffr<"wrffr">; 86 87 defm ADD_ZZZ : sve_int_bin_cons_arit_0<0b000, "add", add>; 88 defm SUB_ZZZ : sve_int_bin_cons_arit_0<0b001, "sub", sub>; 89 defm SQADD_ZZZ : sve_int_bin_cons_arit_0<0b100, "sqadd", saddsat>; 90 defm UQADD_ZZZ : sve_int_bin_cons_arit_0<0b101, "uqadd", uaddsat>; 91 defm SQSUB_ZZZ : sve_int_bin_cons_arit_0<0b110, "sqsub", ssubsat>; 92 defm UQSUB_ZZZ : sve_int_bin_cons_arit_0<0b111, "uqsub", usubsat>; 93 94 defm AND_ZZZ : sve_int_bin_cons_log<0b00, "and", and>; 95 defm ORR_ZZZ : sve_int_bin_cons_log<0b01, "orr", or>; 96 defm EOR_ZZZ : sve_int_bin_cons_log<0b10, "eor", xor>; 97 defm BIC_ZZZ : sve_int_bin_cons_log<0b11, "bic", null_frag>; 98 99 defm ADD_ZPmZ : sve_int_bin_pred_arit_0<0b000, "add", int_aarch64_sve_add>; 100 defm SUB_ZPmZ : sve_int_bin_pred_arit_0<0b001, "sub", int_aarch64_sve_sub>; 101 defm SUBR_ZPmZ : sve_int_bin_pred_arit_0<0b011, "subr", int_aarch64_sve_subr>; 102 103 defm ORR_ZPmZ : sve_int_bin_pred_log<0b000, "orr", int_aarch64_sve_orr>; 104 defm EOR_ZPmZ : sve_int_bin_pred_log<0b001, "eor", int_aarch64_sve_eor>; 105 defm AND_ZPmZ : sve_int_bin_pred_log<0b010, "and", int_aarch64_sve_and>; 106 defm BIC_ZPmZ : sve_int_bin_pred_log<0b011, "bic", int_aarch64_sve_bic>; 107 108 defm ADD_ZI : sve_int_arith_imm0<0b000, "add", add>; 109 defm SUB_ZI : sve_int_arith_imm0<0b001, "sub", sub>; 110 defm SUBR_ZI : sve_int_arith_imm0_subr<0b011, "subr", sub>; 111 defm SQADD_ZI : sve_int_arith_imm0<0b100, "sqadd", saddsat>; 112 defm UQADD_ZI : sve_int_arith_imm0<0b101, "uqadd", uaddsat>; 113 defm SQSUB_ZI : sve_int_arith_imm0<0b110, "sqsub", ssubsat>; 114 defm UQSUB_ZI : sve_int_arith_imm0<0b111, "uqsub", usubsat>; 115 116 defm MAD_ZPmZZ : sve_int_mladdsub_vvv_pred<0b0, "mad", int_aarch64_sve_mad>; 117 defm MSB_ZPmZZ : sve_int_mladdsub_vvv_pred<0b1, "msb", int_aarch64_sve_msb>; 118 defm MLA_ZPmZZ : sve_int_mlas_vvv_pred<0b0, "mla", int_aarch64_sve_mla>; 119 defm MLS_ZPmZZ : sve_int_mlas_vvv_pred<0b1, "mls", int_aarch64_sve_mls>; 120 121 // SVE predicated integer reductions. 122 defm SADDV_VPZ : sve_int_reduce_0_saddv<0b000, "saddv", int_aarch64_sve_saddv>; 123 defm UADDV_VPZ : sve_int_reduce_0_uaddv<0b001, "uaddv", int_aarch64_sve_uaddv, int_aarch64_sve_saddv>; 124 defm SMAXV_VPZ : sve_int_reduce_1<0b000, "smaxv", AArch64smaxv_pred>; 125 defm UMAXV_VPZ : sve_int_reduce_1<0b001, "umaxv", AArch64umaxv_pred>; 126 defm SMINV_VPZ : sve_int_reduce_1<0b010, "sminv", AArch64sminv_pred>; 127 defm UMINV_VPZ : sve_int_reduce_1<0b011, "uminv", AArch64uminv_pred>; 128 defm ORV_VPZ : sve_int_reduce_2<0b000, "orv", AArch64orv_pred>; 129 defm EORV_VPZ : sve_int_reduce_2<0b001, "eorv", AArch64eorv_pred>; 130 defm ANDV_VPZ : sve_int_reduce_2<0b010, "andv", AArch64andv_pred>; 131 132 defm ORR_ZI : sve_int_log_imm<0b00, "orr", "orn", or>; 133 defm EOR_ZI : sve_int_log_imm<0b01, "eor", "eon", xor>; 134 defm AND_ZI : sve_int_log_imm<0b10, "and", "bic", and>; 135 136 defm SMAX_ZI : sve_int_arith_imm1<0b00, "smax", smax>; 137 defm SMIN_ZI : sve_int_arith_imm1<0b10, "smin", smin>; 138 defm UMAX_ZI : sve_int_arith_imm1_unsigned<0b01, "umax", umax>; 139 defm UMIN_ZI : sve_int_arith_imm1_unsigned<0b11, "umin", umin>; 140 141 defm MUL_ZI : sve_int_arith_imm2<"mul", mul>; 142 defm MUL_ZPmZ : sve_int_bin_pred_arit_2<0b000, "mul", int_aarch64_sve_mul>; 143 defm SMULH_ZPmZ : sve_int_bin_pred_arit_2<0b010, "smulh", int_aarch64_sve_smulh>; 144 defm UMULH_ZPmZ : sve_int_bin_pred_arit_2<0b011, "umulh", int_aarch64_sve_umulh>; 145 146 defm SDIV_ZPmZ : sve_int_bin_pred_arit_2_div<0b100, "sdiv", int_aarch64_sve_sdiv>; 147 defm UDIV_ZPmZ : sve_int_bin_pred_arit_2_div<0b101, "udiv", int_aarch64_sve_udiv>; 148 defm SDIVR_ZPmZ : sve_int_bin_pred_arit_2_div<0b110, "sdivr", int_aarch64_sve_sdivr>; 149 defm UDIVR_ZPmZ : sve_int_bin_pred_arit_2_div<0b111, "udivr", int_aarch64_sve_udivr>; 150 151 defm SDOT_ZZZ : sve_intx_dot<0b0, "sdot", int_aarch64_sve_sdot>; 152 defm UDOT_ZZZ : sve_intx_dot<0b1, "udot", int_aarch64_sve_udot>; 153 154 defm SDOT_ZZZI : sve_intx_dot_by_indexed_elem<0b0, "sdot", int_aarch64_sve_sdot_lane>; 155 defm UDOT_ZZZI : sve_intx_dot_by_indexed_elem<0b1, "udot", int_aarch64_sve_udot_lane>; 156 157 defm SXTB_ZPmZ : sve_int_un_pred_arit_0_h<0b000, "sxtb", int_aarch64_sve_sxtb>; 158 defm UXTB_ZPmZ : sve_int_un_pred_arit_0_h<0b001, "uxtb", int_aarch64_sve_uxtb>; 159 defm SXTH_ZPmZ : sve_int_un_pred_arit_0_w<0b010, "sxth", int_aarch64_sve_sxth>; 160 defm UXTH_ZPmZ : sve_int_un_pred_arit_0_w<0b011, "uxth", int_aarch64_sve_uxth>; 161 defm SXTW_ZPmZ : sve_int_un_pred_arit_0_d<0b100, "sxtw", int_aarch64_sve_sxtw>; 162 defm UXTW_ZPmZ : sve_int_un_pred_arit_0_d<0b101, "uxtw", int_aarch64_sve_uxtw>; 163 defm ABS_ZPmZ : sve_int_un_pred_arit_0< 0b110, "abs", int_aarch64_sve_abs>; 164 defm NEG_ZPmZ : sve_int_un_pred_arit_0< 0b111, "neg", int_aarch64_sve_neg>; 165 166 defm CLS_ZPmZ : sve_int_un_pred_arit_1< 0b000, "cls", int_aarch64_sve_cls>; 167 defm CLZ_ZPmZ : sve_int_un_pred_arit_1< 0b001, "clz", int_aarch64_sve_clz>; 168 defm CNT_ZPmZ : sve_int_un_pred_arit_1< 0b010, "cnt", int_aarch64_sve_cnt>; 169 defm CNOT_ZPmZ : sve_int_un_pred_arit_1< 0b011, "cnot", int_aarch64_sve_cnot>; 170 defm NOT_ZPmZ : sve_int_un_pred_arit_1< 0b110, "not", int_aarch64_sve_not>; 171 defm FABS_ZPmZ : sve_int_un_pred_arit_1_fp<0b100, "fabs", int_aarch64_sve_fabs>; 172 defm FNEG_ZPmZ : sve_int_un_pred_arit_1_fp<0b101, "fneg", int_aarch64_sve_fneg>; 173 174 defm SMAX_ZPmZ : sve_int_bin_pred_arit_1<0b000, "smax", int_aarch64_sve_smax>; 175 defm UMAX_ZPmZ : sve_int_bin_pred_arit_1<0b001, "umax", int_aarch64_sve_umax>; 176 defm SMIN_ZPmZ : sve_int_bin_pred_arit_1<0b010, "smin", int_aarch64_sve_smin>; 177 defm UMIN_ZPmZ : sve_int_bin_pred_arit_1<0b011, "umin", int_aarch64_sve_umin>; 178 defm SABD_ZPmZ : sve_int_bin_pred_arit_1<0b100, "sabd", int_aarch64_sve_sabd>; 179 defm UABD_ZPmZ : sve_int_bin_pred_arit_1<0b101, "uabd", int_aarch64_sve_uabd>; 180 181 defm FRECPE_ZZ : sve_fp_2op_u_zd<0b110, "frecpe", int_aarch64_sve_frecpe_x>; 182 defm FRSQRTE_ZZ : sve_fp_2op_u_zd<0b111, "frsqrte", int_aarch64_sve_frsqrte_x>; 183 184 defm FADD_ZPmI : sve_fp_2op_i_p_zds<0b000, "fadd", sve_fpimm_half_one>; 185 defm FSUB_ZPmI : sve_fp_2op_i_p_zds<0b001, "fsub", sve_fpimm_half_one>; 186 defm FMUL_ZPmI : sve_fp_2op_i_p_zds<0b010, "fmul", sve_fpimm_half_two>; 187 defm FSUBR_ZPmI : sve_fp_2op_i_p_zds<0b011, "fsubr", sve_fpimm_half_one>; 188 defm FMAXNM_ZPmI : sve_fp_2op_i_p_zds<0b100, "fmaxnm", sve_fpimm_zero_one>; 189 defm FMINNM_ZPmI : sve_fp_2op_i_p_zds<0b101, "fminnm", sve_fpimm_zero_one>; 190 defm FMAX_ZPmI : sve_fp_2op_i_p_zds<0b110, "fmax", sve_fpimm_zero_one>; 191 defm FMIN_ZPmI : sve_fp_2op_i_p_zds<0b111, "fmin", sve_fpimm_zero_one>; 192 193 defm FADD_ZPmZ : sve_fp_2op_p_zds<0b0000, "fadd", int_aarch64_sve_fadd>; 194 defm FSUB_ZPmZ : sve_fp_2op_p_zds<0b0001, "fsub", int_aarch64_sve_fsub>; 195 defm FMUL_ZPmZ : sve_fp_2op_p_zds<0b0010, "fmul", int_aarch64_sve_fmul>; 196 defm FSUBR_ZPmZ : sve_fp_2op_p_zds<0b0011, "fsubr", int_aarch64_sve_fsubr>; 197 defm FMAXNM_ZPmZ : sve_fp_2op_p_zds<0b0100, "fmaxnm", int_aarch64_sve_fmaxnm>; 198 defm FMINNM_ZPmZ : sve_fp_2op_p_zds<0b0101, "fminnm", int_aarch64_sve_fminnm>; 199 defm FMAX_ZPmZ : sve_fp_2op_p_zds<0b0110, "fmax", int_aarch64_sve_fmax>; 200 defm FMIN_ZPmZ : sve_fp_2op_p_zds<0b0111, "fmin", int_aarch64_sve_fmin>; 201 defm FABD_ZPmZ : sve_fp_2op_p_zds<0b1000, "fabd", int_aarch64_sve_fabd>; 202 defm FSCALE_ZPmZ : sve_fp_2op_p_zds_fscale<0b1001, "fscale", int_aarch64_sve_fscale>; 203 defm FMULX_ZPmZ : sve_fp_2op_p_zds<0b1010, "fmulx", int_aarch64_sve_fmulx>; 204 defm FDIVR_ZPmZ : sve_fp_2op_p_zds<0b1100, "fdivr", int_aarch64_sve_fdivr>; 205 defm FDIV_ZPmZ : sve_fp_2op_p_zds<0b1101, "fdiv", int_aarch64_sve_fdiv>; 206 207 defm FADD_ZZZ : sve_fp_3op_u_zd<0b000, "fadd", fadd>; 208 defm FSUB_ZZZ : sve_fp_3op_u_zd<0b001, "fsub", fsub>; 209 defm FMUL_ZZZ : sve_fp_3op_u_zd<0b010, "fmul", fmul>; 210 defm FTSMUL_ZZZ : sve_fp_3op_u_zd_ftsmul<0b011, "ftsmul", int_aarch64_sve_ftsmul_x>; 211 defm FRECPS_ZZZ : sve_fp_3op_u_zd<0b110, "frecps", int_aarch64_sve_frecps_x>; 212 defm FRSQRTS_ZZZ : sve_fp_3op_u_zd<0b111, "frsqrts", int_aarch64_sve_frsqrts_x>; 213 214 defm FTSSEL_ZZZ : sve_int_bin_cons_misc_0_b<"ftssel", int_aarch64_sve_ftssel_x>; 215 216 defm FCADD_ZPmZ : sve_fp_fcadd<"fcadd", int_aarch64_sve_fcadd>; 217 defm FCMLA_ZPmZZ : sve_fp_fcmla<"fcmla", int_aarch64_sve_fcmla>; 218 219 defm FMLA_ZPmZZ : sve_fp_3op_p_zds_a<0b00, "fmla", int_aarch64_sve_fmla>; 220 defm FMLS_ZPmZZ : sve_fp_3op_p_zds_a<0b01, "fmls", int_aarch64_sve_fmls>; 221 defm FNMLA_ZPmZZ : sve_fp_3op_p_zds_a<0b10, "fnmla", int_aarch64_sve_fnmla>; 222 defm FNMLS_ZPmZZ : sve_fp_3op_p_zds_a<0b11, "fnmls", int_aarch64_sve_fnmls>; 223 224 defm FMAD_ZPmZZ : sve_fp_3op_p_zds_b<0b00, "fmad", int_aarch64_sve_fmad>; 225 defm FMSB_ZPmZZ : sve_fp_3op_p_zds_b<0b01, "fmsb", int_aarch64_sve_fmsb>; 226 defm FNMAD_ZPmZZ : sve_fp_3op_p_zds_b<0b10, "fnmad", int_aarch64_sve_fnmad>; 227 defm FNMSB_ZPmZZ : sve_fp_3op_p_zds_b<0b11, "fnmsb", int_aarch64_sve_fnmsb>; 228 229 defm FTMAD_ZZI : sve_fp_ftmad<"ftmad", int_aarch64_sve_ftmad_x>; 230 231 defm FMLA_ZZZI : sve_fp_fma_by_indexed_elem<0b0, "fmla", int_aarch64_sve_fmla_lane>; 232 defm FMLS_ZZZI : sve_fp_fma_by_indexed_elem<0b1, "fmls", int_aarch64_sve_fmls_lane>; 233 234 defm FCMLA_ZZZI : sve_fp_fcmla_by_indexed_elem<"fcmla", int_aarch64_sve_fcmla_lane>; 235 defm FMUL_ZZZI : sve_fp_fmul_by_indexed_elem<"fmul", int_aarch64_sve_fmul_lane>; 236 237 // SVE floating point reductions. 238 defm FADDA_VPZ : sve_fp_2op_p_vd<0b000, "fadda", int_aarch64_sve_fadda>; 239 defm FADDV_VPZ : sve_fp_fast_red<0b000, "faddv", int_aarch64_sve_faddv>; 240 defm FMAXNMV_VPZ : sve_fp_fast_red<0b100, "fmaxnmv", int_aarch64_sve_fmaxnmv>; 241 defm FMINNMV_VPZ : sve_fp_fast_red<0b101, "fminnmv", int_aarch64_sve_fminnmv>; 242 defm FMAXV_VPZ : sve_fp_fast_red<0b110, "fmaxv", int_aarch64_sve_fmaxv>; 243 defm FMINV_VPZ : sve_fp_fast_red<0b111, "fminv", int_aarch64_sve_fminv>; 244 245 // Splat immediate (unpredicated) 246 defm DUP_ZI : sve_int_dup_imm<"dup">; 247 defm FDUP_ZI : sve_int_dup_fpimm<"fdup">; 248 defm DUPM_ZI : sve_int_dup_mask_imm<"dupm">; 249 250 // Splat immediate (predicated) 251 defm CPY_ZPmI : sve_int_dup_imm_pred_merge<"cpy">; 252 defm CPY_ZPzI : sve_int_dup_imm_pred_zero<"cpy">; 253 defm FCPY_ZPmI : sve_int_dup_fpimm_pred<"fcpy">; 254 255 // Splat scalar register (unpredicated, GPR or vector + element index) 256 defm DUP_ZR : sve_int_perm_dup_r<"dup", AArch64dup>; 257 defm DUP_ZZI : sve_int_perm_dup_i<"dup">; 258 259 // Splat scalar register (predicated) 260 defm CPY_ZPmR : sve_int_perm_cpy_r<"cpy">; 261 defm CPY_ZPmV : sve_int_perm_cpy_v<"cpy">; 262 263 // Select elements from either vector (predicated) 264 defm SEL_ZPZZ : sve_int_sel_vvv<"sel", vselect>; 265 266 defm SPLICE_ZPZ : sve_int_perm_splice<"splice", int_aarch64_sve_splice>; 267 defm COMPACT_ZPZ : sve_int_perm_compact<"compact", int_aarch64_sve_compact>; 268 defm INSR_ZR : sve_int_perm_insrs<"insr", AArch64insr>; 269 defm INSR_ZV : sve_int_perm_insrv<"insr", AArch64insr>; 270 defm EXT_ZZI : sve_int_perm_extract_i<"ext", AArch64ext>; 271 272 defm RBIT_ZPmZ : sve_int_perm_rev_rbit<"rbit", int_aarch64_sve_rbit>; 273 defm REVB_ZPmZ : sve_int_perm_rev_revb<"revb", int_aarch64_sve_revb, bswap>; 274 defm REVH_ZPmZ : sve_int_perm_rev_revh<"revh", int_aarch64_sve_revh>; 275 defm REVW_ZPmZ : sve_int_perm_rev_revw<"revw", int_aarch64_sve_revw>; 276 277 defm REV_PP : sve_int_perm_reverse_p<"rev", AArch64rev>; 278 defm REV_ZZ : sve_int_perm_reverse_z<"rev", AArch64rev>; 279 280 defm SUNPKLO_ZZ : sve_int_perm_unpk<0b00, "sunpklo", AArch64sunpklo>; 281 defm SUNPKHI_ZZ : sve_int_perm_unpk<0b01, "sunpkhi", AArch64sunpkhi>; 282 defm UUNPKLO_ZZ : sve_int_perm_unpk<0b10, "uunpklo", AArch64uunpklo>; 283 defm UUNPKHI_ZZ : sve_int_perm_unpk<0b11, "uunpkhi", AArch64uunpkhi>; 284 285 defm PUNPKLO_PP : sve_int_perm_punpk<0b0, "punpklo", int_aarch64_sve_punpklo>; 286 defm PUNPKHI_PP : sve_int_perm_punpk<0b1, "punpkhi", int_aarch64_sve_punpkhi>; 287 288 defm MOVPRFX_ZPzZ : sve_int_movprfx_pred_zero<0b000, "movprfx">; 289 defm MOVPRFX_ZPmZ : sve_int_movprfx_pred_merge<0b001, "movprfx">; 290 def MOVPRFX_ZZ : sve_int_bin_cons_misc_0_c<0b00000001, "movprfx", ZPRAny>; 291 defm FEXPA_ZZ : sve_int_bin_cons_misc_0_c_fexpa<"fexpa", int_aarch64_sve_fexpa_x>; 292 293 def BRKPA_PPzPP : sve_int_brkp<0b00, "brkpa">; 294 def BRKPAS_PPzPP : sve_int_brkp<0b10, "brkpas">; 295 def BRKPB_PPzPP : sve_int_brkp<0b01, "brkpb">; 296 def BRKPBS_PPzPP : sve_int_brkp<0b11, "brkpbs">; 297 298 def BRKN_PPzP : sve_int_brkn<0b0, "brkn">; 299 def BRKNS_PPzP : sve_int_brkn<0b1, "brkns">; 300 301 defm BRKA_PPzP : sve_int_break_z<0b000, "brka">; 302 defm BRKA_PPmP : sve_int_break_m<0b001, "brka">; 303 defm BRKAS_PPzP : sve_int_break_z<0b010, "brkas">; 304 defm BRKB_PPzP : sve_int_break_z<0b100, "brkb">; 305 defm BRKB_PPmP : sve_int_break_m<0b101, "brkb">; 306 defm BRKBS_PPzP : sve_int_break_z<0b110, "brkbs">; 307 308 def PTEST_PP : sve_int_ptest<0b010000, "ptest">; 309 def PFALSE : sve_int_pfalse<0b000000, "pfalse">; 310 defm PFIRST : sve_int_pfirst<0b00000, "pfirst", int_aarch64_sve_pfirst>; 311 defm PNEXT : sve_int_pnext<0b00110, "pnext", int_aarch64_sve_pnext>; 312 313 defm AND_PPzPP : sve_int_pred_log<0b0000, "and", int_aarch64_sve_and_z>; 314 defm BIC_PPzPP : sve_int_pred_log<0b0001, "bic", int_aarch64_sve_bic_z>; 315 defm EOR_PPzPP : sve_int_pred_log<0b0010, "eor", int_aarch64_sve_eor_z>; 316 defm SEL_PPPP : sve_int_pred_log<0b0011, "sel", vselect>; 317 defm ANDS_PPzPP : sve_int_pred_log<0b0100, "ands", null_frag>; 318 defm BICS_PPzPP : sve_int_pred_log<0b0101, "bics", null_frag>; 319 defm EORS_PPzPP : sve_int_pred_log<0b0110, "eors", null_frag>; 320 defm ORR_PPzPP : sve_int_pred_log<0b1000, "orr", int_aarch64_sve_orr_z>; 321 defm ORN_PPzPP : sve_int_pred_log<0b1001, "orn", int_aarch64_sve_orn_z>; 322 defm NOR_PPzPP : sve_int_pred_log<0b1010, "nor", int_aarch64_sve_nor_z>; 323 defm NAND_PPzPP : sve_int_pred_log<0b1011, "nand", int_aarch64_sve_nand_z>; 324 defm ORRS_PPzPP : sve_int_pred_log<0b1100, "orrs", null_frag>; 325 defm ORNS_PPzPP : sve_int_pred_log<0b1101, "orns", null_frag>; 326 defm NORS_PPzPP : sve_int_pred_log<0b1110, "nors", null_frag>; 327 defm NANDS_PPzPP : sve_int_pred_log<0b1111, "nands", null_frag>; 328 329 defm CLASTA_RPZ : sve_int_perm_clast_rz<0, "clasta", AArch64clasta_n>; 330 defm CLASTB_RPZ : sve_int_perm_clast_rz<1, "clastb", AArch64clastb_n>; 331 defm CLASTA_VPZ : sve_int_perm_clast_vz<0, "clasta", AArch64clasta_n>; 332 defm CLASTB_VPZ : sve_int_perm_clast_vz<1, "clastb", AArch64clastb_n>; 333 defm CLASTA_ZPZ : sve_int_perm_clast_zz<0, "clasta", int_aarch64_sve_clasta>; 334 defm CLASTB_ZPZ : sve_int_perm_clast_zz<1, "clastb", int_aarch64_sve_clastb>; 335 336 defm LASTA_RPZ : sve_int_perm_last_r<0, "lasta", AArch64lasta>; 337 defm LASTB_RPZ : sve_int_perm_last_r<1, "lastb", AArch64lastb>; 338 defm LASTA_VPZ : sve_int_perm_last_v<0, "lasta", AArch64lasta>; 339 defm LASTB_VPZ : sve_int_perm_last_v<1, "lastb", AArch64lastb>; 340 341 // continuous load with reg+immediate 342 defm LD1B_IMM : sve_mem_cld_si<0b0000, "ld1b", Z_b, ZPR8>; 343 defm LD1B_H_IMM : sve_mem_cld_si<0b0001, "ld1b", Z_h, ZPR16>; 344 defm LD1B_S_IMM : sve_mem_cld_si<0b0010, "ld1b", Z_s, ZPR32>; 345 defm LD1B_D_IMM : sve_mem_cld_si<0b0011, "ld1b", Z_d, ZPR64>; 346 defm LD1SW_D_IMM : sve_mem_cld_si<0b0100, "ld1sw", Z_d, ZPR64>; 347 defm LD1H_IMM : sve_mem_cld_si<0b0101, "ld1h", Z_h, ZPR16>; 348 defm LD1H_S_IMM : sve_mem_cld_si<0b0110, "ld1h", Z_s, ZPR32>; 349 defm LD1H_D_IMM : sve_mem_cld_si<0b0111, "ld1h", Z_d, ZPR64>; 350 defm LD1SH_D_IMM : sve_mem_cld_si<0b1000, "ld1sh", Z_d, ZPR64>; 351 defm LD1SH_S_IMM : sve_mem_cld_si<0b1001, "ld1sh", Z_s, ZPR32>; 352 defm LD1W_IMM : sve_mem_cld_si<0b1010, "ld1w", Z_s, ZPR32>; 353 defm LD1W_D_IMM : sve_mem_cld_si<0b1011, "ld1w", Z_d, ZPR64>; 354 defm LD1SB_D_IMM : sve_mem_cld_si<0b1100, "ld1sb", Z_d, ZPR64>; 355 defm LD1SB_S_IMM : sve_mem_cld_si<0b1101, "ld1sb", Z_s, ZPR32>; 356 defm LD1SB_H_IMM : sve_mem_cld_si<0b1110, "ld1sb", Z_h, ZPR16>; 357 defm LD1D_IMM : sve_mem_cld_si<0b1111, "ld1d", Z_d, ZPR64>; 358 359 // LD1R loads (splat scalar to vector) 360 defm LD1RB_IMM : sve_mem_ld_dup<0b00, 0b00, "ld1rb", Z_b, ZPR8, uimm6s1>; 361 defm LD1RB_H_IMM : sve_mem_ld_dup<0b00, 0b01, "ld1rb", Z_h, ZPR16, uimm6s1>; 362 defm LD1RB_S_IMM : sve_mem_ld_dup<0b00, 0b10, "ld1rb", Z_s, ZPR32, uimm6s1>; 363 defm LD1RB_D_IMM : sve_mem_ld_dup<0b00, 0b11, "ld1rb", Z_d, ZPR64, uimm6s1>; 364 defm LD1RSW_IMM : sve_mem_ld_dup<0b01, 0b00, "ld1rsw", Z_d, ZPR64, uimm6s4>; 365 defm LD1RH_IMM : sve_mem_ld_dup<0b01, 0b01, "ld1rh", Z_h, ZPR16, uimm6s2>; 366 defm LD1RH_S_IMM : sve_mem_ld_dup<0b01, 0b10, "ld1rh", Z_s, ZPR32, uimm6s2>; 367 defm LD1RH_D_IMM : sve_mem_ld_dup<0b01, 0b11, "ld1rh", Z_d, ZPR64, uimm6s2>; 368 defm LD1RSH_D_IMM : sve_mem_ld_dup<0b10, 0b00, "ld1rsh", Z_d, ZPR64, uimm6s2>; 369 defm LD1RSH_S_IMM : sve_mem_ld_dup<0b10, 0b01, "ld1rsh", Z_s, ZPR32, uimm6s2>; 370 defm LD1RW_IMM : sve_mem_ld_dup<0b10, 0b10, "ld1rw", Z_s, ZPR32, uimm6s4>; 371 defm LD1RW_D_IMM : sve_mem_ld_dup<0b10, 0b11, "ld1rw", Z_d, ZPR64, uimm6s4>; 372 defm LD1RSB_D_IMM : sve_mem_ld_dup<0b11, 0b00, "ld1rsb", Z_d, ZPR64, uimm6s1>; 373 defm LD1RSB_S_IMM : sve_mem_ld_dup<0b11, 0b01, "ld1rsb", Z_s, ZPR32, uimm6s1>; 374 defm LD1RSB_H_IMM : sve_mem_ld_dup<0b11, 0b10, "ld1rsb", Z_h, ZPR16, uimm6s1>; 375 defm LD1RD_IMM : sve_mem_ld_dup<0b11, 0b11, "ld1rd", Z_d, ZPR64, uimm6s8>; 376 377 // LD1RQ loads (load quadword-vector and splat to scalable vector) 378 defm LD1RQ_B_IMM : sve_mem_ldqr_si<0b00, "ld1rqb", Z_b, ZPR8>; 379 defm LD1RQ_H_IMM : sve_mem_ldqr_si<0b01, "ld1rqh", Z_h, ZPR16>; 380 defm LD1RQ_W_IMM : sve_mem_ldqr_si<0b10, "ld1rqw", Z_s, ZPR32>; 381 defm LD1RQ_D_IMM : sve_mem_ldqr_si<0b11, "ld1rqd", Z_d, ZPR64>; 382 defm LD1RQ_B : sve_mem_ldqr_ss<0b00, "ld1rqb", Z_b, ZPR8, GPR64NoXZRshifted8>; 383 defm LD1RQ_H : sve_mem_ldqr_ss<0b01, "ld1rqh", Z_h, ZPR16, GPR64NoXZRshifted16>; 384 defm LD1RQ_W : sve_mem_ldqr_ss<0b10, "ld1rqw", Z_s, ZPR32, GPR64NoXZRshifted32>; 385 defm LD1RQ_D : sve_mem_ldqr_ss<0b11, "ld1rqd", Z_d, ZPR64, GPR64NoXZRshifted64>; 386 387 // continuous load with reg+reg addressing. 388 defm LD1B : sve_mem_cld_ss<0b0000, "ld1b", Z_b, ZPR8, GPR64NoXZRshifted8>; 389 defm LD1B_H : sve_mem_cld_ss<0b0001, "ld1b", Z_h, ZPR16, GPR64NoXZRshifted8>; 390 defm LD1B_S : sve_mem_cld_ss<0b0010, "ld1b", Z_s, ZPR32, GPR64NoXZRshifted8>; 391 defm LD1B_D : sve_mem_cld_ss<0b0011, "ld1b", Z_d, ZPR64, GPR64NoXZRshifted8>; 392 defm LD1SW_D : sve_mem_cld_ss<0b0100, "ld1sw", Z_d, ZPR64, GPR64NoXZRshifted32>; 393 defm LD1H : sve_mem_cld_ss<0b0101, "ld1h", Z_h, ZPR16, GPR64NoXZRshifted16>; 394 defm LD1H_S : sve_mem_cld_ss<0b0110, "ld1h", Z_s, ZPR32, GPR64NoXZRshifted16>; 395 defm LD1H_D : sve_mem_cld_ss<0b0111, "ld1h", Z_d, ZPR64, GPR64NoXZRshifted16>; 396 defm LD1SH_D : sve_mem_cld_ss<0b1000, "ld1sh", Z_d, ZPR64, GPR64NoXZRshifted16>; 397 defm LD1SH_S : sve_mem_cld_ss<0b1001, "ld1sh", Z_s, ZPR32, GPR64NoXZRshifted16>; 398 defm LD1W : sve_mem_cld_ss<0b1010, "ld1w", Z_s, ZPR32, GPR64NoXZRshifted32>; 399 defm LD1W_D : sve_mem_cld_ss<0b1011, "ld1w", Z_d, ZPR64, GPR64NoXZRshifted32>; 400 defm LD1SB_D : sve_mem_cld_ss<0b1100, "ld1sb", Z_d, ZPR64, GPR64NoXZRshifted8>; 401 defm LD1SB_S : sve_mem_cld_ss<0b1101, "ld1sb", Z_s, ZPR32, GPR64NoXZRshifted8>; 402 defm LD1SB_H : sve_mem_cld_ss<0b1110, "ld1sb", Z_h, ZPR16, GPR64NoXZRshifted8>; 403 defm LD1D : sve_mem_cld_ss<0b1111, "ld1d", Z_d, ZPR64, GPR64NoXZRshifted64>; 404 405 // non-faulting continuous load with reg+immediate 406 defm LDNF1B_IMM : sve_mem_cldnf_si<0b0000, "ldnf1b", Z_b, ZPR8>; 407 defm LDNF1B_H_IMM : sve_mem_cldnf_si<0b0001, "ldnf1b", Z_h, ZPR16>; 408 defm LDNF1B_S_IMM : sve_mem_cldnf_si<0b0010, "ldnf1b", Z_s, ZPR32>; 409 defm LDNF1B_D_IMM : sve_mem_cldnf_si<0b0011, "ldnf1b", Z_d, ZPR64>; 410 defm LDNF1SW_D_IMM : sve_mem_cldnf_si<0b0100, "ldnf1sw", Z_d, ZPR64>; 411 defm LDNF1H_IMM : sve_mem_cldnf_si<0b0101, "ldnf1h", Z_h, ZPR16>; 412 defm LDNF1H_S_IMM : sve_mem_cldnf_si<0b0110, "ldnf1h", Z_s, ZPR32>; 413 defm LDNF1H_D_IMM : sve_mem_cldnf_si<0b0111, "ldnf1h", Z_d, ZPR64>; 414 defm LDNF1SH_D_IMM : sve_mem_cldnf_si<0b1000, "ldnf1sh", Z_d, ZPR64>; 415 defm LDNF1SH_S_IMM : sve_mem_cldnf_si<0b1001, "ldnf1sh", Z_s, ZPR32>; 416 defm LDNF1W_IMM : sve_mem_cldnf_si<0b1010, "ldnf1w", Z_s, ZPR32>; 417 defm LDNF1W_D_IMM : sve_mem_cldnf_si<0b1011, "ldnf1w", Z_d, ZPR64>; 418 defm LDNF1SB_D_IMM : sve_mem_cldnf_si<0b1100, "ldnf1sb", Z_d, ZPR64>; 419 defm LDNF1SB_S_IMM : sve_mem_cldnf_si<0b1101, "ldnf1sb", Z_s, ZPR32>; 420 defm LDNF1SB_H_IMM : sve_mem_cldnf_si<0b1110, "ldnf1sb", Z_h, ZPR16>; 421 defm LDNF1D_IMM : sve_mem_cldnf_si<0b1111, "ldnf1d", Z_d, ZPR64>; 422 423 // First-faulting loads with reg+reg addressing. 424 defm LDFF1B : sve_mem_cldff_ss<0b0000, "ldff1b", Z_b, ZPR8, GPR64shifted8>; 425 defm LDFF1B_H : sve_mem_cldff_ss<0b0001, "ldff1b", Z_h, ZPR16, GPR64shifted8>; 426 defm LDFF1B_S : sve_mem_cldff_ss<0b0010, "ldff1b", Z_s, ZPR32, GPR64shifted8>; 427 defm LDFF1B_D : sve_mem_cldff_ss<0b0011, "ldff1b", Z_d, ZPR64, GPR64shifted8>; 428 defm LDFF1SW_D : sve_mem_cldff_ss<0b0100, "ldff1sw", Z_d, ZPR64, GPR64shifted32>; 429 defm LDFF1H : sve_mem_cldff_ss<0b0101, "ldff1h", Z_h, ZPR16, GPR64shifted16>; 430 defm LDFF1H_S : sve_mem_cldff_ss<0b0110, "ldff1h", Z_s, ZPR32, GPR64shifted16>; 431 defm LDFF1H_D : sve_mem_cldff_ss<0b0111, "ldff1h", Z_d, ZPR64, GPR64shifted16>; 432 defm LDFF1SH_D : sve_mem_cldff_ss<0b1000, "ldff1sh", Z_d, ZPR64, GPR64shifted16>; 433 defm LDFF1SH_S : sve_mem_cldff_ss<0b1001, "ldff1sh", Z_s, ZPR32, GPR64shifted16>; 434 defm LDFF1W : sve_mem_cldff_ss<0b1010, "ldff1w", Z_s, ZPR32, GPR64shifted32>; 435 defm LDFF1W_D : sve_mem_cldff_ss<0b1011, "ldff1w", Z_d, ZPR64, GPR64shifted32>; 436 defm LDFF1SB_D : sve_mem_cldff_ss<0b1100, "ldff1sb", Z_d, ZPR64, GPR64shifted8>; 437 defm LDFF1SB_S : sve_mem_cldff_ss<0b1101, "ldff1sb", Z_s, ZPR32, GPR64shifted8>; 438 defm LDFF1SB_H : sve_mem_cldff_ss<0b1110, "ldff1sb", Z_h, ZPR16, GPR64shifted8>; 439 defm LDFF1D : sve_mem_cldff_ss<0b1111, "ldff1d", Z_d, ZPR64, GPR64shifted64>; 440 441 // LD(2|3|4) structured loads with reg+immediate 442 defm LD2B_IMM : sve_mem_eld_si<0b00, 0b01, ZZ_b, "ld2b", simm4s2>; 443 defm LD3B_IMM : sve_mem_eld_si<0b00, 0b10, ZZZ_b, "ld3b", simm4s3>; 444 defm LD4B_IMM : sve_mem_eld_si<0b00, 0b11, ZZZZ_b, "ld4b", simm4s4>; 445 defm LD2H_IMM : sve_mem_eld_si<0b01, 0b01, ZZ_h, "ld2h", simm4s2>; 446 defm LD3H_IMM : sve_mem_eld_si<0b01, 0b10, ZZZ_h, "ld3h", simm4s3>; 447 defm LD4H_IMM : sve_mem_eld_si<0b01, 0b11, ZZZZ_h, "ld4h", simm4s4>; 448 defm LD2W_IMM : sve_mem_eld_si<0b10, 0b01, ZZ_s, "ld2w", simm4s2>; 449 defm LD3W_IMM : sve_mem_eld_si<0b10, 0b10, ZZZ_s, "ld3w", simm4s3>; 450 defm LD4W_IMM : sve_mem_eld_si<0b10, 0b11, ZZZZ_s, "ld4w", simm4s4>; 451 defm LD2D_IMM : sve_mem_eld_si<0b11, 0b01, ZZ_d, "ld2d", simm4s2>; 452 defm LD3D_IMM : sve_mem_eld_si<0b11, 0b10, ZZZ_d, "ld3d", simm4s3>; 453 defm LD4D_IMM : sve_mem_eld_si<0b11, 0b11, ZZZZ_d, "ld4d", simm4s4>; 454 455 // LD(2|3|4) structured loads (register + register) 456 def LD2B : sve_mem_eld_ss<0b00, 0b01, ZZ_b, "ld2b", GPR64NoXZRshifted8>; 457 def LD3B : sve_mem_eld_ss<0b00, 0b10, ZZZ_b, "ld3b", GPR64NoXZRshifted8>; 458 def LD4B : sve_mem_eld_ss<0b00, 0b11, ZZZZ_b, "ld4b", GPR64NoXZRshifted8>; 459 def LD2H : sve_mem_eld_ss<0b01, 0b01, ZZ_h, "ld2h", GPR64NoXZRshifted16>; 460 def LD3H : sve_mem_eld_ss<0b01, 0b10, ZZZ_h, "ld3h", GPR64NoXZRshifted16>; 461 def LD4H : sve_mem_eld_ss<0b01, 0b11, ZZZZ_h, "ld4h", GPR64NoXZRshifted16>; 462 def LD2W : sve_mem_eld_ss<0b10, 0b01, ZZ_s, "ld2w", GPR64NoXZRshifted32>; 463 def LD3W : sve_mem_eld_ss<0b10, 0b10, ZZZ_s, "ld3w", GPR64NoXZRshifted32>; 464 def LD4W : sve_mem_eld_ss<0b10, 0b11, ZZZZ_s, "ld4w", GPR64NoXZRshifted32>; 465 def LD2D : sve_mem_eld_ss<0b11, 0b01, ZZ_d, "ld2d", GPR64NoXZRshifted64>; 466 def LD3D : sve_mem_eld_ss<0b11, 0b10, ZZZ_d, "ld3d", GPR64NoXZRshifted64>; 467 def LD4D : sve_mem_eld_ss<0b11, 0b11, ZZZZ_d, "ld4d", GPR64NoXZRshifted64>; 468 469 // Gathers using unscaled 32-bit offsets, e.g. 470 // ld1h z0.s, p0/z, [x0, z0.s, uxtw] 471 defm GLD1SB_S : sve_mem_32b_gld_vs_32_unscaled<0b0000, "ld1sb", AArch64ld1s_gather_sxtw, AArch64ld1s_gather_uxtw, ZPR32ExtSXTW8Only, ZPR32ExtUXTW8Only, nxv4i8>; 472 defm GLDFF1SB_S : sve_mem_32b_gld_vs_32_unscaled<0b0001, "ldff1sb", null_frag, null_frag, ZPR32ExtSXTW8Only, ZPR32ExtUXTW8Only, nxv4i8>; 473 defm GLD1B_S : sve_mem_32b_gld_vs_32_unscaled<0b0010, "ld1b", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR32ExtSXTW8Only, ZPR32ExtUXTW8Only, nxv4i8>; 474 defm GLDFF1B_S : sve_mem_32b_gld_vs_32_unscaled<0b0011, "ldff1b", null_frag, null_frag, ZPR32ExtSXTW8Only, ZPR32ExtUXTW8Only, nxv4i8>; 475 defm GLD1SH_S : sve_mem_32b_gld_vs_32_unscaled<0b0100, "ld1sh", AArch64ld1s_gather_sxtw, AArch64ld1s_gather_uxtw, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i16>; 476 defm GLDFF1SH_S : sve_mem_32b_gld_vs_32_unscaled<0b0101, "ldff1sh", null_frag, null_frag, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i16>; 477 defm GLD1H_S : sve_mem_32b_gld_vs_32_unscaled<0b0110, "ld1h", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i16>; 478 defm GLDFF1H_S : sve_mem_32b_gld_vs_32_unscaled<0b0111, "ldff1h", null_frag, null_frag, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i16>; 479 defm GLD1W : sve_mem_32b_gld_vs_32_unscaled<0b1010, "ld1w", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i32>; 480 defm GLDFF1W : sve_mem_32b_gld_vs_32_unscaled<0b1011, "ldff1w", null_frag, null_frag, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i32>; 481 482 // Gathers using scaled 32-bit offsets, e.g. 483 // ld1h z0.s, p0/z, [x0, z0.s, uxtw #1] 484 defm GLD1SH_S : sve_mem_32b_gld_sv_32_scaled<0b0100, "ld1sh", AArch64ld1s_gather_sxtw_scaled, AArch64ld1s_gather_uxtw_scaled, ZPR32ExtSXTW16, ZPR32ExtUXTW16, nxv4i16>; 485 defm GLDFF1SH_S : sve_mem_32b_gld_sv_32_scaled<0b0101, "ldff1sh", null_frag, null_frag, ZPR32ExtSXTW16, ZPR32ExtUXTW16, nxv4i16>; 486 defm GLD1H_S : sve_mem_32b_gld_sv_32_scaled<0b0110, "ld1h", AArch64ld1_gather_sxtw_scaled, AArch64ld1_gather_uxtw_scaled, ZPR32ExtSXTW16, ZPR32ExtUXTW16, nxv4i16>; 487 defm GLDFF1H_S : sve_mem_32b_gld_sv_32_scaled<0b0111, "ldff1h", null_frag, null_frag, ZPR32ExtSXTW16, ZPR32ExtUXTW16, nxv4i16>; 488 defm GLD1W : sve_mem_32b_gld_sv_32_scaled<0b1010, "ld1w", AArch64ld1_gather_sxtw_scaled, AArch64ld1_gather_uxtw_scaled, ZPR32ExtSXTW32, ZPR32ExtUXTW32, nxv4i32>; 489 defm GLDFF1W : sve_mem_32b_gld_sv_32_scaled<0b1011, "ldff1w", null_frag, null_frag, ZPR32ExtSXTW32, ZPR32ExtUXTW32, nxv4i32>; 490 491 // Gathers using 32-bit pointers with scaled offset, e.g. 492 // ld1h z0.s, p0/z, [z0.s, #16] 493 defm GLD1SB_S : sve_mem_32b_gld_vi_32_ptrs<0b0000, "ld1sb", imm0_31, AArch64ld1s_gather_imm, nxv4i8>; 494 defm GLDFF1SB_S : sve_mem_32b_gld_vi_32_ptrs<0b0001, "ldff1sb", imm0_31, null_frag, nxv4i8>; 495 defm GLD1B_S : sve_mem_32b_gld_vi_32_ptrs<0b0010, "ld1b", imm0_31, AArch64ld1_gather_imm, nxv4i8>; 496 defm GLDFF1B_S : sve_mem_32b_gld_vi_32_ptrs<0b0011, "ldff1b", imm0_31, null_frag, nxv4i8>; 497 defm GLD1SH_S : sve_mem_32b_gld_vi_32_ptrs<0b0100, "ld1sh", uimm5s2, AArch64ld1s_gather_imm, nxv4i16>; 498 defm GLDFF1SH_S : sve_mem_32b_gld_vi_32_ptrs<0b0101, "ldff1sh", uimm5s2, null_frag, nxv4i16>; 499 defm GLD1H_S : sve_mem_32b_gld_vi_32_ptrs<0b0110, "ld1h", uimm5s2, AArch64ld1_gather_imm, nxv4i16>; 500 defm GLDFF1H_S : sve_mem_32b_gld_vi_32_ptrs<0b0111, "ldff1h", uimm5s2, null_frag, nxv4i16>; 501 defm GLD1W : sve_mem_32b_gld_vi_32_ptrs<0b1010, "ld1w", uimm5s4, AArch64ld1_gather_imm, nxv4i32>; 502 defm GLDFF1W : sve_mem_32b_gld_vi_32_ptrs<0b1011, "ldff1w", uimm5s4, null_frag, nxv4i32>; 503 504 // Gathers using 64-bit pointers with scaled offset, e.g. 505 // ld1h z0.d, p0/z, [z0.d, #16] 506 defm GLD1SB_D : sve_mem_64b_gld_vi_64_ptrs<0b0000, "ld1sb", imm0_31, AArch64ld1s_gather_imm, nxv2i8>; 507 defm GLDFF1SB_D : sve_mem_64b_gld_vi_64_ptrs<0b0001, "ldff1sb", imm0_31, null_frag, nxv2i8>; 508 defm GLD1B_D : sve_mem_64b_gld_vi_64_ptrs<0b0010, "ld1b", imm0_31, AArch64ld1_gather_imm, nxv2i8>; 509 defm GLDFF1B_D : sve_mem_64b_gld_vi_64_ptrs<0b0011, "ldff1b", imm0_31, null_frag, nxv2i8>; 510 defm GLD1SH_D : sve_mem_64b_gld_vi_64_ptrs<0b0100, "ld1sh", uimm5s2, AArch64ld1s_gather_imm, nxv2i16>; 511 defm GLDFF1SH_D : sve_mem_64b_gld_vi_64_ptrs<0b0101, "ldff1sh", uimm5s2, null_frag, nxv2i16>; 512 defm GLD1H_D : sve_mem_64b_gld_vi_64_ptrs<0b0110, "ld1h", uimm5s2, AArch64ld1_gather_imm, nxv2i16>; 513 defm GLDFF1H_D : sve_mem_64b_gld_vi_64_ptrs<0b0111, "ldff1h", uimm5s2, null_frag, nxv2i16>; 514 defm GLD1SW_D : sve_mem_64b_gld_vi_64_ptrs<0b1000, "ld1sw", uimm5s4, AArch64ld1s_gather_imm, nxv2i32>; 515 defm GLDFF1SW_D : sve_mem_64b_gld_vi_64_ptrs<0b1001, "ldff1sw", uimm5s4, null_frag, nxv2i32>; 516 defm GLD1W_D : sve_mem_64b_gld_vi_64_ptrs<0b1010, "ld1w", uimm5s4, AArch64ld1_gather_imm, nxv2i32>; 517 defm GLDFF1W_D : sve_mem_64b_gld_vi_64_ptrs<0b1011, "ldff1w", uimm5s4, null_frag, nxv2i32>; 518 defm GLD1D : sve_mem_64b_gld_vi_64_ptrs<0b1110, "ld1d", uimm5s8, AArch64ld1_gather_imm, nxv2i64>; 519 defm GLDFF1D : sve_mem_64b_gld_vi_64_ptrs<0b1111, "ldff1d", uimm5s8, null_frag, nxv2i64>; 520 521 // Gathers using unscaled 64-bit offsets, e.g. 522 // ld1h z0.d, p0/z, [x0, z0.d] 523 defm GLD1SB_D : sve_mem_64b_gld_vs2_64_unscaled<0b0000, "ld1sb", AArch64ld1s_gather, nxv2i8>; 524 defm GLDFF1SB_D : sve_mem_64b_gld_vs2_64_unscaled<0b0001, "ldff1sb", null_frag, nxv2i8>; 525 defm GLD1B_D : sve_mem_64b_gld_vs2_64_unscaled<0b0010, "ld1b", AArch64ld1_gather, nxv2i8>; 526 defm GLDFF1B_D : sve_mem_64b_gld_vs2_64_unscaled<0b0011, "ldff1b", null_frag, nxv2i8>; 527 defm GLD1SH_D : sve_mem_64b_gld_vs2_64_unscaled<0b0100, "ld1sh", AArch64ld1s_gather, nxv2i16>; 528 defm GLDFF1SH_D : sve_mem_64b_gld_vs2_64_unscaled<0b0101, "ldff1sh", null_frag, nxv2i16>; 529 defm GLD1H_D : sve_mem_64b_gld_vs2_64_unscaled<0b0110, "ld1h", AArch64ld1_gather, nxv2i16>; 530 defm GLDFF1H_D : sve_mem_64b_gld_vs2_64_unscaled<0b0111, "ldff1h", null_frag, nxv2i16>; 531 defm GLD1SW_D : sve_mem_64b_gld_vs2_64_unscaled<0b1000, "ld1sw", AArch64ld1s_gather, nxv2i32>; 532 defm GLDFF1SW_D : sve_mem_64b_gld_vs2_64_unscaled<0b1001, "ldff1sw", null_frag, nxv2i32>; 533 defm GLD1W_D : sve_mem_64b_gld_vs2_64_unscaled<0b1010, "ld1w", AArch64ld1_gather, nxv2i32>; 534 defm GLDFF1W_D : sve_mem_64b_gld_vs2_64_unscaled<0b1011, "ldff1w", null_frag, nxv2i32>; 535 defm GLD1D : sve_mem_64b_gld_vs2_64_unscaled<0b1110, "ld1d", AArch64ld1_gather, nxv2i64>; 536 defm GLDFF1D : sve_mem_64b_gld_vs2_64_unscaled<0b1111, "ldff1d", null_frag, nxv2i64>; 537 538 // Gathers using scaled 64-bit offsets, e.g. 539 // ld1h z0.d, p0/z, [x0, z0.d, lsl #1] 540 defm GLD1SH_D : sve_mem_64b_gld_sv2_64_scaled<0b0100, "ld1sh", AArch64ld1s_gather_scaled, ZPR64ExtLSL16, nxv2i16>; 541 defm GLDFF1SH_D : sve_mem_64b_gld_sv2_64_scaled<0b0101, "ldff1sh", null_frag, ZPR64ExtLSL16, nxv2i16>; 542 defm GLD1H_D : sve_mem_64b_gld_sv2_64_scaled<0b0110, "ld1h", AArch64ld1_gather_scaled, ZPR64ExtLSL16, nxv2i16>; 543 defm GLDFF1H_D : sve_mem_64b_gld_sv2_64_scaled<0b0111, "ldff1h", null_frag, ZPR64ExtLSL16, nxv2i16>; 544 defm GLD1SW_D : sve_mem_64b_gld_sv2_64_scaled<0b1000, "ld1sw", AArch64ld1s_gather_scaled, ZPR64ExtLSL32, nxv2i32>; 545 defm GLDFF1SW_D : sve_mem_64b_gld_sv2_64_scaled<0b1001, "ldff1sw", null_frag, ZPR64ExtLSL32, nxv2i32>; 546 defm GLD1W_D : sve_mem_64b_gld_sv2_64_scaled<0b1010, "ld1w", AArch64ld1_gather_scaled, ZPR64ExtLSL32, nxv2i32>; 547 defm GLDFF1W_D : sve_mem_64b_gld_sv2_64_scaled<0b1011, "ldff1w", null_frag, ZPR64ExtLSL32, nxv2i32>; 548 defm GLD1D : sve_mem_64b_gld_sv2_64_scaled<0b1110, "ld1d", AArch64ld1_gather_scaled, ZPR64ExtLSL64, nxv2i64>; 549 defm GLDFF1D : sve_mem_64b_gld_sv2_64_scaled<0b1111, "ldff1d", null_frag, ZPR64ExtLSL64, nxv2i64>; 550 551 // Gathers using unscaled 32-bit offsets unpacked in 64-bits elements, e.g. 552 // ld1h z0.d, p0/z, [x0, z0.d, uxtw] 553 defm GLD1SB_D : sve_mem_64b_gld_vs_32_unscaled<0b0000, "ld1sb", AArch64ld1s_gather_sxtw, AArch64ld1s_gather_uxtw, ZPR64ExtSXTW8Only, ZPR64ExtUXTW8Only, nxv2i8>; 554 defm GLDFF1SB_D : sve_mem_64b_gld_vs_32_unscaled<0b0001, "ldff1sb", null_frag, null_frag, ZPR64ExtSXTW8Only, ZPR64ExtUXTW8Only, nxv2i8>; 555 defm GLD1B_D : sve_mem_64b_gld_vs_32_unscaled<0b0010, "ld1b", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR64ExtSXTW8Only, ZPR64ExtUXTW8Only, nxv2i8>; 556 defm GLDFF1B_D : sve_mem_64b_gld_vs_32_unscaled<0b0011, "ldff1b", null_frag, null_frag, ZPR64ExtSXTW8Only, ZPR64ExtUXTW8Only, nxv2i8>; 557 defm GLD1SH_D : sve_mem_64b_gld_vs_32_unscaled<0b0100, "ld1sh", AArch64ld1s_gather_sxtw, AArch64ld1s_gather_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i16>; 558 defm GLDFF1SH_D : sve_mem_64b_gld_vs_32_unscaled<0b0101, "ldff1sh", null_frag, null_frag, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i16>; 559 defm GLD1H_D : sve_mem_64b_gld_vs_32_unscaled<0b0110, "ld1h", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i16>; 560 defm GLDFF1H_D : sve_mem_64b_gld_vs_32_unscaled<0b0111, "ldff1h", null_frag, null_frag, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i16>; 561 defm GLD1SW_D : sve_mem_64b_gld_vs_32_unscaled<0b1000, "ld1sw", AArch64ld1s_gather_sxtw, AArch64ld1s_gather_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i32>; 562 defm GLDFF1SW_D : sve_mem_64b_gld_vs_32_unscaled<0b1001, "ldff1sw", null_frag, null_frag, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i32>; 563 defm GLD1W_D : sve_mem_64b_gld_vs_32_unscaled<0b1010, "ld1w", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i32>; 564 defm GLDFF1W_D : sve_mem_64b_gld_vs_32_unscaled<0b1011, "ldff1w", null_frag, null_frag, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i32>; 565 defm GLD1D : sve_mem_64b_gld_vs_32_unscaled<0b1110, "ld1d", AArch64ld1_gather_sxtw, AArch64ld1_gather_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i64>; 566 defm GLDFF1D : sve_mem_64b_gld_vs_32_unscaled<0b1111, "ldff1d", null_frag, null_frag, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i64>; 567 568 // Gathers using scaled 32-bit offsets unpacked in 64-bits elements, e.g. 569 // ld1h z0.d, p0/z, [x0, z0.d, uxtw #1] 570 defm GLD1SH_D : sve_mem_64b_gld_sv_32_scaled<0b0100, "ld1sh", AArch64ld1s_gather_sxtw_scaled, AArch64ld1s_gather_uxtw_scaled, ZPR64ExtSXTW16, ZPR64ExtUXTW16, nxv2i16>; 571 defm GLDFF1SH_D : sve_mem_64b_gld_sv_32_scaled<0b0101, "ldff1sh", null_frag, null_frag, ZPR64ExtSXTW16, ZPR64ExtUXTW16, nxv2i16>; 572 defm GLD1H_D : sve_mem_64b_gld_sv_32_scaled<0b0110, "ld1h", AArch64ld1_gather_sxtw_scaled, AArch64ld1_gather_uxtw_scaled, ZPR64ExtSXTW16, ZPR64ExtUXTW16, nxv2i16>; 573 defm GLDFF1H_D : sve_mem_64b_gld_sv_32_scaled<0b0111, "ldff1h", null_frag, null_frag, ZPR64ExtSXTW16, ZPR64ExtUXTW16, nxv2i16>; 574 defm GLD1SW_D : sve_mem_64b_gld_sv_32_scaled<0b1000, "ld1sw", AArch64ld1s_gather_sxtw_scaled, AArch64ld1s_gather_uxtw_scaled, ZPR64ExtSXTW32, ZPR64ExtUXTW32, nxv2i32>; 575 defm GLDFF1SW_D : sve_mem_64b_gld_sv_32_scaled<0b1001, "ldff1sw", null_frag, null_frag, ZPR64ExtSXTW32, ZPR64ExtUXTW32, nxv2i32>; 576 defm GLD1W_D : sve_mem_64b_gld_sv_32_scaled<0b1010, "ld1w", AArch64ld1_gather_sxtw_scaled, AArch64ld1_gather_uxtw_scaled, ZPR64ExtSXTW32, ZPR64ExtUXTW32, nxv2i32>; 577 defm GLDFF1W_D : sve_mem_64b_gld_sv_32_scaled<0b1011, "ldff1w", null_frag, null_frag, ZPR64ExtSXTW32, ZPR64ExtUXTW32, nxv2i32>; 578 defm GLD1D : sve_mem_64b_gld_sv_32_scaled<0b1110, "ld1d", AArch64ld1_gather_sxtw_scaled, AArch64ld1_gather_uxtw_scaled, ZPR64ExtSXTW64, ZPR64ExtUXTW64, nxv2i64>; 579 defm GLDFF1D : sve_mem_64b_gld_sv_32_scaled<0b1111, "ldff1d", null_frag, null_frag, ZPR64ExtSXTW64, ZPR64ExtUXTW64, nxv2i64>; 580 581 // Non-temporal contiguous loads (register + immediate) 582 defm LDNT1B_ZRI : sve_mem_cldnt_si<0b00, "ldnt1b", Z_b, ZPR8>; 583 defm LDNT1H_ZRI : sve_mem_cldnt_si<0b01, "ldnt1h", Z_h, ZPR16>; 584 defm LDNT1W_ZRI : sve_mem_cldnt_si<0b10, "ldnt1w", Z_s, ZPR32>; 585 defm LDNT1D_ZRI : sve_mem_cldnt_si<0b11, "ldnt1d", Z_d, ZPR64>; 586 587 // Non-temporal contiguous loads (register + register) 588 defm LDNT1B_ZRR : sve_mem_cldnt_ss<0b00, "ldnt1b", Z_b, ZPR8, GPR64NoXZRshifted8>; 589 defm LDNT1H_ZRR : sve_mem_cldnt_ss<0b01, "ldnt1h", Z_h, ZPR16, GPR64NoXZRshifted16>; 590 defm LDNT1W_ZRR : sve_mem_cldnt_ss<0b10, "ldnt1w", Z_s, ZPR32, GPR64NoXZRshifted32>; 591 defm LDNT1D_ZRR : sve_mem_cldnt_ss<0b11, "ldnt1d", Z_d, ZPR64, GPR64NoXZRshifted64>; 592 593 // contiguous store with immediates 594 defm ST1B_IMM : sve_mem_cst_si<0b00, 0b00, "st1b", Z_b, ZPR8>; 595 defm ST1B_H_IMM : sve_mem_cst_si<0b00, 0b01, "st1b", Z_h, ZPR16>; 596 defm ST1B_S_IMM : sve_mem_cst_si<0b00, 0b10, "st1b", Z_s, ZPR32>; 597 defm ST1B_D_IMM : sve_mem_cst_si<0b00, 0b11, "st1b", Z_d, ZPR64>; 598 defm ST1H_IMM : sve_mem_cst_si<0b01, 0b01, "st1h", Z_h, ZPR16>; 599 defm ST1H_S_IMM : sve_mem_cst_si<0b01, 0b10, "st1h", Z_s, ZPR32>; 600 defm ST1H_D_IMM : sve_mem_cst_si<0b01, 0b11, "st1h", Z_d, ZPR64>; 601 defm ST1W_IMM : sve_mem_cst_si<0b10, 0b10, "st1w", Z_s, ZPR32>; 602 defm ST1W_D_IMM : sve_mem_cst_si<0b10, 0b11, "st1w", Z_d, ZPR64>; 603 defm ST1D_IMM : sve_mem_cst_si<0b11, 0b11, "st1d", Z_d, ZPR64>; 604 605 // contiguous store with reg+reg addressing. 606 defm ST1B : sve_mem_cst_ss<0b0000, "st1b", Z_b, ZPR8, GPR64NoXZRshifted8>; 607 defm ST1B_H : sve_mem_cst_ss<0b0001, "st1b", Z_h, ZPR16, GPR64NoXZRshifted8>; 608 defm ST1B_S : sve_mem_cst_ss<0b0010, "st1b", Z_s, ZPR32, GPR64NoXZRshifted8>; 609 defm ST1B_D : sve_mem_cst_ss<0b0011, "st1b", Z_d, ZPR64, GPR64NoXZRshifted8>; 610 defm ST1H : sve_mem_cst_ss<0b0101, "st1h", Z_h, ZPR16, GPR64NoXZRshifted16>; 611 defm ST1H_S : sve_mem_cst_ss<0b0110, "st1h", Z_s, ZPR32, GPR64NoXZRshifted16>; 612 defm ST1H_D : sve_mem_cst_ss<0b0111, "st1h", Z_d, ZPR64, GPR64NoXZRshifted16>; 613 defm ST1W : sve_mem_cst_ss<0b1010, "st1w", Z_s, ZPR32, GPR64NoXZRshifted32>; 614 defm ST1W_D : sve_mem_cst_ss<0b1011, "st1w", Z_d, ZPR64, GPR64NoXZRshifted32>; 615 defm ST1D : sve_mem_cst_ss<0b1111, "st1d", Z_d, ZPR64, GPR64NoXZRshifted64>; 616 617 // Scatters using unpacked, unscaled 32-bit offsets, e.g. 618 // st1h z0.d, p0, [x0, z0.d, uxtw] 619 defm SST1B_D : sve_mem_64b_sst_sv_32_unscaled<0b000, "st1b", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR64ExtSXTW8Only, ZPR64ExtUXTW8Only, nxv2i8>; 620 defm SST1H_D : sve_mem_64b_sst_sv_32_unscaled<0b010, "st1h", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i16>; 621 defm SST1W_D : sve_mem_64b_sst_sv_32_unscaled<0b100, "st1w", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8,nxv2i32>; 622 defm SST1D : sve_mem_64b_sst_sv_32_unscaled<0b110, "st1d", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR64ExtSXTW8, ZPR64ExtUXTW8, nxv2i64>; 623 624 // Scatters using packed, unscaled 32-bit offsets, e.g. 625 // st1h z0.s, p0, [x0, z0.s, uxtw] 626 defm SST1B_S : sve_mem_32b_sst_sv_32_unscaled<0b001, "st1b", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR32ExtSXTW8Only, ZPR32ExtUXTW8Only, nxv4i8>; 627 defm SST1H_S : sve_mem_32b_sst_sv_32_unscaled<0b011, "st1h", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i16>; 628 defm SST1W : sve_mem_32b_sst_sv_32_unscaled<0b101, "st1w", AArch64st1_scatter_sxtw, AArch64st1_scatter_uxtw, ZPR32ExtSXTW8, ZPR32ExtUXTW8, nxv4i32>; 629 630 // Scatters using packed, scaled 32-bit offsets, e.g. 631 // st1h z0.s, p0, [x0, z0.s, uxtw #1] 632 defm SST1H_S : sve_mem_32b_sst_sv_32_scaled<0b011, "st1h", AArch64st1_scatter_sxtw_scaled, AArch64st1_scatter_uxtw_scaled, ZPR32ExtSXTW16, ZPR32ExtUXTW16, nxv4i16>; 633 defm SST1W : sve_mem_32b_sst_sv_32_scaled<0b101, "st1w", AArch64st1_scatter_sxtw_scaled, AArch64st1_scatter_uxtw_scaled, ZPR32ExtSXTW32, ZPR32ExtUXTW32, nxv4i32>; 634 635 // Scatters using unpacked, scaled 32-bit offsets, e.g. 636 // st1h z0.d, p0, [x0, z0.d, uxtw #1] 637 defm SST1H_D : sve_mem_64b_sst_sv_32_scaled<0b010, "st1h", AArch64st1_scatter_sxtw_scaled, AArch64st1_scatter_uxtw_scaled, ZPR64ExtSXTW16, ZPR64ExtUXTW16, nxv2i16>; 638 defm SST1W_D : sve_mem_64b_sst_sv_32_scaled<0b100, "st1w", AArch64st1_scatter_sxtw_scaled, AArch64st1_scatter_uxtw_scaled, ZPR64ExtSXTW32, ZPR64ExtUXTW32, nxv2i32>; 639 defm SST1D : sve_mem_64b_sst_sv_32_scaled<0b110, "st1d", AArch64st1_scatter_sxtw_scaled, AArch64st1_scatter_uxtw_scaled, ZPR64ExtSXTW64, ZPR64ExtUXTW64, nxv2i64>; 640 641 // Scatters using 32/64-bit pointers with offset, e.g. 642 // st1h z0.s, p0, [z0.s, #16] 643 defm SST1B_S : sve_mem_32b_sst_vi_ptrs<0b001, "st1b", timm0_31, AArch64st1_scatter_imm, nxv4i8>; 644 defm SST1H_S : sve_mem_32b_sst_vi_ptrs<0b011, "st1h", tuimm5s2, AArch64st1_scatter_imm, nxv4i16>; 645 defm SST1W : sve_mem_32b_sst_vi_ptrs<0b101, "st1w", tuimm5s4, AArch64st1_scatter_imm, nxv4i32>; 646 647 // Scatters using 32/64-bit pointers with offset, e.g. 648 // st1h z0.d, p0, [z0.d, #16] 649 defm SST1B_D : sve_mem_64b_sst_vi_ptrs<0b000, "st1b", timm0_31, AArch64st1_scatter_imm, nxv2i8>; 650 defm SST1H_D : sve_mem_64b_sst_vi_ptrs<0b010, "st1h", tuimm5s2, AArch64st1_scatter_imm, nxv2i16>; 651 defm SST1W_D : sve_mem_64b_sst_vi_ptrs<0b100, "st1w", tuimm5s4, AArch64st1_scatter_imm, nxv2i32>; 652 defm SST1D : sve_mem_64b_sst_vi_ptrs<0b110, "st1d", tuimm5s8, AArch64st1_scatter_imm, nxv2i64>; 653 654 // Scatters using unscaled 64-bit offsets, e.g. 655 // st1h z0.d, p0, [x0, z0.d] 656 defm SST1B_D : sve_mem_sst_sv_64_unscaled<0b00, "st1b", AArch64st1_scatter, nxv2i8>; 657 defm SST1H_D : sve_mem_sst_sv_64_unscaled<0b01, "st1h", AArch64st1_scatter, nxv2i16>; 658 defm SST1W_D : sve_mem_sst_sv_64_unscaled<0b10, "st1w", AArch64st1_scatter, nxv2i32>; 659 defm SST1D : sve_mem_sst_sv_64_unscaled<0b11, "st1d", AArch64st1_scatter, nxv2i64>; 660 661 // Scatters using scaled 64-bit offsets, e.g. 662 // st1h z0.d, p0, [x0, z0.d, lsl #1] 663 defm SST1H_D_SCALED : sve_mem_sst_sv_64_scaled<0b01, "st1h", AArch64st1_scatter_scaled, ZPR64ExtLSL16, nxv2i16>; 664 defm SST1W_D_SCALED : sve_mem_sst_sv_64_scaled<0b10, "st1w", AArch64st1_scatter_scaled, ZPR64ExtLSL32, nxv2i32>; 665 defm SST1D_SCALED : sve_mem_sst_sv_64_scaled<0b11, "st1d", AArch64st1_scatter_scaled, ZPR64ExtLSL64, nxv2i64>; 666 667 // ST(2|3|4) structured stores (register + immediate) 668 defm ST2B_IMM : sve_mem_est_si<0b00, 0b01, ZZ_b, "st2b", simm4s2>; 669 defm ST3B_IMM : sve_mem_est_si<0b00, 0b10, ZZZ_b, "st3b", simm4s3>; 670 defm ST4B_IMM : sve_mem_est_si<0b00, 0b11, ZZZZ_b, "st4b", simm4s4>; 671 defm ST2H_IMM : sve_mem_est_si<0b01, 0b01, ZZ_h, "st2h", simm4s2>; 672 defm ST3H_IMM : sve_mem_est_si<0b01, 0b10, ZZZ_h, "st3h", simm4s3>; 673 defm ST4H_IMM : sve_mem_est_si<0b01, 0b11, ZZZZ_h, "st4h", simm4s4>; 674 defm ST2W_IMM : sve_mem_est_si<0b10, 0b01, ZZ_s, "st2w", simm4s2>; 675 defm ST3W_IMM : sve_mem_est_si<0b10, 0b10, ZZZ_s, "st3w", simm4s3>; 676 defm ST4W_IMM : sve_mem_est_si<0b10, 0b11, ZZZZ_s, "st4w", simm4s4>; 677 defm ST2D_IMM : sve_mem_est_si<0b11, 0b01, ZZ_d, "st2d", simm4s2>; 678 defm ST3D_IMM : sve_mem_est_si<0b11, 0b10, ZZZ_d, "st3d", simm4s3>; 679 defm ST4D_IMM : sve_mem_est_si<0b11, 0b11, ZZZZ_d, "st4d", simm4s4>; 680 681 // ST(2|3|4) structured stores (register + register) 682 def ST2B : sve_mem_est_ss<0b00, 0b01, ZZ_b, "st2b", GPR64NoXZRshifted8>; 683 def ST3B : sve_mem_est_ss<0b00, 0b10, ZZZ_b, "st3b", GPR64NoXZRshifted8>; 684 def ST4B : sve_mem_est_ss<0b00, 0b11, ZZZZ_b, "st4b", GPR64NoXZRshifted8>; 685 def ST2H : sve_mem_est_ss<0b01, 0b01, ZZ_h, "st2h", GPR64NoXZRshifted16>; 686 def ST3H : sve_mem_est_ss<0b01, 0b10, ZZZ_h, "st3h", GPR64NoXZRshifted16>; 687 def ST4H : sve_mem_est_ss<0b01, 0b11, ZZZZ_h, "st4h", GPR64NoXZRshifted16>; 688 def ST2W : sve_mem_est_ss<0b10, 0b01, ZZ_s, "st2w", GPR64NoXZRshifted32>; 689 def ST3W : sve_mem_est_ss<0b10, 0b10, ZZZ_s, "st3w", GPR64NoXZRshifted32>; 690 def ST4W : sve_mem_est_ss<0b10, 0b11, ZZZZ_s, "st4w", GPR64NoXZRshifted32>; 691 def ST2D : sve_mem_est_ss<0b11, 0b01, ZZ_d, "st2d", GPR64NoXZRshifted64>; 692 def ST3D : sve_mem_est_ss<0b11, 0b10, ZZZ_d, "st3d", GPR64NoXZRshifted64>; 693 def ST4D : sve_mem_est_ss<0b11, 0b11, ZZZZ_d, "st4d", GPR64NoXZRshifted64>; 694 695 // Non-temporal contiguous stores (register + immediate) 696 defm STNT1B_ZRI : sve_mem_cstnt_si<0b00, "stnt1b", Z_b, ZPR8>; 697 defm STNT1H_ZRI : sve_mem_cstnt_si<0b01, "stnt1h", Z_h, ZPR16>; 698 defm STNT1W_ZRI : sve_mem_cstnt_si<0b10, "stnt1w", Z_s, ZPR32>; 699 defm STNT1D_ZRI : sve_mem_cstnt_si<0b11, "stnt1d", Z_d, ZPR64>; 700 701 // Non-temporal contiguous stores (register + register) 702 defm STNT1B_ZRR : sve_mem_cstnt_ss<0b00, "stnt1b", Z_b, ZPR8, GPR64NoXZRshifted8>; 703 defm STNT1H_ZRR : sve_mem_cstnt_ss<0b01, "stnt1h", Z_h, ZPR16, GPR64NoXZRshifted16>; 704 defm STNT1W_ZRR : sve_mem_cstnt_ss<0b10, "stnt1w", Z_s, ZPR32, GPR64NoXZRshifted32>; 705 defm STNT1D_ZRR : sve_mem_cstnt_ss<0b11, "stnt1d", Z_d, ZPR64, GPR64NoXZRshifted64>; 706 707 // Fill/Spill 708 defm LDR_ZXI : sve_mem_z_fill<"ldr">; 709 defm LDR_PXI : sve_mem_p_fill<"ldr">; 710 defm STR_ZXI : sve_mem_z_spill<"str">; 711 defm STR_PXI : sve_mem_p_spill<"str">; 712 713 // Contiguous prefetch (register + immediate) 714 defm PRFB_PRI : sve_mem_prfm_si<0b00, "prfb">; 715 defm PRFH_PRI : sve_mem_prfm_si<0b01, "prfh">; 716 defm PRFW_PRI : sve_mem_prfm_si<0b10, "prfw">; 717 defm PRFD_PRI : sve_mem_prfm_si<0b11, "prfd">; 718 719 // Contiguous prefetch (register + register) 720 def PRFB_PRR : sve_mem_prfm_ss<0b001, "prfb", GPR64NoXZRshifted8>; 721 def PRFH_PRR : sve_mem_prfm_ss<0b011, "prfh", GPR64NoXZRshifted16>; 722 def PRFS_PRR : sve_mem_prfm_ss<0b101, "prfw", GPR64NoXZRshifted32>; 723 def PRFD_PRR : sve_mem_prfm_ss<0b111, "prfd", GPR64NoXZRshifted64>; 724 725 // Gather prefetch using scaled 32-bit offsets, e.g. 726 // prfh pldl1keep, p0, [x0, z0.s, uxtw #1] 727 defm PRFB_S : sve_mem_32b_prfm_sv_scaled<0b00, "prfb", ZPR32ExtSXTW8Only, ZPR32ExtUXTW8Only>; 728 defm PRFH_S : sve_mem_32b_prfm_sv_scaled<0b01, "prfh", ZPR32ExtSXTW16, ZPR32ExtUXTW16>; 729 defm PRFW_S : sve_mem_32b_prfm_sv_scaled<0b10, "prfw", ZPR32ExtSXTW32, ZPR32ExtUXTW32>; 730 defm PRFD_S : sve_mem_32b_prfm_sv_scaled<0b11, "prfd", ZPR32ExtSXTW64, ZPR32ExtUXTW64>; 731 732 // Gather prefetch using unpacked, scaled 32-bit offsets, e.g. 733 // prfh pldl1keep, p0, [x0, z0.d, uxtw #1] 734 defm PRFB_D : sve_mem_64b_prfm_sv_ext_scaled<0b00, "prfb", ZPR64ExtSXTW8Only, ZPR64ExtUXTW8Only>; 735 defm PRFH_D : sve_mem_64b_prfm_sv_ext_scaled<0b01, "prfh", ZPR64ExtSXTW16, ZPR64ExtUXTW16>; 736 defm PRFW_D : sve_mem_64b_prfm_sv_ext_scaled<0b10, "prfw", ZPR64ExtSXTW32, ZPR64ExtUXTW32>; 737 defm PRFD_D : sve_mem_64b_prfm_sv_ext_scaled<0b11, "prfd", ZPR64ExtSXTW64, ZPR64ExtUXTW64>; 738 739 // Gather prefetch using scaled 64-bit offsets, e.g. 740 // prfh pldl1keep, p0, [x0, z0.d, lsl #1] 741 defm PRFB_D_SCALED : sve_mem_64b_prfm_sv_lsl_scaled<0b00, "prfb", ZPR64ExtLSL8>; 742 defm PRFH_D_SCALED : sve_mem_64b_prfm_sv_lsl_scaled<0b01, "prfh", ZPR64ExtLSL16>; 743 defm PRFW_D_SCALED : sve_mem_64b_prfm_sv_lsl_scaled<0b10, "prfw", ZPR64ExtLSL32>; 744 defm PRFD_D_SCALED : sve_mem_64b_prfm_sv_lsl_scaled<0b11, "prfd", ZPR64ExtLSL64>; 745 746 // Gather prefetch using 32/64-bit pointers with offset, e.g. 747 // prfh pldl1keep, p0, [z0.s, #16] 748 // prfh pldl1keep, p0, [z0.d, #16] 749 defm PRFB_S_PZI : sve_mem_32b_prfm_vi<0b00, "prfb", imm0_31>; 750 defm PRFH_S_PZI : sve_mem_32b_prfm_vi<0b01, "prfh", uimm5s2>; 751 defm PRFW_S_PZI : sve_mem_32b_prfm_vi<0b10, "prfw", uimm5s4>; 752 defm PRFD_S_PZI : sve_mem_32b_prfm_vi<0b11, "prfd", uimm5s8>; 753 754 defm PRFB_D_PZI : sve_mem_64b_prfm_vi<0b00, "prfb", imm0_31>; 755 defm PRFH_D_PZI : sve_mem_64b_prfm_vi<0b01, "prfh", uimm5s2>; 756 defm PRFW_D_PZI : sve_mem_64b_prfm_vi<0b10, "prfw", uimm5s4>; 757 defm PRFD_D_PZI : sve_mem_64b_prfm_vi<0b11, "prfd", uimm5s8>; 758 759 defm ADR_SXTW_ZZZ_D : sve_int_bin_cons_misc_0_a_sxtw<0b00, "adr">; 760 defm ADR_UXTW_ZZZ_D : sve_int_bin_cons_misc_0_a_uxtw<0b01, "adr">; 761 defm ADR_LSL_ZZZ_S : sve_int_bin_cons_misc_0_a_32_lsl<0b10, "adr">; 762 defm ADR_LSL_ZZZ_D : sve_int_bin_cons_misc_0_a_64_lsl<0b11, "adr">; 763 764 defm TBL_ZZZ : sve_int_perm_tbl<"tbl", AArch64tbl>; 765 766 defm ZIP1_ZZZ : sve_int_perm_bin_perm_zz<0b000, "zip1", AArch64zip1>; 767 defm ZIP2_ZZZ : sve_int_perm_bin_perm_zz<0b001, "zip2", AArch64zip2>; 768 defm UZP1_ZZZ : sve_int_perm_bin_perm_zz<0b010, "uzp1", AArch64uzp1>; 769 defm UZP2_ZZZ : sve_int_perm_bin_perm_zz<0b011, "uzp2", AArch64uzp2>; 770 defm TRN1_ZZZ : sve_int_perm_bin_perm_zz<0b100, "trn1", AArch64trn1>; 771 defm TRN2_ZZZ : sve_int_perm_bin_perm_zz<0b101, "trn2", AArch64trn2>; 772 773 defm ZIP1_PPP : sve_int_perm_bin_perm_pp<0b000, "zip1", AArch64zip1>; 774 defm ZIP2_PPP : sve_int_perm_bin_perm_pp<0b001, "zip2", AArch64zip2>; 775 defm UZP1_PPP : sve_int_perm_bin_perm_pp<0b010, "uzp1", AArch64uzp1>; 776 defm UZP2_PPP : sve_int_perm_bin_perm_pp<0b011, "uzp2", AArch64uzp2>; 777 defm TRN1_PPP : sve_int_perm_bin_perm_pp<0b100, "trn1", AArch64trn1>; 778 defm TRN2_PPP : sve_int_perm_bin_perm_pp<0b101, "trn2", AArch64trn2>; 779 780 defm CMPHS_PPzZZ : sve_int_cmp_0<0b000, "cmphs", int_aarch64_sve_cmphs, SETUGE>; 781 defm CMPHI_PPzZZ : sve_int_cmp_0<0b001, "cmphi", int_aarch64_sve_cmphi, SETUGT>; 782 defm CMPGE_PPzZZ : sve_int_cmp_0<0b100, "cmpge", int_aarch64_sve_cmpge, SETGE>; 783 defm CMPGT_PPzZZ : sve_int_cmp_0<0b101, "cmpgt", int_aarch64_sve_cmpgt, SETGT>; 784 defm CMPEQ_PPzZZ : sve_int_cmp_0<0b110, "cmpeq", int_aarch64_sve_cmpeq, SETEQ>; 785 defm CMPNE_PPzZZ : sve_int_cmp_0<0b111, "cmpne", int_aarch64_sve_cmpne, SETNE>; 786 787 defm CMPEQ_WIDE_PPzZZ : sve_int_cmp_0_wide<0b010, "cmpeq", int_aarch64_sve_cmpeq_wide>; 788 defm CMPNE_WIDE_PPzZZ : sve_int_cmp_0_wide<0b011, "cmpne", int_aarch64_sve_cmpne_wide>; 789 defm CMPGE_WIDE_PPzZZ : sve_int_cmp_1_wide<0b000, "cmpge", int_aarch64_sve_cmpge_wide>; 790 defm CMPGT_WIDE_PPzZZ : sve_int_cmp_1_wide<0b001, "cmpgt", int_aarch64_sve_cmpgt_wide>; 791 defm CMPLT_WIDE_PPzZZ : sve_int_cmp_1_wide<0b010, "cmplt", int_aarch64_sve_cmplt_wide>; 792 defm CMPLE_WIDE_PPzZZ : sve_int_cmp_1_wide<0b011, "cmple", int_aarch64_sve_cmple_wide>; 793 defm CMPHS_WIDE_PPzZZ : sve_int_cmp_1_wide<0b100, "cmphs", int_aarch64_sve_cmphs_wide>; 794 defm CMPHI_WIDE_PPzZZ : sve_int_cmp_1_wide<0b101, "cmphi", int_aarch64_sve_cmphi_wide>; 795 defm CMPLO_WIDE_PPzZZ : sve_int_cmp_1_wide<0b110, "cmplo", int_aarch64_sve_cmplo_wide>; 796 defm CMPLS_WIDE_PPzZZ : sve_int_cmp_1_wide<0b111, "cmpls", int_aarch64_sve_cmpls_wide>; 797 798 defm CMPGE_PPzZI : sve_int_scmp_vi<0b000, "cmpge", SETGE, int_aarch64_sve_cmpge>; 799 defm CMPGT_PPzZI : sve_int_scmp_vi<0b001, "cmpgt", SETGT, int_aarch64_sve_cmpgt>; 800 defm CMPLT_PPzZI : sve_int_scmp_vi<0b010, "cmplt", SETLT, null_frag, int_aarch64_sve_cmpgt>; 801 defm CMPLE_PPzZI : sve_int_scmp_vi<0b011, "cmple", SETLE, null_frag, int_aarch64_sve_cmpge>; 802 defm CMPEQ_PPzZI : sve_int_scmp_vi<0b100, "cmpeq", SETEQ, int_aarch64_sve_cmpeq>; 803 defm CMPNE_PPzZI : sve_int_scmp_vi<0b101, "cmpne", SETNE, int_aarch64_sve_cmpne>; 804 defm CMPHS_PPzZI : sve_int_ucmp_vi<0b00, "cmphs", SETUGE, int_aarch64_sve_cmphs>; 805 defm CMPHI_PPzZI : sve_int_ucmp_vi<0b01, "cmphi", SETUGT, int_aarch64_sve_cmphi>; 806 defm CMPLO_PPzZI : sve_int_ucmp_vi<0b10, "cmplo", SETULT, null_frag, int_aarch64_sve_cmphi>; 807 defm CMPLS_PPzZI : sve_int_ucmp_vi<0b11, "cmpls", SETULE, null_frag, int_aarch64_sve_cmphs>; 808 809 defm FCMGE_PPzZZ : sve_fp_3op_p_pd<0b000, "fcmge", int_aarch64_sve_fcmpge>; 810 defm FCMGT_PPzZZ : sve_fp_3op_p_pd<0b001, "fcmgt", int_aarch64_sve_fcmpgt>; 811 defm FCMEQ_PPzZZ : sve_fp_3op_p_pd<0b010, "fcmeq", int_aarch64_sve_fcmpeq>; 812 defm FCMNE_PPzZZ : sve_fp_3op_p_pd<0b011, "fcmne", int_aarch64_sve_fcmpne>; 813 defm FCMUO_PPzZZ : sve_fp_3op_p_pd<0b100, "fcmuo", int_aarch64_sve_fcmpuo>; 814 defm FACGE_PPzZZ : sve_fp_3op_p_pd<0b101, "facge", int_aarch64_sve_facge>; 815 defm FACGT_PPzZZ : sve_fp_3op_p_pd<0b111, "facgt", int_aarch64_sve_facgt>; 816 817 defm FCMGE_PPzZ0 : sve_fp_2op_p_pd<0b000, "fcmge">; 818 defm FCMGT_PPzZ0 : sve_fp_2op_p_pd<0b001, "fcmgt">; 819 defm FCMLT_PPzZ0 : sve_fp_2op_p_pd<0b010, "fcmlt">; 820 defm FCMLE_PPzZ0 : sve_fp_2op_p_pd<0b011, "fcmle">; 821 defm FCMEQ_PPzZ0 : sve_fp_2op_p_pd<0b100, "fcmeq">; 822 defm FCMNE_PPzZ0 : sve_fp_2op_p_pd<0b110, "fcmne">; 823 824 defm WHILELT_PWW : sve_int_while4_rr<0b010, "whilelt", int_aarch64_sve_whilelt>; 825 defm WHILELE_PWW : sve_int_while4_rr<0b011, "whilele", int_aarch64_sve_whilele>; 826 defm WHILELO_PWW : sve_int_while4_rr<0b110, "whilelo", int_aarch64_sve_whilelo>; 827 defm WHILELS_PWW : sve_int_while4_rr<0b111, "whilels", int_aarch64_sve_whilels>; 828 829 defm WHILELT_PXX : sve_int_while8_rr<0b010, "whilelt", int_aarch64_sve_whilelt>; 830 defm WHILELE_PXX : sve_int_while8_rr<0b011, "whilele", int_aarch64_sve_whilele>; 831 defm WHILELO_PXX : sve_int_while8_rr<0b110, "whilelo", int_aarch64_sve_whilelo>; 832 defm WHILELS_PXX : sve_int_while8_rr<0b111, "whilels", int_aarch64_sve_whilels>; 833 834 def CTERMEQ_WW : sve_int_cterm<0b0, 0b0, "ctermeq", GPR32>; 835 def CTERMNE_WW : sve_int_cterm<0b0, 0b1, "ctermne", GPR32>; 836 def CTERMEQ_XX : sve_int_cterm<0b1, 0b0, "ctermeq", GPR64>; 837 def CTERMNE_XX : sve_int_cterm<0b1, 0b1, "ctermne", GPR64>; 838 839 def RDVLI_XI : sve_int_read_vl_a<0b0, 0b11111, "rdvl">; 840 def ADDVL_XXI : sve_int_arith_vl<0b0, "addvl">; 841 def ADDPL_XXI : sve_int_arith_vl<0b1, "addpl">; 842 843 defm CNTB_XPiI : sve_int_count<0b000, "cntb", int_aarch64_sve_cntb>; 844 defm CNTH_XPiI : sve_int_count<0b010, "cnth", int_aarch64_sve_cnth>; 845 defm CNTW_XPiI : sve_int_count<0b100, "cntw", int_aarch64_sve_cntw>; 846 defm CNTD_XPiI : sve_int_count<0b110, "cntd", int_aarch64_sve_cntd>; 847 defm CNTP_XPP : sve_int_pcount_pred<0b0000, "cntp", int_aarch64_sve_cntp>; 848 849 defm INCB_XPiI : sve_int_pred_pattern_a<0b000, "incb">; 850 defm DECB_XPiI : sve_int_pred_pattern_a<0b001, "decb">; 851 defm INCH_XPiI : sve_int_pred_pattern_a<0b010, "inch">; 852 defm DECH_XPiI : sve_int_pred_pattern_a<0b011, "dech">; 853 defm INCW_XPiI : sve_int_pred_pattern_a<0b100, "incw">; 854 defm DECW_XPiI : sve_int_pred_pattern_a<0b101, "decw">; 855 defm INCD_XPiI : sve_int_pred_pattern_a<0b110, "incd">; 856 defm DECD_XPiI : sve_int_pred_pattern_a<0b111, "decd">; 857 858 defm SQINCB_XPiWdI : sve_int_pred_pattern_b_s32<0b00000, "sqincb", int_aarch64_sve_sqincb_n32>; 859 defm UQINCB_WPiI : sve_int_pred_pattern_b_u32<0b00001, "uqincb", int_aarch64_sve_uqincb_n32>; 860 defm SQDECB_XPiWdI : sve_int_pred_pattern_b_s32<0b00010, "sqdecb", int_aarch64_sve_sqdecb_n32>; 861 defm UQDECB_WPiI : sve_int_pred_pattern_b_u32<0b00011, "uqdecb", int_aarch64_sve_uqdecb_n32>; 862 defm SQINCB_XPiI : sve_int_pred_pattern_b_x64<0b00100, "sqincb", int_aarch64_sve_sqincb_n64>; 863 defm UQINCB_XPiI : sve_int_pred_pattern_b_x64<0b00101, "uqincb", int_aarch64_sve_uqincb_n64>; 864 defm SQDECB_XPiI : sve_int_pred_pattern_b_x64<0b00110, "sqdecb", int_aarch64_sve_sqdecb_n64>; 865 defm UQDECB_XPiI : sve_int_pred_pattern_b_x64<0b00111, "uqdecb", int_aarch64_sve_uqdecb_n64>; 866 867 defm SQINCH_XPiWdI : sve_int_pred_pattern_b_s32<0b01000, "sqinch", int_aarch64_sve_sqinch_n32>; 868 defm UQINCH_WPiI : sve_int_pred_pattern_b_u32<0b01001, "uqinch", int_aarch64_sve_uqinch_n32>; 869 defm SQDECH_XPiWdI : sve_int_pred_pattern_b_s32<0b01010, "sqdech", int_aarch64_sve_sqdech_n32>; 870 defm UQDECH_WPiI : sve_int_pred_pattern_b_u32<0b01011, "uqdech", int_aarch64_sve_uqdech_n32>; 871 defm SQINCH_XPiI : sve_int_pred_pattern_b_x64<0b01100, "sqinch", int_aarch64_sve_sqinch_n64>; 872 defm UQINCH_XPiI : sve_int_pred_pattern_b_x64<0b01101, "uqinch", int_aarch64_sve_uqinch_n64>; 873 defm SQDECH_XPiI : sve_int_pred_pattern_b_x64<0b01110, "sqdech", int_aarch64_sve_sqdech_n64>; 874 defm UQDECH_XPiI : sve_int_pred_pattern_b_x64<0b01111, "uqdech", int_aarch64_sve_uqdech_n64>; 875 876 defm SQINCW_XPiWdI : sve_int_pred_pattern_b_s32<0b10000, "sqincw", int_aarch64_sve_sqincw_n32>; 877 defm UQINCW_WPiI : sve_int_pred_pattern_b_u32<0b10001, "uqincw", int_aarch64_sve_uqincw_n32>; 878 defm SQDECW_XPiWdI : sve_int_pred_pattern_b_s32<0b10010, "sqdecw", int_aarch64_sve_sqdecw_n32>; 879 defm UQDECW_WPiI : sve_int_pred_pattern_b_u32<0b10011, "uqdecw", int_aarch64_sve_uqdecw_n32>; 880 defm SQINCW_XPiI : sve_int_pred_pattern_b_x64<0b10100, "sqincw", int_aarch64_sve_sqincw_n64>; 881 defm UQINCW_XPiI : sve_int_pred_pattern_b_x64<0b10101, "uqincw", int_aarch64_sve_uqincw_n64>; 882 defm SQDECW_XPiI : sve_int_pred_pattern_b_x64<0b10110, "sqdecw", int_aarch64_sve_sqdecw_n64>; 883 defm UQDECW_XPiI : sve_int_pred_pattern_b_x64<0b10111, "uqdecw", int_aarch64_sve_uqdecw_n64>; 884 885 defm SQINCD_XPiWdI : sve_int_pred_pattern_b_s32<0b11000, "sqincd", int_aarch64_sve_sqincd_n32>; 886 defm UQINCD_WPiI : sve_int_pred_pattern_b_u32<0b11001, "uqincd", int_aarch64_sve_uqincd_n32>; 887 defm SQDECD_XPiWdI : sve_int_pred_pattern_b_s32<0b11010, "sqdecd", int_aarch64_sve_sqdecd_n32>; 888 defm UQDECD_WPiI : sve_int_pred_pattern_b_u32<0b11011, "uqdecd", int_aarch64_sve_uqdecd_n32>; 889 defm SQINCD_XPiI : sve_int_pred_pattern_b_x64<0b11100, "sqincd", int_aarch64_sve_sqincd_n64>; 890 defm UQINCD_XPiI : sve_int_pred_pattern_b_x64<0b11101, "uqincd", int_aarch64_sve_uqincd_n64>; 891 defm SQDECD_XPiI : sve_int_pred_pattern_b_x64<0b11110, "sqdecd", int_aarch64_sve_sqdecd_n64>; 892 defm UQDECD_XPiI : sve_int_pred_pattern_b_x64<0b11111, "uqdecd", int_aarch64_sve_uqdecd_n64>; 893 894 defm SQINCH_ZPiI : sve_int_countvlv<0b01000, "sqinch", ZPR16, int_aarch64_sve_sqinch, nxv8i16>; 895 defm UQINCH_ZPiI : sve_int_countvlv<0b01001, "uqinch", ZPR16, int_aarch64_sve_uqinch, nxv8i16>; 896 defm SQDECH_ZPiI : sve_int_countvlv<0b01010, "sqdech", ZPR16, int_aarch64_sve_sqdech, nxv8i16>; 897 defm UQDECH_ZPiI : sve_int_countvlv<0b01011, "uqdech", ZPR16, int_aarch64_sve_uqdech, nxv8i16>; 898 defm INCH_ZPiI : sve_int_countvlv<0b01100, "inch", ZPR16>; 899 defm DECH_ZPiI : sve_int_countvlv<0b01101, "dech", ZPR16>; 900 defm SQINCW_ZPiI : sve_int_countvlv<0b10000, "sqincw", ZPR32, int_aarch64_sve_sqincw, nxv4i32>; 901 defm UQINCW_ZPiI : sve_int_countvlv<0b10001, "uqincw", ZPR32, int_aarch64_sve_uqincw, nxv4i32>; 902 defm SQDECW_ZPiI : sve_int_countvlv<0b10010, "sqdecw", ZPR32, int_aarch64_sve_sqdecw, nxv4i32>; 903 defm UQDECW_ZPiI : sve_int_countvlv<0b10011, "uqdecw", ZPR32, int_aarch64_sve_uqdecw, nxv4i32>; 904 defm INCW_ZPiI : sve_int_countvlv<0b10100, "incw", ZPR32>; 905 defm DECW_ZPiI : sve_int_countvlv<0b10101, "decw", ZPR32>; 906 defm SQINCD_ZPiI : sve_int_countvlv<0b11000, "sqincd", ZPR64, int_aarch64_sve_sqincd, nxv2i64>; 907 defm UQINCD_ZPiI : sve_int_countvlv<0b11001, "uqincd", ZPR64, int_aarch64_sve_uqincd, nxv2i64>; 908 defm SQDECD_ZPiI : sve_int_countvlv<0b11010, "sqdecd", ZPR64, int_aarch64_sve_sqdecd, nxv2i64>; 909 defm UQDECD_ZPiI : sve_int_countvlv<0b11011, "uqdecd", ZPR64, int_aarch64_sve_uqdecd, nxv2i64>; 910 defm INCD_ZPiI : sve_int_countvlv<0b11100, "incd", ZPR64>; 911 defm DECD_ZPiI : sve_int_countvlv<0b11101, "decd", ZPR64>; 912 913 defm SQINCP_XPWd : sve_int_count_r_s32<0b00000, "sqincp", int_aarch64_sve_sqincp_n32>; 914 defm SQINCP_XP : sve_int_count_r_x64<0b00010, "sqincp", int_aarch64_sve_sqincp_n64>; 915 defm UQINCP_WP : sve_int_count_r_u32<0b00100, "uqincp", int_aarch64_sve_uqincp_n32>; 916 defm UQINCP_XP : sve_int_count_r_x64<0b00110, "uqincp", int_aarch64_sve_uqincp_n64>; 917 defm SQDECP_XPWd : sve_int_count_r_s32<0b01000, "sqdecp", int_aarch64_sve_sqdecp_n32>; 918 defm SQDECP_XP : sve_int_count_r_x64<0b01010, "sqdecp", int_aarch64_sve_sqdecp_n64>; 919 defm UQDECP_WP : sve_int_count_r_u32<0b01100, "uqdecp", int_aarch64_sve_uqdecp_n32>; 920 defm UQDECP_XP : sve_int_count_r_x64<0b01110, "uqdecp", int_aarch64_sve_uqdecp_n64>; 921 defm INCP_XP : sve_int_count_r_x64<0b10000, "incp">; 922 defm DECP_XP : sve_int_count_r_x64<0b10100, "decp">; 923 924 defm SQINCP_ZP : sve_int_count_v<0b00000, "sqincp", int_aarch64_sve_sqincp>; 925 defm UQINCP_ZP : sve_int_count_v<0b00100, "uqincp", int_aarch64_sve_uqincp>; 926 defm SQDECP_ZP : sve_int_count_v<0b01000, "sqdecp", int_aarch64_sve_sqdecp>; 927 defm UQDECP_ZP : sve_int_count_v<0b01100, "uqdecp", int_aarch64_sve_uqdecp>; 928 defm INCP_ZP : sve_int_count_v<0b10000, "incp">; 929 defm DECP_ZP : sve_int_count_v<0b10100, "decp">; 930 931 defm INDEX_RR : sve_int_index_rr<"index">; 932 defm INDEX_IR : sve_int_index_ir<"index">; 933 defm INDEX_RI : sve_int_index_ri<"index">; 934 defm INDEX_II : sve_int_index_ii<"index">; 935 936 // Unpredicated shifts 937 defm ASR_ZZI : sve_int_bin_cons_shift_imm_right<0b00, "asr">; 938 defm LSR_ZZI : sve_int_bin_cons_shift_imm_right<0b01, "lsr">; 939 defm LSL_ZZI : sve_int_bin_cons_shift_imm_left< 0b11, "lsl">; 940 941 defm ASR_WIDE_ZZZ : sve_int_bin_cons_shift_wide<0b00, "asr">; 942 defm LSR_WIDE_ZZZ : sve_int_bin_cons_shift_wide<0b01, "lsr">; 943 defm LSL_WIDE_ZZZ : sve_int_bin_cons_shift_wide<0b11, "lsl">; 944 945 // Predicated shifts 946 defm ASR_ZPmI : sve_int_bin_pred_shift_imm_right<0b0000, "asr">; 947 defm LSR_ZPmI : sve_int_bin_pred_shift_imm_right<0b0001, "lsr">; 948 defm LSL_ZPmI : sve_int_bin_pred_shift_imm_left< 0b0011, "lsl">; 949 defm ASRD_ZPmI : sve_int_bin_pred_shift_imm_right<0b0100, "asrd", int_aarch64_sve_asrd>; 950 951 defm ASR_ZPmZ : sve_int_bin_pred_shift<0b000, "asr", int_aarch64_sve_asr>; 952 defm LSR_ZPmZ : sve_int_bin_pred_shift<0b001, "lsr", int_aarch64_sve_lsr>; 953 defm LSL_ZPmZ : sve_int_bin_pred_shift<0b011, "lsl", int_aarch64_sve_lsl>; 954 defm ASRR_ZPmZ : sve_int_bin_pred_shift<0b100, "asrr", null_frag>; 955 defm LSRR_ZPmZ : sve_int_bin_pred_shift<0b101, "lsrr", null_frag>; 956 defm LSLR_ZPmZ : sve_int_bin_pred_shift<0b111, "lslr", null_frag>; 957 958 defm ASR_WIDE_ZPmZ : sve_int_bin_pred_shift_wide<0b000, "asr", int_aarch64_sve_asr_wide>; 959 defm LSR_WIDE_ZPmZ : sve_int_bin_pred_shift_wide<0b001, "lsr", int_aarch64_sve_lsr_wide>; 960 defm LSL_WIDE_ZPmZ : sve_int_bin_pred_shift_wide<0b011, "lsl", int_aarch64_sve_lsl_wide>; 961 962 defm FCVT_ZPmZ_StoH : sve_fp_2op_p_zd<0b1001000, "fcvt", ZPR32, ZPR16, int_aarch64_sve_fcvt_f16f32, nxv8f16, nxv16i1, nxv4f32, ElementSizeS>; 963 defm FCVT_ZPmZ_HtoS : sve_fp_2op_p_zd<0b1001001, "fcvt", ZPR16, ZPR32, int_aarch64_sve_fcvt_f32f16, nxv4f32, nxv16i1, nxv8f16, ElementSizeS>; 964 defm SCVTF_ZPmZ_HtoH : sve_fp_2op_p_zd<0b0110010, "scvtf", ZPR16, ZPR16, int_aarch64_sve_scvtf, nxv8f16, nxv8i1, nxv8i16, ElementSizeH>; 965 defm SCVTF_ZPmZ_StoS : sve_fp_2op_p_zd<0b1010100, "scvtf", ZPR32, ZPR32, int_aarch64_sve_scvtf, nxv4f32, nxv4i1, nxv4i32, ElementSizeS>; 966 defm UCVTF_ZPmZ_StoS : sve_fp_2op_p_zd<0b1010101, "ucvtf", ZPR32, ZPR32, int_aarch64_sve_ucvtf, nxv4f32, nxv4i1, nxv4i32, ElementSizeS>; 967 defm UCVTF_ZPmZ_HtoH : sve_fp_2op_p_zd<0b0110011, "ucvtf", ZPR16, ZPR16, int_aarch64_sve_ucvtf, nxv8f16, nxv8i1, nxv8i16, ElementSizeH>; 968 defm FCVTZS_ZPmZ_HtoH : sve_fp_2op_p_zd<0b0111010, "fcvtzs", ZPR16, ZPR16, int_aarch64_sve_fcvtzs, nxv8i16, nxv8i1, nxv8f16, ElementSizeH>; 969 defm FCVTZS_ZPmZ_StoS : sve_fp_2op_p_zd<0b1011100, "fcvtzs", ZPR32, ZPR32, int_aarch64_sve_fcvtzs, nxv4i32, nxv4i1, nxv4f32, ElementSizeS>; 970 defm FCVTZU_ZPmZ_HtoH : sve_fp_2op_p_zd<0b0111011, "fcvtzu", ZPR16, ZPR16, int_aarch64_sve_fcvtzu, nxv8i16, nxv8i1, nxv8f16, ElementSizeH>; 971 defm FCVTZU_ZPmZ_StoS : sve_fp_2op_p_zd<0b1011101, "fcvtzu", ZPR32, ZPR32, int_aarch64_sve_fcvtzu, nxv4i32, nxv4i1, nxv4f32, ElementSizeS>; 972 defm FCVT_ZPmZ_DtoH : sve_fp_2op_p_zd<0b1101000, "fcvt", ZPR64, ZPR16, int_aarch64_sve_fcvt_f16f64, nxv8f16, nxv16i1, nxv2f64, ElementSizeD>; 973 defm FCVT_ZPmZ_HtoD : sve_fp_2op_p_zd<0b1101001, "fcvt", ZPR16, ZPR64, int_aarch64_sve_fcvt_f64f16, nxv2f64, nxv16i1, nxv8f16, ElementSizeD>; 974 defm FCVT_ZPmZ_DtoS : sve_fp_2op_p_zd<0b1101010, "fcvt", ZPR64, ZPR32, int_aarch64_sve_fcvt_f32f64, nxv4f32, nxv16i1, nxv2f64, ElementSizeD>; 975 defm FCVT_ZPmZ_StoD : sve_fp_2op_p_zd<0b1101011, "fcvt", ZPR32, ZPR64, int_aarch64_sve_fcvt_f64f32, nxv2f64, nxv16i1, nxv4f32, ElementSizeD>; 976 defm SCVTF_ZPmZ_StoD : sve_fp_2op_p_zd<0b1110000, "scvtf", ZPR32, ZPR64, int_aarch64_sve_scvtf_f64i32, nxv2f64, nxv16i1, nxv4i32, ElementSizeD>; 977 defm UCVTF_ZPmZ_StoD : sve_fp_2op_p_zd<0b1110001, "ucvtf", ZPR32, ZPR64, int_aarch64_sve_ucvtf_f64i32, nxv2f64, nxv16i1, nxv4i32, ElementSizeD>; 978 defm UCVTF_ZPmZ_StoH : sve_fp_2op_p_zd<0b0110101, "ucvtf", ZPR32, ZPR16, int_aarch64_sve_ucvtf_f16i32, nxv8f16, nxv16i1, nxv4i32, ElementSizeS>; 979 defm SCVTF_ZPmZ_DtoS : sve_fp_2op_p_zd<0b1110100, "scvtf", ZPR64, ZPR32, int_aarch64_sve_scvtf_f32i64, nxv4f32, nxv16i1, nxv2i64, ElementSizeD>; 980 defm SCVTF_ZPmZ_StoH : sve_fp_2op_p_zd<0b0110100, "scvtf", ZPR32, ZPR16, int_aarch64_sve_scvtf_f16i32, nxv8f16, nxv16i1, nxv4i32, ElementSizeS>; 981 defm SCVTF_ZPmZ_DtoH : sve_fp_2op_p_zd<0b0110110, "scvtf", ZPR64, ZPR16, int_aarch64_sve_scvtf_f16i64, nxv8f16, nxv16i1, nxv2i64, ElementSizeD>; 982 defm UCVTF_ZPmZ_DtoS : sve_fp_2op_p_zd<0b1110101, "ucvtf", ZPR64, ZPR32, int_aarch64_sve_ucvtf_f32i64, nxv4f32, nxv16i1, nxv2i64, ElementSizeD>; 983 defm UCVTF_ZPmZ_DtoH : sve_fp_2op_p_zd<0b0110111, "ucvtf", ZPR64, ZPR16, int_aarch64_sve_ucvtf_f16i64, nxv8f16, nxv16i1, nxv2i64, ElementSizeD>; 984 defm SCVTF_ZPmZ_DtoD : sve_fp_2op_p_zd<0b1110110, "scvtf", ZPR64, ZPR64, int_aarch64_sve_scvtf, nxv2f64, nxv2i1, nxv2i64, ElementSizeD>; 985 defm UCVTF_ZPmZ_DtoD : sve_fp_2op_p_zd<0b1110111, "ucvtf", ZPR64, ZPR64, int_aarch64_sve_ucvtf, nxv2f64, nxv2i1, nxv2i64, ElementSizeD>; 986 defm FCVTZS_ZPmZ_DtoS : sve_fp_2op_p_zd<0b1111000, "fcvtzs", ZPR64, ZPR32, int_aarch64_sve_fcvtzs_i32f64, nxv4i32, nxv16i1, nxv2f64, ElementSizeD>; 987 defm FCVTZU_ZPmZ_DtoS : sve_fp_2op_p_zd<0b1111001, "fcvtzu", ZPR64, ZPR32, int_aarch64_sve_fcvtzu_i32f64, nxv4i32, nxv16i1, nxv2f64, ElementSizeD>; 988 defm FCVTZS_ZPmZ_StoD : sve_fp_2op_p_zd<0b1111100, "fcvtzs", ZPR32, ZPR64, int_aarch64_sve_fcvtzs_i64f32, nxv2i64, nxv16i1, nxv4f32, ElementSizeD>; 989 defm FCVTZS_ZPmZ_HtoS : sve_fp_2op_p_zd<0b0111100, "fcvtzs", ZPR16, ZPR32, int_aarch64_sve_fcvtzs_i32f16, nxv4i32, nxv16i1, nxv8f16, ElementSizeS>; 990 defm FCVTZS_ZPmZ_HtoD : sve_fp_2op_p_zd<0b0111110, "fcvtzs", ZPR16, ZPR64, int_aarch64_sve_fcvtzs_i64f16, nxv2i64, nxv16i1, nxv8f16, ElementSizeD>; 991 defm FCVTZU_ZPmZ_HtoS : sve_fp_2op_p_zd<0b0111101, "fcvtzu", ZPR16, ZPR32, int_aarch64_sve_fcvtzu_i32f16, nxv4i32, nxv16i1, nxv8f16, ElementSizeS>; 992 defm FCVTZU_ZPmZ_HtoD : sve_fp_2op_p_zd<0b0111111, "fcvtzu", ZPR16, ZPR64, int_aarch64_sve_fcvtzu_i64f16, nxv2i64, nxv16i1, nxv8f16, ElementSizeD>; 993 defm FCVTZU_ZPmZ_StoD : sve_fp_2op_p_zd<0b1111101, "fcvtzu", ZPR32, ZPR64, int_aarch64_sve_fcvtzu_i64f32, nxv2i64, nxv16i1, nxv4f32, ElementSizeD>; 994 defm FCVTZS_ZPmZ_DtoD : sve_fp_2op_p_zd<0b1111110, "fcvtzs", ZPR64, ZPR64, int_aarch64_sve_fcvtzs, nxv2i64, nxv2i1, nxv2f64, ElementSizeD>; 995 defm FCVTZU_ZPmZ_DtoD : sve_fp_2op_p_zd<0b1111111, "fcvtzu", ZPR64, ZPR64, int_aarch64_sve_fcvtzu, nxv2i64, nxv2i1, nxv2f64, ElementSizeD>; 996 997 defm FRINTN_ZPmZ : sve_fp_2op_p_zd_HSD<0b00000, "frintn", int_aarch64_sve_frintn>; 998 defm FRINTP_ZPmZ : sve_fp_2op_p_zd_HSD<0b00001, "frintp", int_aarch64_sve_frintp>; 999 defm FRINTM_ZPmZ : sve_fp_2op_p_zd_HSD<0b00010, "frintm", int_aarch64_sve_frintm>; 1000 defm FRINTZ_ZPmZ : sve_fp_2op_p_zd_HSD<0b00011, "frintz", int_aarch64_sve_frintz>; 1001 defm FRINTA_ZPmZ : sve_fp_2op_p_zd_HSD<0b00100, "frinta", int_aarch64_sve_frinta>; 1002 defm FRINTX_ZPmZ : sve_fp_2op_p_zd_HSD<0b00110, "frintx", int_aarch64_sve_frintx>; 1003 defm FRINTI_ZPmZ : sve_fp_2op_p_zd_HSD<0b00111, "frinti", int_aarch64_sve_frinti>; 1004 defm FRECPX_ZPmZ : sve_fp_2op_p_zd_HSD<0b01100, "frecpx", int_aarch64_sve_frecpx>; 1005 defm FSQRT_ZPmZ : sve_fp_2op_p_zd_HSD<0b01101, "fsqrt", int_aarch64_sve_fsqrt>; 1006 1007 // InstAliases 1008 def : InstAlias<"mov $Zd, $Zn", 1009 (ORR_ZZZ ZPR64:$Zd, ZPR64:$Zn, ZPR64:$Zn), 1>; 1010 def : InstAlias<"mov $Pd, $Pg/m, $Pn", 1011 (SEL_PPPP PPR8:$Pd, PPRAny:$Pg, PPR8:$Pn, PPR8:$Pd), 1>; 1012 def : InstAlias<"mov $Pd, $Pn", 1013 (ORR_PPzPP PPR8:$Pd, PPR8:$Pn, PPR8:$Pn, PPR8:$Pn), 1>; 1014 def : InstAlias<"mov $Pd, $Pg/z, $Pn", 1015 (AND_PPzPP PPR8:$Pd, PPRAny:$Pg, PPR8:$Pn, PPR8:$Pn), 1>; 1016 1017 def : InstAlias<"movs $Pd, $Pn", 1018 (ORRS_PPzPP PPR8:$Pd, PPR8:$Pn, PPR8:$Pn, PPR8:$Pn), 1>; 1019 def : InstAlias<"movs $Pd, $Pg/z, $Pn", 1020 (ANDS_PPzPP PPR8:$Pd, PPRAny:$Pg, PPR8:$Pn, PPR8:$Pn), 1>; 1021 1022 def : InstAlias<"not $Pd, $Pg/z, $Pn", 1023 (EOR_PPzPP PPR8:$Pd, PPRAny:$Pg, PPR8:$Pn, PPRAny:$Pg), 1>; 1024 1025 def : InstAlias<"nots $Pd, $Pg/z, $Pn", 1026 (EORS_PPzPP PPR8:$Pd, PPRAny:$Pg, PPR8:$Pn, PPRAny:$Pg), 1>; 1027 1028 def : InstAlias<"cmple $Zd, $Pg/z, $Zm, $Zn", 1029 (CMPGE_PPzZZ_B PPR8:$Zd, PPR3bAny:$Pg, ZPR8:$Zn, ZPR8:$Zm), 0>; 1030 def : InstAlias<"cmple $Zd, $Pg/z, $Zm, $Zn", 1031 (CMPGE_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1032 def : InstAlias<"cmple $Zd, $Pg/z, $Zm, $Zn", 1033 (CMPGE_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1034 def : InstAlias<"cmple $Zd, $Pg/z, $Zm, $Zn", 1035 (CMPGE_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1036 1037 def : InstAlias<"cmplo $Zd, $Pg/z, $Zm, $Zn", 1038 (CMPHI_PPzZZ_B PPR8:$Zd, PPR3bAny:$Pg, ZPR8:$Zn, ZPR8:$Zm), 0>; 1039 def : InstAlias<"cmplo $Zd, $Pg/z, $Zm, $Zn", 1040 (CMPHI_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1041 def : InstAlias<"cmplo $Zd, $Pg/z, $Zm, $Zn", 1042 (CMPHI_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1043 def : InstAlias<"cmplo $Zd, $Pg/z, $Zm, $Zn", 1044 (CMPHI_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1045 1046 def : InstAlias<"cmpls $Zd, $Pg/z, $Zm, $Zn", 1047 (CMPHS_PPzZZ_B PPR8:$Zd, PPR3bAny:$Pg, ZPR8:$Zn, ZPR8:$Zm), 0>; 1048 def : InstAlias<"cmpls $Zd, $Pg/z, $Zm, $Zn", 1049 (CMPHS_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1050 def : InstAlias<"cmpls $Zd, $Pg/z, $Zm, $Zn", 1051 (CMPHS_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1052 def : InstAlias<"cmpls $Zd, $Pg/z, $Zm, $Zn", 1053 (CMPHS_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1054 1055 def : InstAlias<"cmplt $Zd, $Pg/z, $Zm, $Zn", 1056 (CMPGT_PPzZZ_B PPR8:$Zd, PPR3bAny:$Pg, ZPR8:$Zn, ZPR8:$Zm), 0>; 1057 def : InstAlias<"cmplt $Zd, $Pg/z, $Zm, $Zn", 1058 (CMPGT_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1059 def : InstAlias<"cmplt $Zd, $Pg/z, $Zm, $Zn", 1060 (CMPGT_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1061 def : InstAlias<"cmplt $Zd, $Pg/z, $Zm, $Zn", 1062 (CMPGT_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1063 1064 def : InstAlias<"facle $Zd, $Pg/z, $Zm, $Zn", 1065 (FACGE_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1066 def : InstAlias<"facle $Zd, $Pg/z, $Zm, $Zn", 1067 (FACGE_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1068 def : InstAlias<"facle $Zd, $Pg/z, $Zm, $Zn", 1069 (FACGE_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1070 1071 def : InstAlias<"faclt $Zd, $Pg/z, $Zm, $Zn", 1072 (FACGT_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1073 def : InstAlias<"faclt $Zd, $Pg/z, $Zm, $Zn", 1074 (FACGT_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1075 def : InstAlias<"faclt $Zd, $Pg/z, $Zm, $Zn", 1076 (FACGT_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1077 1078 def : InstAlias<"fcmle $Zd, $Pg/z, $Zm, $Zn", 1079 (FCMGE_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1080 def : InstAlias<"fcmle $Zd, $Pg/z, $Zm, $Zn", 1081 (FCMGE_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1082 def : InstAlias<"fcmle $Zd, $Pg/z, $Zm, $Zn", 1083 (FCMGE_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1084 1085 def : InstAlias<"fcmlt $Zd, $Pg/z, $Zm, $Zn", 1086 (FCMGT_PPzZZ_H PPR16:$Zd, PPR3bAny:$Pg, ZPR16:$Zn, ZPR16:$Zm), 0>; 1087 def : InstAlias<"fcmlt $Zd, $Pg/z, $Zm, $Zn", 1088 (FCMGT_PPzZZ_S PPR32:$Zd, PPR3bAny:$Pg, ZPR32:$Zn, ZPR32:$Zm), 0>; 1089 def : InstAlias<"fcmlt $Zd, $Pg/z, $Zm, $Zn", 1090 (FCMGT_PPzZZ_D PPR64:$Zd, PPR3bAny:$Pg, ZPR64:$Zn, ZPR64:$Zm), 0>; 1091 1092 def : Pat<(AArch64ptest (nxv16i1 PPR:$pg), (nxv16i1 PPR:$src)), 1093 (PTEST_PP PPR:$pg, PPR:$src)>; 1094 def : Pat<(AArch64ptest (nxv8i1 PPR:$pg), (nxv8i1 PPR:$src)), 1095 (PTEST_PP PPR:$pg, PPR:$src)>; 1096 def : Pat<(AArch64ptest (nxv4i1 PPR:$pg), (nxv4i1 PPR:$src)), 1097 (PTEST_PP PPR:$pg, PPR:$src)>; 1098 def : Pat<(AArch64ptest (nxv2i1 PPR:$pg), (nxv2i1 PPR:$src)), 1099 (PTEST_PP PPR:$pg, PPR:$src)>; 1100 1101 def : Pat<(sext_inreg (nxv2i64 ZPR:$Zs), nxv2i32), (SXTW_ZPmZ_D (IMPLICIT_DEF), (PTRUE_D 31), ZPR:$Zs)>; 1102 def : Pat<(sext_inreg (nxv2i64 ZPR:$Zs), nxv2i16), (SXTH_ZPmZ_D (IMPLICIT_DEF), (PTRUE_D 31), ZPR:$Zs)>; 1103 def : Pat<(sext_inreg (nxv2i64 ZPR:$Zs), nxv2i8), (SXTB_ZPmZ_D (IMPLICIT_DEF), (PTRUE_D 31), ZPR:$Zs)>; 1104 def : Pat<(sext_inreg (nxv4i32 ZPR:$Zs), nxv4i16), (SXTH_ZPmZ_S (IMPLICIT_DEF), (PTRUE_S 31), ZPR:$Zs)>; 1105 def : Pat<(sext_inreg (nxv4i32 ZPR:$Zs), nxv4i8), (SXTB_ZPmZ_S (IMPLICIT_DEF), (PTRUE_S 31), ZPR:$Zs)>; 1106 def : Pat<(sext_inreg (nxv8i16 ZPR:$Zs), nxv8i8), (SXTB_ZPmZ_H (IMPLICIT_DEF), (PTRUE_H 31), ZPR:$Zs)>; 1107 1108 def : Pat<(nxv16i8 (bitconvert (nxv8i16 ZPR:$src))), (nxv16i8 ZPR:$src)>; 1109 def : Pat<(nxv16i8 (bitconvert (nxv4i32 ZPR:$src))), (nxv16i8 ZPR:$src)>; 1110 def : Pat<(nxv16i8 (bitconvert (nxv2i64 ZPR:$src))), (nxv16i8 ZPR:$src)>; 1111 def : Pat<(nxv16i8 (bitconvert (nxv8f16 ZPR:$src))), (nxv16i8 ZPR:$src)>; 1112 def : Pat<(nxv16i8 (bitconvert (nxv4f32 ZPR:$src))), (nxv16i8 ZPR:$src)>; 1113 def : Pat<(nxv16i8 (bitconvert (nxv2f64 ZPR:$src))), (nxv16i8 ZPR:$src)>; 1114 1115 def : Pat<(nxv8i16 (bitconvert (nxv16i8 ZPR:$src))), (nxv8i16 ZPR:$src)>; 1116 def : Pat<(nxv8i16 (bitconvert (nxv4i32 ZPR:$src))), (nxv8i16 ZPR:$src)>; 1117 def : Pat<(nxv8i16 (bitconvert (nxv2i64 ZPR:$src))), (nxv8i16 ZPR:$src)>; 1118 def : Pat<(nxv8i16 (bitconvert (nxv8f16 ZPR:$src))), (nxv8i16 ZPR:$src)>; 1119 def : Pat<(nxv8i16 (bitconvert (nxv4f32 ZPR:$src))), (nxv8i16 ZPR:$src)>; 1120 def : Pat<(nxv8i16 (bitconvert (nxv2f64 ZPR:$src))), (nxv8i16 ZPR:$src)>; 1121 1122 def : Pat<(nxv4i32 (bitconvert (nxv16i8 ZPR:$src))), (nxv4i32 ZPR:$src)>; 1123 def : Pat<(nxv4i32 (bitconvert (nxv8i16 ZPR:$src))), (nxv4i32 ZPR:$src)>; 1124 def : Pat<(nxv4i32 (bitconvert (nxv2i64 ZPR:$src))), (nxv4i32 ZPR:$src)>; 1125 def : Pat<(nxv4i32 (bitconvert (nxv8f16 ZPR:$src))), (nxv4i32 ZPR:$src)>; 1126 def : Pat<(nxv4i32 (bitconvert (nxv4f32 ZPR:$src))), (nxv4i32 ZPR:$src)>; 1127 def : Pat<(nxv4i32 (bitconvert (nxv2f64 ZPR:$src))), (nxv4i32 ZPR:$src)>; 1128 1129 def : Pat<(nxv2i64 (bitconvert (nxv16i8 ZPR:$src))), (nxv2i64 ZPR:$src)>; 1130 def : Pat<(nxv2i64 (bitconvert (nxv8i16 ZPR:$src))), (nxv2i64 ZPR:$src)>; 1131 def : Pat<(nxv2i64 (bitconvert (nxv4i32 ZPR:$src))), (nxv2i64 ZPR:$src)>; 1132 def : Pat<(nxv2i64 (bitconvert (nxv8f16 ZPR:$src))), (nxv2i64 ZPR:$src)>; 1133 def : Pat<(nxv2i64 (bitconvert (nxv4f32 ZPR:$src))), (nxv2i64 ZPR:$src)>; 1134 def : Pat<(nxv2i64 (bitconvert (nxv2f64 ZPR:$src))), (nxv2i64 ZPR:$src)>; 1135 1136 def : Pat<(nxv8f16 (bitconvert (nxv16i8 ZPR:$src))), (nxv8f16 ZPR:$src)>; 1137 def : Pat<(nxv8f16 (bitconvert (nxv8i16 ZPR:$src))), (nxv8f16 ZPR:$src)>; 1138 def : Pat<(nxv8f16 (bitconvert (nxv4i32 ZPR:$src))), (nxv8f16 ZPR:$src)>; 1139 def : Pat<(nxv8f16 (bitconvert (nxv2i64 ZPR:$src))), (nxv8f16 ZPR:$src)>; 1140 def : Pat<(nxv8f16 (bitconvert (nxv4f32 ZPR:$src))), (nxv8f16 ZPR:$src)>; 1141 def : Pat<(nxv8f16 (bitconvert (nxv2f64 ZPR:$src))), (nxv8f16 ZPR:$src)>; 1142 1143 def : Pat<(nxv4f32 (bitconvert (nxv16i8 ZPR:$src))), (nxv4f32 ZPR:$src)>; 1144 def : Pat<(nxv4f32 (bitconvert (nxv8i16 ZPR:$src))), (nxv4f32 ZPR:$src)>; 1145 def : Pat<(nxv4f32 (bitconvert (nxv4i32 ZPR:$src))), (nxv4f32 ZPR:$src)>; 1146 def : Pat<(nxv4f32 (bitconvert (nxv2i64 ZPR:$src))), (nxv4f32 ZPR:$src)>; 1147 def : Pat<(nxv4f32 (bitconvert (nxv8f16 ZPR:$src))), (nxv4f32 ZPR:$src)>; 1148 def : Pat<(nxv4f32 (bitconvert (nxv2f64 ZPR:$src))), (nxv4f32 ZPR:$src)>; 1149 1150 def : Pat<(nxv2f64 (bitconvert (nxv16i8 ZPR:$src))), (nxv2f64 ZPR:$src)>; 1151 def : Pat<(nxv2f64 (bitconvert (nxv8i16 ZPR:$src))), (nxv2f64 ZPR:$src)>; 1152 def : Pat<(nxv2f64 (bitconvert (nxv4i32 ZPR:$src))), (nxv2f64 ZPR:$src)>; 1153 def : Pat<(nxv2f64 (bitconvert (nxv2i64 ZPR:$src))), (nxv2f64 ZPR:$src)>; 1154 def : Pat<(nxv2f64 (bitconvert (nxv8f16 ZPR:$src))), (nxv2f64 ZPR:$src)>; 1155 def : Pat<(nxv2f64 (bitconvert (nxv4f32 ZPR:$src))), (nxv2f64 ZPR:$src)>; 1156 1157 // Add more complex addressing modes here as required 1158 multiclass pred_load<ValueType Ty, ValueType PredTy, SDPatternOperator Load, 1159 Instruction RegImmInst> { 1160 1161 def _default_z : Pat<(Ty (Load GPR64:$base, (PredTy PPR:$gp), (SVEDup0Undef))), 1162 (RegImmInst PPR:$gp, GPR64:$base, (i64 0))>; 1163 } 1164 1165 // 2-element contiguous loads 1166 defm : pred_load<nxv2i64, nxv2i1, zext_masked_load_i8, LD1B_D_IMM>; 1167 defm : pred_load<nxv2i64, nxv2i1, asext_masked_load_i8, LD1SB_D_IMM>; 1168 defm : pred_load<nxv2i64, nxv2i1, zext_masked_load_i16, LD1H_D_IMM>; 1169 defm : pred_load<nxv2i64, nxv2i1, asext_masked_load_i16, LD1SH_D_IMM>; 1170 defm : pred_load<nxv2i64, nxv2i1, zext_masked_load_i32, LD1W_D_IMM>; 1171 defm : pred_load<nxv2i64, nxv2i1, asext_masked_load_i32, LD1SW_D_IMM>; 1172 defm : pred_load<nxv2i64, nxv2i1, nonext_masked_load, LD1D_IMM>; 1173 defm : pred_load<nxv2f16, nxv2i1, nonext_masked_load, LD1H_D_IMM>; 1174 defm : pred_load<nxv2f32, nxv2i1, nonext_masked_load, LD1W_D_IMM>; 1175 defm : pred_load<nxv2f64, nxv2i1, nonext_masked_load, LD1D_IMM>; 1176 1177 // 4-element contiguous loads 1178 defm : pred_load<nxv4i32, nxv4i1, zext_masked_load_i8, LD1B_S_IMM>; 1179 defm : pred_load<nxv4i32, nxv4i1, asext_masked_load_i8, LD1SB_S_IMM>; 1180 defm : pred_load<nxv4i32, nxv4i1, zext_masked_load_i16, LD1H_S_IMM>; 1181 defm : pred_load<nxv4i32, nxv4i1, asext_masked_load_i16, LD1SH_S_IMM>; 1182 defm : pred_load<nxv4i32, nxv4i1, nonext_masked_load, LD1W_IMM>; 1183 defm : pred_load<nxv4f16, nxv4i1, nonext_masked_load, LD1H_S_IMM>; 1184 defm : pred_load<nxv4f32, nxv4i1, nonext_masked_load, LD1W_IMM>; 1185 1186 // 8-element contiguous loads 1187 defm : pred_load<nxv8i16, nxv8i1, zext_masked_load_i8, LD1B_H_IMM>; 1188 defm : pred_load<nxv8i16, nxv8i1, asext_masked_load_i8, LD1SB_H_IMM>; 1189 defm : pred_load<nxv8i16, nxv8i1, nonext_masked_load, LD1H_IMM>; 1190 defm : pred_load<nxv8f16, nxv8i1, nonext_masked_load, LD1H_IMM>; 1191 1192 // 16-element contiguous loads 1193 defm : pred_load<nxv16i8, nxv16i1, nonext_masked_load, LD1B_IMM>; 1194 1195 multiclass pred_store<ValueType Ty, ValueType PredTy, SDPatternOperator Store, 1196 Instruction RegImmInst> { 1197 def _default : Pat<(Store (Ty ZPR:$vec), GPR64:$base, (PredTy PPR:$gp)), 1198 (RegImmInst ZPR:$vec, PPR:$gp, GPR64:$base, (i64 0))>; 1199 } 1200 1201 // 2-element contiguous stores 1202 defm : pred_store<nxv2i64, nxv2i1, trunc_masked_store_i8, ST1B_D_IMM>; 1203 defm : pred_store<nxv2i64, nxv2i1, trunc_masked_store_i16, ST1H_D_IMM>; 1204 defm : pred_store<nxv2i64, nxv2i1, trunc_masked_store_i32, ST1W_D_IMM>; 1205 defm : pred_store<nxv2i64, nxv2i1, nontrunc_masked_store, ST1D_IMM>; 1206 defm : pred_store<nxv2f16, nxv2i1, nontrunc_masked_store, ST1H_D_IMM>; 1207 defm : pred_store<nxv2f32, nxv2i1, nontrunc_masked_store, ST1W_D_IMM>; 1208 defm : pred_store<nxv2f64, nxv2i1, nontrunc_masked_store, ST1D_IMM>; 1209 1210 // 4-element contiguous stores 1211 defm : pred_store<nxv4i32, nxv4i1, trunc_masked_store_i8, ST1B_S_IMM>; 1212 defm : pred_store<nxv4i32, nxv4i1, trunc_masked_store_i16, ST1H_S_IMM>; 1213 defm : pred_store<nxv4i32, nxv4i1, nontrunc_masked_store, ST1W_IMM>; 1214 defm : pred_store<nxv4f16, nxv4i1, nontrunc_masked_store, ST1H_S_IMM>; 1215 defm : pred_store<nxv4f32, nxv4i1, nontrunc_masked_store, ST1W_IMM>; 1216 1217 // 8-element contiguous stores 1218 defm : pred_store<nxv8i16, nxv8i1, trunc_masked_store_i8, ST1B_H_IMM>; 1219 defm : pred_store<nxv8i16, nxv8i1, nontrunc_masked_store, ST1H_IMM>; 1220 defm : pred_store<nxv8f16, nxv8i1, nontrunc_masked_store, ST1H_IMM>; 1221 1222 // 16-element contiguous stores 1223 defm : pred_store<nxv16i8, nxv16i1, nontrunc_masked_store, ST1B_IMM>; 1224 1225 defm : pred_load<nxv16i8, nxv16i1, non_temporal_load, LDNT1B_ZRI>; 1226 defm : pred_load<nxv8i16, nxv8i1, non_temporal_load, LDNT1H_ZRI>; 1227 defm : pred_load<nxv4i32, nxv4i1, non_temporal_load, LDNT1W_ZRI>; 1228 defm : pred_load<nxv2i64, nxv2i1, non_temporal_load, LDNT1D_ZRI>; 1229 1230 defm : pred_store<nxv16i8, nxv16i1, non_temporal_store, STNT1B_ZRI>; 1231 defm : pred_store<nxv8i16, nxv8i1, non_temporal_store, STNT1H_ZRI>; 1232 defm : pred_store<nxv4i32, nxv4i1, non_temporal_store, STNT1W_ZRI>; 1233 defm : pred_store<nxv2i64, nxv2i1, non_temporal_store, STNT1D_ZRI>; 1234} 1235 1236let Predicates = [HasSVE2] in { 1237 // SVE2 integer multiply-add (indexed) 1238 defm MLA_ZZZI : sve2_int_mla_by_indexed_elem<0b01, 0b0, "mla">; 1239 defm MLS_ZZZI : sve2_int_mla_by_indexed_elem<0b01, 0b1, "mls">; 1240 1241 // SVE2 saturating multiply-add high (indexed) 1242 defm SQRDMLAH_ZZZI : sve2_int_mla_by_indexed_elem<0b10, 0b0, "sqrdmlah">; 1243 defm SQRDMLSH_ZZZI : sve2_int_mla_by_indexed_elem<0b10, 0b1, "sqrdmlsh">; 1244 1245 // SVE2 saturating multiply-add high (vectors, unpredicated) 1246 defm SQRDMLAH_ZZZ : sve2_int_mla<0b0, "sqrdmlah">; 1247 defm SQRDMLSH_ZZZ : sve2_int_mla<0b1, "sqrdmlsh">; 1248 1249 // SVE2 integer multiply (indexed) 1250 defm MUL_ZZZI : sve2_int_mul_by_indexed_elem<0b1110, "mul">; 1251 1252 // SVE2 saturating multiply high (indexed) 1253 defm SQDMULH_ZZZI : sve2_int_mul_by_indexed_elem<0b1100, "sqdmulh">; 1254 defm SQRDMULH_ZZZI : sve2_int_mul_by_indexed_elem<0b1101, "sqrdmulh">; 1255 1256 // SVE2 signed saturating doubling multiply high (unpredicated) 1257 defm SQDMULH_ZZZ : sve2_int_mul<0b100, "sqdmulh">; 1258 defm SQRDMULH_ZZZ : sve2_int_mul<0b101, "sqrdmulh">; 1259 1260 // SVE2 integer multiply vectors (unpredicated) 1261 defm MUL_ZZZ : sve2_int_mul<0b000, "mul">; 1262 defm SMULH_ZZZ : sve2_int_mul<0b010, "smulh">; 1263 defm UMULH_ZZZ : sve2_int_mul<0b011, "umulh">; 1264 def PMUL_ZZZ_B : sve2_int_mul<0b00, 0b001, "pmul", ZPR8>; 1265 1266 // SVE2 complex integer dot product (indexed) 1267 defm CDOT_ZZZI : sve2_cintx_dot_by_indexed_elem<"cdot">; 1268 1269 // SVE2 complex integer dot product 1270 defm CDOT_ZZZ : sve2_cintx_dot<"cdot">; 1271 1272 // SVE2 complex integer multiply-add (indexed) 1273 defm CMLA_ZZZI : sve2_cmla_by_indexed_elem<0b0, "cmla">; 1274 // SVE2 complex saturating multiply-add (indexed) 1275 defm SQRDCMLAH_ZZZI : sve2_cmla_by_indexed_elem<0b1, "sqrdcmlah">; 1276 1277 // SVE2 complex integer multiply-add 1278 defm CMLA_ZZZ : sve2_int_cmla<0b0, "cmla">; 1279 defm SQRDCMLAH_ZZZ : sve2_int_cmla<0b1, "sqrdcmlah">; 1280 1281 // SVE2 integer multiply long (indexed) 1282 defm SMULLB_ZZZI : sve2_int_mul_long_by_indexed_elem<0b000, "smullb">; 1283 defm SMULLT_ZZZI : sve2_int_mul_long_by_indexed_elem<0b001, "smullt">; 1284 defm UMULLB_ZZZI : sve2_int_mul_long_by_indexed_elem<0b010, "umullb">; 1285 defm UMULLT_ZZZI : sve2_int_mul_long_by_indexed_elem<0b011, "umullt">; 1286 1287 // SVE2 saturating multiply (indexed) 1288 defm SQDMULLB_ZZZI : sve2_int_mul_long_by_indexed_elem<0b100, "sqdmullb">; 1289 defm SQDMULLT_ZZZI : sve2_int_mul_long_by_indexed_elem<0b101, "sqdmullt">; 1290 1291 // SVE2 integer multiply-add long (indexed) 1292 defm SMLALB_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1000, "smlalb">; 1293 defm SMLALT_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1001, "smlalt">; 1294 defm UMLALB_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1010, "umlalb">; 1295 defm UMLALT_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1011, "umlalt">; 1296 defm SMLSLB_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1100, "smlslb">; 1297 defm SMLSLT_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1101, "smlslt">; 1298 defm UMLSLB_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1110, "umlslb">; 1299 defm UMLSLT_ZZZI : sve2_int_mla_long_by_indexed_elem<0b1111, "umlslt">; 1300 1301 // SVE2 integer multiply-add long (vectors, unpredicated) 1302 defm SMLALB_ZZZ : sve2_int_mla_long<0b10000, "smlalb">; 1303 defm SMLALT_ZZZ : sve2_int_mla_long<0b10001, "smlalt">; 1304 defm UMLALB_ZZZ : sve2_int_mla_long<0b10010, "umlalb">; 1305 defm UMLALT_ZZZ : sve2_int_mla_long<0b10011, "umlalt">; 1306 defm SMLSLB_ZZZ : sve2_int_mla_long<0b10100, "smlslb">; 1307 defm SMLSLT_ZZZ : sve2_int_mla_long<0b10101, "smlslt">; 1308 defm UMLSLB_ZZZ : sve2_int_mla_long<0b10110, "umlslb">; 1309 defm UMLSLT_ZZZ : sve2_int_mla_long<0b10111, "umlslt">; 1310 1311 // SVE2 saturating multiply-add long (indexed) 1312 defm SQDMLALB_ZZZI : sve2_int_mla_long_by_indexed_elem<0b0100, "sqdmlalb">; 1313 defm SQDMLALT_ZZZI : sve2_int_mla_long_by_indexed_elem<0b0101, "sqdmlalt">; 1314 defm SQDMLSLB_ZZZI : sve2_int_mla_long_by_indexed_elem<0b0110, "sqdmlslb">; 1315 defm SQDMLSLT_ZZZI : sve2_int_mla_long_by_indexed_elem<0b0111, "sqdmlslt">; 1316 1317 // SVE2 saturating multiply-add long (vectors, unpredicated) 1318 defm SQDMLALB_ZZZ : sve2_int_mla_long<0b11000, "sqdmlalb">; 1319 defm SQDMLALT_ZZZ : sve2_int_mla_long<0b11001, "sqdmlalt">; 1320 defm SQDMLSLB_ZZZ : sve2_int_mla_long<0b11010, "sqdmlslb">; 1321 defm SQDMLSLT_ZZZ : sve2_int_mla_long<0b11011, "sqdmlslt">; 1322 1323 // SVE2 saturating multiply-add interleaved long 1324 defm SQDMLALBT_ZZZ : sve2_int_mla_long<0b00010, "sqdmlalbt">; 1325 defm SQDMLSLBT_ZZZ : sve2_int_mla_long<0b00011, "sqdmlslbt">; 1326 1327 // SVE2 integer halving add/subtract (predicated) 1328 defm SHADD_ZPmZ : sve2_int_arith_pred<0b100000, "shadd">; 1329 defm UHADD_ZPmZ : sve2_int_arith_pred<0b100010, "uhadd">; 1330 defm SHSUB_ZPmZ : sve2_int_arith_pred<0b100100, "shsub">; 1331 defm UHSUB_ZPmZ : sve2_int_arith_pred<0b100110, "uhsub">; 1332 defm SRHADD_ZPmZ : sve2_int_arith_pred<0b101000, "srhadd">; 1333 defm URHADD_ZPmZ : sve2_int_arith_pred<0b101010, "urhadd">; 1334 defm SHSUBR_ZPmZ : sve2_int_arith_pred<0b101100, "shsubr">; 1335 defm UHSUBR_ZPmZ : sve2_int_arith_pred<0b101110, "uhsubr">; 1336 1337 // SVE2 integer pairwise add and accumulate long 1338 defm SADALP_ZPmZ : sve2_int_sadd_long_accum_pairwise<0, "sadalp">; 1339 defm UADALP_ZPmZ : sve2_int_sadd_long_accum_pairwise<1, "uadalp">; 1340 1341 // SVE2 integer pairwise arithmetic 1342 defm ADDP_ZPmZ : sve2_int_arith_pred<0b100011, "addp">; 1343 defm SMAXP_ZPmZ : sve2_int_arith_pred<0b101001, "smaxp">; 1344 defm UMAXP_ZPmZ : sve2_int_arith_pred<0b101011, "umaxp">; 1345 defm SMINP_ZPmZ : sve2_int_arith_pred<0b101101, "sminp">; 1346 defm UMINP_ZPmZ : sve2_int_arith_pred<0b101111, "uminp">; 1347 1348 // SVE2 integer unary operations (predicated) 1349 defm URECPE_ZPmZ : sve2_int_un_pred_arit_s<0b000, "urecpe">; 1350 defm URSQRTE_ZPmZ : sve2_int_un_pred_arit_s<0b001, "ursqrte">; 1351 defm SQABS_ZPmZ : sve2_int_un_pred_arit<0b100, "sqabs">; 1352 defm SQNEG_ZPmZ : sve2_int_un_pred_arit<0b101, "sqneg">; 1353 1354 // SVE2 saturating add/subtract 1355 defm SQADD_ZPmZ : sve2_int_arith_pred<0b110000, "sqadd">; 1356 defm UQADD_ZPmZ : sve2_int_arith_pred<0b110010, "uqadd">; 1357 defm SQSUB_ZPmZ : sve2_int_arith_pred<0b110100, "sqsub">; 1358 defm UQSUB_ZPmZ : sve2_int_arith_pred<0b110110, "uqsub">; 1359 defm SUQADD_ZPmZ : sve2_int_arith_pred<0b111000, "suqadd">; 1360 defm USQADD_ZPmZ : sve2_int_arith_pred<0b111010, "usqadd">; 1361 defm SQSUBR_ZPmZ : sve2_int_arith_pred<0b111100, "sqsubr">; 1362 defm UQSUBR_ZPmZ : sve2_int_arith_pred<0b111110, "uqsubr">; 1363 1364 // SVE2 saturating/rounding bitwise shift left (predicated) 1365 defm SRSHL_ZPmZ : sve2_int_arith_pred<0b000100, "srshl">; 1366 defm URSHL_ZPmZ : sve2_int_arith_pred<0b000110, "urshl">; 1367 defm SRSHLR_ZPmZ : sve2_int_arith_pred<0b001100, "srshlr">; 1368 defm URSHLR_ZPmZ : sve2_int_arith_pred<0b001110, "urshlr">; 1369 defm SQSHL_ZPmZ : sve2_int_arith_pred<0b010000, "sqshl">; 1370 defm UQSHL_ZPmZ : sve2_int_arith_pred<0b010010, "uqshl">; 1371 defm SQRSHL_ZPmZ : sve2_int_arith_pred<0b010100, "sqrshl">; 1372 defm UQRSHL_ZPmZ : sve2_int_arith_pred<0b010110, "uqrshl">; 1373 defm SQSHLR_ZPmZ : sve2_int_arith_pred<0b011000, "sqshlr">; 1374 defm UQSHLR_ZPmZ : sve2_int_arith_pred<0b011010, "uqshlr">; 1375 defm SQRSHLR_ZPmZ : sve2_int_arith_pred<0b011100, "sqrshlr">; 1376 defm UQRSHLR_ZPmZ : sve2_int_arith_pred<0b011110, "uqrshlr">; 1377 1378 // SVE2 predicated shifts 1379 defm SQSHL_ZPmI : sve_int_bin_pred_shift_imm_left< 0b0110, "sqshl">; 1380 defm UQSHL_ZPmI : sve_int_bin_pred_shift_imm_left< 0b0111, "uqshl">; 1381 defm SRSHR_ZPmI : sve_int_bin_pred_shift_imm_right<0b1100, "srshr">; 1382 defm URSHR_ZPmI : sve_int_bin_pred_shift_imm_right<0b1101, "urshr">; 1383 defm SQSHLU_ZPmI : sve_int_bin_pred_shift_imm_left< 0b1111, "sqshlu">; 1384 1385 // SVE2 integer add/subtract long 1386 defm SADDLB_ZZZ : sve2_wide_int_arith_long<0b00000, "saddlb">; 1387 defm SADDLT_ZZZ : sve2_wide_int_arith_long<0b00001, "saddlt">; 1388 defm UADDLB_ZZZ : sve2_wide_int_arith_long<0b00010, "uaddlb">; 1389 defm UADDLT_ZZZ : sve2_wide_int_arith_long<0b00011, "uaddlt">; 1390 defm SSUBLB_ZZZ : sve2_wide_int_arith_long<0b00100, "ssublb">; 1391 defm SSUBLT_ZZZ : sve2_wide_int_arith_long<0b00101, "ssublt">; 1392 defm USUBLB_ZZZ : sve2_wide_int_arith_long<0b00110, "usublb">; 1393 defm USUBLT_ZZZ : sve2_wide_int_arith_long<0b00111, "usublt">; 1394 defm SABDLB_ZZZ : sve2_wide_int_arith_long<0b01100, "sabdlb">; 1395 defm SABDLT_ZZZ : sve2_wide_int_arith_long<0b01101, "sabdlt">; 1396 defm UABDLB_ZZZ : sve2_wide_int_arith_long<0b01110, "uabdlb">; 1397 defm UABDLT_ZZZ : sve2_wide_int_arith_long<0b01111, "uabdlt">; 1398 1399 // SVE2 integer add/subtract wide 1400 defm SADDWB_ZZZ : sve2_wide_int_arith_wide<0b000, "saddwb">; 1401 defm SADDWT_ZZZ : sve2_wide_int_arith_wide<0b001, "saddwt">; 1402 defm UADDWB_ZZZ : sve2_wide_int_arith_wide<0b010, "uaddwb">; 1403 defm UADDWT_ZZZ : sve2_wide_int_arith_wide<0b011, "uaddwt">; 1404 defm SSUBWB_ZZZ : sve2_wide_int_arith_wide<0b100, "ssubwb">; 1405 defm SSUBWT_ZZZ : sve2_wide_int_arith_wide<0b101, "ssubwt">; 1406 defm USUBWB_ZZZ : sve2_wide_int_arith_wide<0b110, "usubwb">; 1407 defm USUBWT_ZZZ : sve2_wide_int_arith_wide<0b111, "usubwt">; 1408 1409 // SVE2 integer multiply long 1410 defm SQDMULLB_ZZZ : sve2_wide_int_arith_long<0b11000, "sqdmullb">; 1411 defm SQDMULLT_ZZZ : sve2_wide_int_arith_long<0b11001, "sqdmullt">; 1412 defm SMULLB_ZZZ : sve2_wide_int_arith_long<0b11100, "smullb">; 1413 defm SMULLT_ZZZ : sve2_wide_int_arith_long<0b11101, "smullt">; 1414 defm UMULLB_ZZZ : sve2_wide_int_arith_long<0b11110, "umullb">; 1415 defm UMULLT_ZZZ : sve2_wide_int_arith_long<0b11111, "umullt">; 1416 defm PMULLB_ZZZ : sve2_pmul_long<0b0, "pmullb">; 1417 defm PMULLT_ZZZ : sve2_pmul_long<0b1, "pmullt">; 1418 1419 // SVE2 bitwise shift and insert 1420 defm SRI_ZZI : sve2_int_bin_shift_imm_right<0b0, "sri">; 1421 defm SLI_ZZI : sve2_int_bin_shift_imm_left< 0b1, "sli">; 1422 1423 // SVE2 bitwise shift right and accumulate 1424 defm SSRA_ZZI : sve2_int_bin_accum_shift_imm_right<0b00, "ssra">; 1425 defm USRA_ZZI : sve2_int_bin_accum_shift_imm_right<0b01, "usra">; 1426 defm SRSRA_ZZI : sve2_int_bin_accum_shift_imm_right<0b10, "srsra">; 1427 defm URSRA_ZZI : sve2_int_bin_accum_shift_imm_right<0b11, "ursra">; 1428 1429 // SVE2 complex integer add 1430 defm CADD_ZZI : sve2_int_cadd<0b0, "cadd">; 1431 defm SQCADD_ZZI : sve2_int_cadd<0b1, "sqcadd">; 1432 1433 // SVE2 integer absolute difference and accumulate 1434 defm SABA_ZZZ : sve2_int_absdiff_accum<0b0, "saba">; 1435 defm UABA_ZZZ : sve2_int_absdiff_accum<0b1, "uaba">; 1436 1437 // SVE2 integer absolute difference and accumulate long 1438 defm SABALB_ZZZ : sve2_int_absdiff_accum_long<0b00, "sabalb">; 1439 defm SABALT_ZZZ : sve2_int_absdiff_accum_long<0b01, "sabalt">; 1440 defm UABALB_ZZZ : sve2_int_absdiff_accum_long<0b10, "uabalb">; 1441 defm UABALT_ZZZ : sve2_int_absdiff_accum_long<0b11, "uabalt">; 1442 1443 // SVE2 integer add/subtract long with carry 1444 defm ADCLB_ZZZ : sve2_int_addsub_long_carry<0b00, "adclb">; 1445 defm ADCLT_ZZZ : sve2_int_addsub_long_carry<0b01, "adclt">; 1446 defm SBCLB_ZZZ : sve2_int_addsub_long_carry<0b10, "sbclb">; 1447 defm SBCLT_ZZZ : sve2_int_addsub_long_carry<0b11, "sbclt">; 1448 1449 // SVE2 bitwise shift right narrow (bottom) 1450 defm SQSHRUNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b000, "sqshrunb", int_aarch64_sve_sqshrunb>; 1451 defm SQRSHRUNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b001, "sqrshrunb", int_aarch64_sve_sqrshrunb>; 1452 defm SHRNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b010, "shrnb", int_aarch64_sve_shrnb>; 1453 defm RSHRNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b011, "rshrnb", int_aarch64_sve_rshrnb>; 1454 defm SQSHRNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b100, "sqshrnb", int_aarch64_sve_sqshrnb>; 1455 defm SQRSHRNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b101, "sqrshrnb", int_aarch64_sve_sqrshrnb>; 1456 defm UQSHRNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b110, "uqshrnb", int_aarch64_sve_uqshrnb>; 1457 defm UQRSHRNB_ZZI : sve2_int_bin_shift_imm_right_narrow_bottom<0b111, "uqrshrnb", int_aarch64_sve_uqrshrnb>; 1458 1459 // SVE2 bitwise shift right narrow (top) 1460 defm SQSHRUNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b000, "sqshrunt", int_aarch64_sve_sqshrunt>; 1461 defm SQRSHRUNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b001, "sqrshrunt", int_aarch64_sve_sqrshrunt>; 1462 defm SHRNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b010, "shrnt", int_aarch64_sve_shrnt>; 1463 defm RSHRNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b011, "rshrnt", int_aarch64_sve_rshrnt>; 1464 defm SQSHRNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b100, "sqshrnt", int_aarch64_sve_sqshrnt>; 1465 defm SQRSHRNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b101, "sqrshrnt", int_aarch64_sve_sqrshrnt>; 1466 defm UQSHRNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b110, "uqshrnt", int_aarch64_sve_uqshrnt>; 1467 defm UQRSHRNT_ZZI : sve2_int_bin_shift_imm_right_narrow_top<0b111, "uqrshrnt", int_aarch64_sve_uqrshrnt>; 1468 1469 // SVE2 integer add/subtract narrow high part (bottom) 1470 defm ADDHNB_ZZZ : sve2_int_addsub_narrow_high_bottom<0b00, "addhnb", int_aarch64_sve_addhnb>; 1471 defm RADDHNB_ZZZ : sve2_int_addsub_narrow_high_bottom<0b01, "raddhnb", int_aarch64_sve_raddhnb>; 1472 defm SUBHNB_ZZZ : sve2_int_addsub_narrow_high_bottom<0b10, "subhnb", int_aarch64_sve_subhnb>; 1473 defm RSUBHNB_ZZZ : sve2_int_addsub_narrow_high_bottom<0b11, "rsubhnb", int_aarch64_sve_rsubhnb>; 1474 1475 // SVE2 integer add/subtract narrow high part (top) 1476 defm ADDHNT_ZZZ : sve2_int_addsub_narrow_high_top<0b00, "addhnt", int_aarch64_sve_addhnt>; 1477 defm RADDHNT_ZZZ : sve2_int_addsub_narrow_high_top<0b01, "raddhnt", int_aarch64_sve_raddhnt>; 1478 defm SUBHNT_ZZZ : sve2_int_addsub_narrow_high_top<0b10, "subhnt", int_aarch64_sve_subhnt>; 1479 defm RSUBHNT_ZZZ : sve2_int_addsub_narrow_high_top<0b11, "rsubhnt", int_aarch64_sve_rsubhnt>; 1480 1481 // SVE2 saturating extract narrow (bottom) 1482 defm SQXTNB_ZZ : sve2_int_sat_extract_narrow_bottom<0b00, "sqxtnb", int_aarch64_sve_sqxtnb>; 1483 defm UQXTNB_ZZ : sve2_int_sat_extract_narrow_bottom<0b01, "uqxtnb", int_aarch64_sve_uqxtnb>; 1484 defm SQXTUNB_ZZ : sve2_int_sat_extract_narrow_bottom<0b10, "sqxtunb", int_aarch64_sve_sqxtunb>; 1485 1486 // SVE2 saturating extract narrow (top) 1487 defm SQXTNT_ZZ : sve2_int_sat_extract_narrow_top<0b00, "sqxtnt", int_aarch64_sve_sqxtnt>; 1488 defm UQXTNT_ZZ : sve2_int_sat_extract_narrow_top<0b01, "uqxtnt", int_aarch64_sve_uqxtnt>; 1489 defm SQXTUNT_ZZ : sve2_int_sat_extract_narrow_top<0b10, "sqxtunt", int_aarch64_sve_sqxtunt>; 1490 1491 // SVE2 character match 1492 defm MATCH_PPzZZ : sve2_char_match<0b0, "match">; 1493 defm NMATCH_PPzZZ : sve2_char_match<0b1, "nmatch">; 1494 1495 // SVE2 bitwise exclusive-or interleaved 1496 defm EORBT_ZZZ : sve2_bitwise_xor_interleaved<0b0, "eorbt">; 1497 defm EORTB_ZZZ : sve2_bitwise_xor_interleaved<0b1, "eortb">; 1498 1499 // SVE2 bitwise shift left long 1500 defm SSHLLB_ZZI : sve2_bitwise_shift_left_long<0b00, "sshllb">; 1501 defm SSHLLT_ZZI : sve2_bitwise_shift_left_long<0b01, "sshllt">; 1502 defm USHLLB_ZZI : sve2_bitwise_shift_left_long<0b10, "ushllb">; 1503 defm USHLLT_ZZI : sve2_bitwise_shift_left_long<0b11, "ushllt">; 1504 1505 // SVE2 integer add/subtract interleaved long 1506 defm SADDLBT_ZZZ : sve2_misc_int_addsub_long_interleaved<0b00, "saddlbt">; 1507 defm SSUBLBT_ZZZ : sve2_misc_int_addsub_long_interleaved<0b10, "ssublbt">; 1508 defm SSUBLTB_ZZZ : sve2_misc_int_addsub_long_interleaved<0b11, "ssubltb">; 1509 1510 // SVE2 histogram generation (segment) 1511 def HISTSEG_ZZZ : sve2_hist_gen_segment<"histseg">; 1512 1513 // SVE2 histogram generation (vector) 1514 defm HISTCNT_ZPzZZ : sve2_hist_gen_vector<"histcnt">; 1515 1516 // SVE2 floating-point base 2 logarithm as integer 1517 defm FLOGB_ZPmZ : sve2_fp_flogb<"flogb", int_aarch64_sve_flogb>; 1518 1519 // SVE2 floating-point convert precision 1520 defm FCVTXNT_ZPmZ : sve2_fp_convert_down_odd_rounding_top<"fcvtxnt", "int_aarch64_sve_fcvtxnt">; 1521 defm FCVTX_ZPmZ : sve2_fp_convert_down_odd_rounding<"fcvtx", "int_aarch64_sve_fcvtx">; 1522 defm FCVTNT_ZPmZ : sve2_fp_convert_down_narrow<"fcvtnt", "int_aarch64_sve_fcvtnt">; 1523 defm FCVTLT_ZPmZ : sve2_fp_convert_up_long<"fcvtlt", "int_aarch64_sve_fcvtlt">; 1524 1525 // SVE2 floating-point pairwise operations 1526 defm FADDP_ZPmZZ : sve2_fp_pairwise_pred<0b000, "faddp", int_aarch64_sve_faddp>; 1527 defm FMAXNMP_ZPmZZ : sve2_fp_pairwise_pred<0b100, "fmaxnmp", int_aarch64_sve_fmaxnmp>; 1528 defm FMINNMP_ZPmZZ : sve2_fp_pairwise_pred<0b101, "fminnmp", int_aarch64_sve_fminnmp>; 1529 defm FMAXP_ZPmZZ : sve2_fp_pairwise_pred<0b110, "fmaxp", int_aarch64_sve_fmaxp>; 1530 defm FMINP_ZPmZZ : sve2_fp_pairwise_pred<0b111, "fminp", int_aarch64_sve_fminp>; 1531 1532 // SVE2 floating-point multiply-add long (indexed) 1533 defm FMLALB_ZZZI_SHH : sve2_fp_mla_long_by_indexed_elem<0b00, "fmlalb", int_aarch64_sve_fmlalb_lane>; 1534 defm FMLALT_ZZZI_SHH : sve2_fp_mla_long_by_indexed_elem<0b01, "fmlalt", int_aarch64_sve_fmlalt_lane>; 1535 defm FMLSLB_ZZZI_SHH : sve2_fp_mla_long_by_indexed_elem<0b10, "fmlslb", int_aarch64_sve_fmlslb_lane>; 1536 defm FMLSLT_ZZZI_SHH : sve2_fp_mla_long_by_indexed_elem<0b11, "fmlslt", int_aarch64_sve_fmlslt_lane>; 1537 1538 // SVE2 floating-point multiply-add long 1539 defm FMLALB_ZZZ_SHH : sve2_fp_mla_long<0b00, "fmlalb", int_aarch64_sve_fmlalb>; 1540 defm FMLALT_ZZZ_SHH : sve2_fp_mla_long<0b01, "fmlalt", int_aarch64_sve_fmlalt>; 1541 defm FMLSLB_ZZZ_SHH : sve2_fp_mla_long<0b10, "fmlslb", int_aarch64_sve_fmlslb>; 1542 defm FMLSLT_ZZZ_SHH : sve2_fp_mla_long<0b11, "fmlslt", int_aarch64_sve_fmlslt>; 1543 1544 // SVE2 bitwise ternary operations 1545 defm EOR3_ZZZZ_D : sve2_int_bitwise_ternary_op<0b000, "eor3">; 1546 defm BCAX_ZZZZ_D : sve2_int_bitwise_ternary_op<0b010, "bcax">; 1547 def BSL_ZZZZ_D : sve2_int_bitwise_ternary_op_d<0b001, "bsl">; 1548 def BSL1N_ZZZZ_D : sve2_int_bitwise_ternary_op_d<0b011, "bsl1n">; 1549 def BSL2N_ZZZZ_D : sve2_int_bitwise_ternary_op_d<0b101, "bsl2n">; 1550 def NBSL_ZZZZ_D : sve2_int_bitwise_ternary_op_d<0b111, "nbsl">; 1551 1552 // SVE2 bitwise xor and rotate right by immediate 1553 defm XAR_ZZZI : sve2_int_rotate_right_imm<"xar">; 1554 1555 // SVE2 extract vector (immediate offset, constructive) 1556 def EXT_ZZI_B : sve2_int_perm_extract_i_cons<"ext">; 1557 1558 // SVE2 non-temporal gather loads 1559 defm LDNT1SB_ZZR_S : sve2_mem_gldnt_vs<0b00000, "ldnt1sb", Z_s, ZPR32>; 1560 defm LDNT1B_ZZR_S : sve2_mem_gldnt_vs<0b00001, "ldnt1b", Z_s, ZPR32>; 1561 defm LDNT1SH_ZZR_S : sve2_mem_gldnt_vs<0b00100, "ldnt1sh", Z_s, ZPR32>; 1562 defm LDNT1H_ZZR_S : sve2_mem_gldnt_vs<0b00101, "ldnt1h", Z_s, ZPR32>; 1563 defm LDNT1W_ZZR_S : sve2_mem_gldnt_vs<0b01001, "ldnt1w", Z_s, ZPR32>; 1564 1565 defm LDNT1SB_ZZR_D : sve2_mem_gldnt_vs<0b10000, "ldnt1sb", Z_d, ZPR64>; 1566 defm LDNT1B_ZZR_D : sve2_mem_gldnt_vs<0b10010, "ldnt1b", Z_d, ZPR64>; 1567 defm LDNT1SH_ZZR_D : sve2_mem_gldnt_vs<0b10100, "ldnt1sh", Z_d, ZPR64>; 1568 defm LDNT1H_ZZR_D : sve2_mem_gldnt_vs<0b10110, "ldnt1h", Z_d, ZPR64>; 1569 defm LDNT1SW_ZZR_D : sve2_mem_gldnt_vs<0b11000, "ldnt1sw", Z_d, ZPR64>; 1570 defm LDNT1W_ZZR_D : sve2_mem_gldnt_vs<0b11010, "ldnt1w", Z_d, ZPR64>; 1571 defm LDNT1D_ZZR_D : sve2_mem_gldnt_vs<0b11110, "ldnt1d", Z_d, ZPR64>; 1572 1573 // SVE2 vector splice (constructive) 1574 defm SPLICE_ZPZZ : sve2_int_perm_splice_cons<"splice">; 1575 1576 // SVE2 non-temporal scatter stores 1577 defm STNT1B_ZZR_S : sve2_mem_sstnt_vs<0b001, "stnt1b", Z_s, ZPR32>; 1578 defm STNT1H_ZZR_S : sve2_mem_sstnt_vs<0b011, "stnt1h", Z_s, ZPR32>; 1579 defm STNT1W_ZZR_S : sve2_mem_sstnt_vs<0b101, "stnt1w", Z_s, ZPR32>; 1580 1581 defm STNT1B_ZZR_D : sve2_mem_sstnt_vs<0b000, "stnt1b", Z_d, ZPR64>; 1582 defm STNT1H_ZZR_D : sve2_mem_sstnt_vs<0b010, "stnt1h", Z_d, ZPR64>; 1583 defm STNT1W_ZZR_D : sve2_mem_sstnt_vs<0b100, "stnt1w", Z_d, ZPR64>; 1584 defm STNT1D_ZZR_D : sve2_mem_sstnt_vs<0b110, "stnt1d", Z_d, ZPR64>; 1585 1586 // SVE2 table lookup (three sources) 1587 defm TBL_ZZZZ : sve2_int_perm_tbl<"tbl">; 1588 defm TBX_ZZZ : sve2_int_perm_tbx<"tbx">; 1589 1590 // SVE2 integer compare scalar count and limit 1591 defm WHILEGE_PWW : sve_int_while4_rr<0b000, "whilege", int_aarch64_sve_whilege>; 1592 defm WHILEGT_PWW : sve_int_while4_rr<0b001, "whilegt", int_aarch64_sve_whilegt>; 1593 defm WHILEHS_PWW : sve_int_while4_rr<0b100, "whilehs", int_aarch64_sve_whilehs>; 1594 defm WHILEHI_PWW : sve_int_while4_rr<0b101, "whilehi", int_aarch64_sve_whilehi>; 1595 1596 defm WHILEGE_PXX : sve_int_while8_rr<0b000, "whilege", int_aarch64_sve_whilege>; 1597 defm WHILEGT_PXX : sve_int_while8_rr<0b001, "whilegt", int_aarch64_sve_whilegt>; 1598 defm WHILEHS_PXX : sve_int_while8_rr<0b100, "whilehs", int_aarch64_sve_whilehs>; 1599 defm WHILEHI_PXX : sve_int_while8_rr<0b101, "whilehi", int_aarch64_sve_whilehi>; 1600 1601 // SVE2 pointer conflict compare 1602 defm WHILEWR_PXX : sve2_int_while_rr<0b0, "whilewr">; 1603 defm WHILERW_PXX : sve2_int_while_rr<0b1, "whilerw">; 1604} 1605 1606let Predicates = [HasSVE2AES] in { 1607 // SVE2 crypto destructive binary operations 1608 def AESE_ZZZ_B : sve2_crypto_des_bin_op<0b00, "aese", ZPR8>; 1609 def AESD_ZZZ_B : sve2_crypto_des_bin_op<0b01, "aesd", ZPR8>; 1610 1611 // SVE2 crypto unary operations 1612 def AESMC_ZZ_B : sve2_crypto_unary_op<0b0, "aesmc">; 1613 def AESIMC_ZZ_B : sve2_crypto_unary_op<0b1, "aesimc">; 1614 1615 // PMULLB and PMULLT instructions which operate with 64-bit source and 1616 // 128-bit destination elements are enabled with crypto extensions, similar 1617 // to NEON PMULL2 instruction. 1618 def PMULLB_ZZZ_Q : sve2_wide_int_arith<0b00, 0b11010, "pmullb", 1619 ZPR128, ZPR64, ZPR64>; 1620 def PMULLT_ZZZ_Q : sve2_wide_int_arith<0b00, 0b11011, "pmullt", 1621 ZPR128, ZPR64, ZPR64>; 1622} 1623 1624let Predicates = [HasSVE2SM4] in { 1625 // SVE2 crypto constructive binary operations 1626 def SM4EKEY_ZZZ_S : sve2_crypto_cons_bin_op<0b0, "sm4ekey", ZPR32>; 1627 // SVE2 crypto destructive binary operations 1628 def SM4E_ZZZ_S : sve2_crypto_des_bin_op<0b10, "sm4e", ZPR32>; 1629} 1630 1631let Predicates = [HasSVE2SHA3] in { 1632 // SVE2 crypto constructive binary operations 1633 def RAX1_ZZZ_D : sve2_crypto_cons_bin_op<0b1, "rax1", ZPR64>; 1634} 1635 1636let Predicates = [HasSVE2BitPerm] in { 1637 // SVE2 bitwise permute 1638 defm BEXT_ZZZ : sve2_misc_bitwise<0b1100, "bext">; 1639 defm BDEP_ZZZ : sve2_misc_bitwise<0b1101, "bdep">; 1640 defm BGRP_ZZZ : sve2_misc_bitwise<0b1110, "bgrp">; 1641} 1642