| //=- AArch64SMEInstrInfo.td - AArch64 SME Instructions -*- tablegen -*-----=// |
| // |
| // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| // See https://llvm.org/LICENSE.txt for license information. |
| // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| // |
| //===----------------------------------------------------------------------===// |
| // |
| // AArch64 Scalable Matrix Extension (SME) Instruction definitions. |
| // |
| //===----------------------------------------------------------------------===// |
| |
| def AArch64_smstart : SDNode<"AArch64ISD::SMSTART", SDTypeProfile<0, 3, |
| [SDTCisInt<0>, SDTCisInt<0>, SDTCisInt<0>]>, |
| [SDNPHasChain, SDNPSideEffect, SDNPVariadic, |
| SDNPOptInGlue, SDNPOutGlue]>; |
| def AArch64_smstop : SDNode<"AArch64ISD::SMSTOP", SDTypeProfile<0, 3, |
| [SDTCisInt<0>, SDTCisInt<0>, SDTCisInt<0>]>, |
| [SDNPHasChain, SDNPSideEffect, SDNPVariadic, |
| SDNPOptInGlue, SDNPOutGlue]>; |
| def AArch64_restore_za : SDNode<"AArch64ISD::RESTORE_ZA", SDTypeProfile<0, 3, |
| [SDTCisInt<0>, SDTCisPtrTy<1>]>, |
| [SDNPHasChain, SDNPSideEffect, SDNPVariadic, |
| SDNPOptInGlue]>; |
| |
| def AArch64ObscureCopy : SDNode<"AArch64ISD::OBSCURE_COPY", SDTypeProfile<1, 1, []>, []>; |
| |
| //===----------------------------------------------------------------------===// |
| // Instruction naming conventions. |
| //===----------------------------------------------------------------------===// |
| |
| // M = SME array register (ZA) |
| // P = Predicate register |
| // C = Predicate-as-counter register |
| // I = immediate |
| // Z = SVE vector register |
| // T = ZT0 register |
| // |
| |
| //===----------------------------------------------------------------------===// |
| // Add vector elements horizontally or vertically to ZA tile. |
| //===----------------------------------------------------------------------===// |
| |
| def SDT_AArch64RDSVL : SDTypeProfile<1, 1, [SDTCisInt<0>, SDTCisInt<1>]>; |
| def AArch64rdsvl : SDNode<"AArch64ISD::RDSVL", SDT_AArch64RDSVL>; |
| |
| let Predicates = [HasSME] in { |
| def RDSVLI_XI : sve_int_read_vl_a<0b0, 0b11111, "rdsvl", /*streaming_sve=*/0b1>; |
| def ADDSPL_XXI : sve_int_arith_vl<0b1, "addspl", /*streaming_sve=*/0b1>; |
| def ADDSVL_XXI : sve_int_arith_vl<0b0, "addsvl", /*streaming_sve=*/0b1>; |
| |
| defm ADDHA_MPPZ_S : sme_add_vector_to_tile_u32<0b0, "addha", int_aarch64_sme_addha>; |
| defm ADDVA_MPPZ_S : sme_add_vector_to_tile_u32<0b1, "addva", int_aarch64_sme_addva>; |
| |
| def : Pat<(AArch64rdsvl (i32 simm6_32b:$imm)), (RDSVLI_XI simm6_32b:$imm)>; |
| } |
| |
| let Predicates = [HasSMEI16I64] in { |
| defm ADDHA_MPPZ_D : sme_add_vector_to_tile_u64<0b0, "addha", int_aarch64_sme_addha>; |
| defm ADDVA_MPPZ_D : sme_add_vector_to_tile_u64<0b1, "addva", int_aarch64_sme_addva>; |
| } |
| |
| let Predicates = [HasSME] in { |
| //===----------------------------------------------------------------------===// |
| // Outer products |
| //===----------------------------------------------------------------------===// |
| |
| defm BFMOPA_MPPZZ : sme_bf16_outer_product<0b000, "bfmopa", int_aarch64_sme_mopa_wide>; |
| defm BFMOPS_MPPZZ : sme_bf16_outer_product<0b001, "bfmops", int_aarch64_sme_mops_wide>; |
| |
| defm FMOPA_MPPZZ_S : sme_outer_product_fp32<0b0, "fmopa", int_aarch64_sme_mopa>; |
| defm FMOPS_MPPZZ_S : sme_outer_product_fp32<0b1, "fmops", int_aarch64_sme_mops>; |
| } |
| |
| let Predicates = [HasSMEF64F64] in { |
| defm FMOPA_MPPZZ_D : sme_outer_product_fp64<0b0, "fmopa", int_aarch64_sme_mopa>; |
| defm FMOPS_MPPZZ_D : sme_outer_product_fp64<0b1, "fmops", int_aarch64_sme_mops>; |
| } |
| |
| let Predicates = [HasSME] in { |
| defm FMOPAL_MPPZZ : sme_f16_outer_product<0b010, "fmopa", int_aarch64_sme_mopa_wide>; |
| defm FMOPSL_MPPZZ : sme_f16_outer_product<0b011, "fmops", int_aarch64_sme_mops_wide>; |
| |
| defm SMOPA_MPPZZ_S : sme_int_outer_product_i32<0b000, "smopa", int_aarch64_sme_smopa_wide>; |
| defm SMOPS_MPPZZ_S : sme_int_outer_product_i32<0b001, "smops", int_aarch64_sme_smops_wide>; |
| defm UMOPA_MPPZZ_S : sme_int_outer_product_i32<0b110, "umopa", int_aarch64_sme_umopa_wide>; |
| defm UMOPS_MPPZZ_S : sme_int_outer_product_i32<0b111, "umops", int_aarch64_sme_umops_wide>; |
| defm SUMOPA_MPPZZ_S : sme_int_outer_product_i32<0b010, "sumopa", int_aarch64_sme_sumopa_wide>; |
| defm SUMOPS_MPPZZ_S : sme_int_outer_product_i32<0b011, "sumops", int_aarch64_sme_sumops_wide>; |
| defm USMOPA_MPPZZ_S : sme_int_outer_product_i32<0b100, "usmopa", int_aarch64_sme_usmopa_wide>; |
| defm USMOPS_MPPZZ_S : sme_int_outer_product_i32<0b101, "usmops", int_aarch64_sme_usmops_wide>; |
| } |
| |
| let Predicates = [HasSMEI16I64] in { |
| defm SMOPA_MPPZZ_D : sme_int_outer_product_i64<0b000, "smopa", int_aarch64_sme_smopa_wide>; |
| defm SMOPS_MPPZZ_D : sme_int_outer_product_i64<0b001, "smops", int_aarch64_sme_smops_wide>; |
| defm UMOPA_MPPZZ_D : sme_int_outer_product_i64<0b110, "umopa", int_aarch64_sme_umopa_wide>; |
| defm UMOPS_MPPZZ_D : sme_int_outer_product_i64<0b111, "umops", int_aarch64_sme_umops_wide>; |
| defm SUMOPA_MPPZZ_D : sme_int_outer_product_i64<0b010, "sumopa", int_aarch64_sme_sumopa_wide>; |
| defm SUMOPS_MPPZZ_D : sme_int_outer_product_i64<0b011, "sumops", int_aarch64_sme_sumops_wide>; |
| defm USMOPA_MPPZZ_D : sme_int_outer_product_i64<0b100, "usmopa", int_aarch64_sme_usmopa_wide>; |
| defm USMOPS_MPPZZ_D : sme_int_outer_product_i64<0b101, "usmops", int_aarch64_sme_usmops_wide>; |
| } |
| |
| let Predicates = [HasSME] in { |
| //===----------------------------------------------------------------------===// |
| // Loads and stores |
| //===----------------------------------------------------------------------===// |
| |
| defm LD1_MXIPXX : sme_mem_ld_ss<"ld1">; |
| defm ST1_MXIPXX : sme_mem_st_ss<"st1">; |
| |
| //===----------------------------------------------------------------------===// |
| // Spill + fill |
| //===----------------------------------------------------------------------===// |
| |
| defm LDR_ZA : sme_fill<"ldr">; |
| defm STR_ZA : sme_spill<"str">; |
| |
| //===----------------------------------------------------------------------===// |
| // Move instructions |
| //===----------------------------------------------------------------------===// |
| |
| defm INSERT_MXIPZ : sme_vector_to_tile<"mova">; |
| defm EXTRACT_ZPMXI : sme_tile_to_vector<"mova">; |
| |
| //===----------------------------------------------------------------------===// |
| // Zero instruction |
| //===----------------------------------------------------------------------===// |
| |
| defm ZERO_M : sme_zero<"zero">; |
| |
| //===----------------------------------------------------------------------===// |
| // Mode selection and state access instructions |
| //===----------------------------------------------------------------------===// |
| |
| // SME defines three pstate fields to set or clear PSTATE.SM, PSTATE.ZA, or |
| // both fields: |
| // |
| // MSR SVCRSM, #<imm1> |
| // MSR SVCRZA, #<imm1> |
| // MSR SVCRSMZA, #<imm1> |
| // |
| // It's tricky to using the existing pstate operand defined in |
| // AArch64SystemOperands.td since it only encodes 5 bits including op1;op2, |
| // when these fields are also encoded in CRm[3:1]. |
| def MSRpstatesvcrImm1 |
| : PstateWriteSimple<(ins svcr_op:$pstatefield, timm0_1:$imm), "msr", |
| "\t$pstatefield, $imm">, |
| Sched<[WriteSys]> { |
| bits<3> pstatefield; |
| bit imm; |
| let Inst{18-16} = 0b011; // op1 |
| let Inst{11-9} = pstatefield; |
| let Inst{8} = imm; |
| let Inst{7-5} = 0b011; // op2 |
| } |
| |
| def : InstAlias<"smstart", (MSRpstatesvcrImm1 0b011, 0b1)>; |
| def : InstAlias<"smstart sm", (MSRpstatesvcrImm1 0b001, 0b1)>; |
| def : InstAlias<"smstart za", (MSRpstatesvcrImm1 0b010, 0b1)>; |
| |
| def : InstAlias<"smstop", (MSRpstatesvcrImm1 0b011, 0b0)>; |
| def : InstAlias<"smstop sm", (MSRpstatesvcrImm1 0b001, 0b0)>; |
| def : InstAlias<"smstop za", (MSRpstatesvcrImm1 0b010, 0b0)>; |
| |
| |
| // Pseudo to match to smstart/smstop. This expands: |
| // |
| // pseudonode (pstate_za|pstate_sm), before_call, expected_value |
| // |
| // Into: |
| // |
| // if (before_call != expected_value) |
| // node (pstate_za|pstate_sm) |
| // |
| // where node can be either 'smstart' or 'smstop'. |
| def MSRpstatePseudo : |
| Pseudo<(outs), |
| (ins svcr_op:$pstatefield, timm0_1:$imm, GPR64:$rtpstate, timm0_1:$expected_pstate, variable_ops), []>, |
| Sched<[WriteSys]>; |
| |
| // Pseudo to conditionally restore ZA state. This expands: |
| // |
| // pseudonode tpidr2_el0, tpidr2obj, restore_routine |
| // |
| // Into: |
| // |
| // if (tpidr2_el0 == 0) |
| // BL restore_routine, implicit-use tpidr2obj |
| // |
| def RestoreZAPseudo : |
| Pseudo<(outs), |
| (ins GPR64:$tpidr2_el0, GPR64sp:$tpidr2obj, i64imm:$restore_routine, variable_ops), []>, |
| Sched<[]>; |
| |
| def : Pat<(AArch64_restore_za |
| (i64 GPR64:$tpidr2_el0), (i64 GPR64sp:$tpidr2obj), (i64 texternalsym:$restore_routine)), |
| (RestoreZAPseudo GPR64:$tpidr2_el0, GPR64sp:$tpidr2obj, texternalsym:$restore_routine)>; |
| |
| // Scenario A: |
| // |
| // %pstate.before.call = 1 |
| // if (%pstate.before.call != 0) |
| // smstop (pstate_za|pstate_sm) |
| // call fn() |
| // if (%pstate.before.call != 0) |
| // smstart (pstate_za|pstate_sm) |
| // |
| def : Pat<(AArch64_smstop (i32 svcr_op:$pstate), (i64 1), (i64 0)), // before call |
| (MSRpstatesvcrImm1 svcr_op:$pstate, 0b0)>; |
| def : Pat<(AArch64_smstart (i32 svcr_op:$pstate), (i64 1), (i64 0)), // after call |
| (MSRpstatesvcrImm1 svcr_op:$pstate, 0b1)>; |
| |
| // Scenario B: |
| // |
| // %pstate.before.call = 0 |
| // if (%pstate.before.call != 1) |
| // smstart (pstate_za|pstate_sm) |
| // call fn() |
| // if (%pstate.before.call != 1) |
| // smstop (pstate_za|pstate_sm) |
| // |
| def : Pat<(AArch64_smstart (i32 svcr_op:$pstate), (i64 0), (i64 1)), // before call |
| (MSRpstatesvcrImm1 svcr_op:$pstate, 0b1)>; |
| def : Pat<(AArch64_smstop (i32 svcr_op:$pstate), (i64 0), (i64 1)), // after call |
| (MSRpstatesvcrImm1 svcr_op:$pstate, 0b0)>; |
| |
| // The generic case which gets expanded to a pseudo node. |
| def : Pat<(AArch64_smstart (i32 svcr_op:$pstate), (i64 GPR64:$rtpstate), (i64 timm0_1:$expected_pstate)), |
| (MSRpstatePseudo svcr_op:$pstate, 0b1, GPR64:$rtpstate, timm0_1:$expected_pstate)>; |
| def : Pat<(AArch64_smstop (i32 svcr_op:$pstate), (i64 GPR64:$rtpstate), (i64 timm0_1:$expected_pstate)), |
| (MSRpstatePseudo svcr_op:$pstate, 0b0, GPR64:$rtpstate, timm0_1:$expected_pstate)>; |
| |
| // Read and write TPIDR2_EL0 |
| def : Pat<(int_aarch64_sme_set_tpidr2 i64:$val), |
| (MSR 0xde85, GPR64:$val)>; |
| def : Pat<(i64 (int_aarch64_sme_get_tpidr2)), |
| (MRS 0xde85)>; |
| |
| def OBSCURE_COPY : Pseudo<(outs GPR64:$dst), (ins GPR64:$idx), []>, Sched<[]> { } |
| def : Pat<(i64 (AArch64ObscureCopy (i64 GPR64:$idx))), |
| (OBSCURE_COPY GPR64:$idx)>; |
| } // End let Predicates = [HasSME] |
| |
| //===----------------------------------------------------------------------===// |
| // SME2 Instructions |
| //===----------------------------------------------------------------------===// |
| let Predicates = [HasSME2] in { |
| defm ADD_VG2_M2ZZ_S : sme2_dot_mla_add_sub_array_vg2_single<"add", 0b0011010, MatrixOp32, ZZ_s, ZPR4b32, nxv4i32, int_aarch64_sme_add_write_single_za_vg1x2>; |
| defm ADD_VG4_M4ZZ_S : sme2_dot_mla_add_sub_array_vg4_single<"add", 0b0111010, MatrixOp32, ZZZZ_s, ZPR4b32, nxv4i32, int_aarch64_sme_add_write_single_za_vg1x4>; |
| defm ADD_VG2_M2Z2Z_S : sme2_dot_mla_add_sub_array_vg2_multi<"add", 0b011010, MatrixOp32, ZZ_s_mul_r, nxv4i32, int_aarch64_sme_add_write_za_vg1x2>; |
| defm ADD_VG4_M4Z4Z_S : sme2_dot_mla_add_sub_array_vg4_multi<"add", 0b011010, MatrixOp32, ZZZZ_s_mul_r, nxv4i32, int_aarch64_sme_add_write_za_vg1x4>; |
| |
| defm ADD_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"add", 0b0110000>; |
| defm ADD_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"add", 0b0110000>; |
| |
| defm SUB_VG2_M2ZZ_S : sme2_dot_mla_add_sub_array_vg2_single<"sub", 0b0011011, MatrixOp32, ZZ_s, ZPR4b32, nxv4i32, int_aarch64_sme_sub_write_single_za_vg1x2>; |
| defm SUB_VG4_M4ZZ_S : sme2_dot_mla_add_sub_array_vg4_single<"sub", 0b0111011, MatrixOp32, ZZZZ_s, ZPR4b32, nxv4i32, int_aarch64_sme_sub_write_single_za_vg1x4>; |
| defm SUB_VG2_M2Z2Z_S : sme2_dot_mla_add_sub_array_vg2_multi<"sub", 0b011011, MatrixOp32, ZZ_s_mul_r, nxv4i32, int_aarch64_sme_sub_write_za_vg1x2>; |
| defm SUB_VG4_M4Z4Z_S : sme2_dot_mla_add_sub_array_vg4_multi<"sub", 0b011011, MatrixOp32, ZZZZ_s_mul_r, nxv4i32, int_aarch64_sme_sub_write_za_vg1x4>; |
| |
| defm FMLA_VG2_M2ZZ_S : sme2_dot_mla_add_sub_array_vg2_single<"fmla", 0b0011000, MatrixOp32, ZZ_s, ZPR4b32, nxv4f32, int_aarch64_sme_fmla_single_vg1x2>; |
| defm FMLA_VG4_M4ZZ_S : sme2_dot_mla_add_sub_array_vg4_single<"fmla", 0b0111000, MatrixOp32, ZZZZ_s, ZPR4b32, nxv4f32, int_aarch64_sme_fmla_single_vg1x4>; |
| defm FMLA_VG2_M2Z2Z_S : sme2_dot_mla_add_sub_array_vg2_multi<"fmla", 0b011000, MatrixOp32, ZZ_s_mul_r, nxv4f32, int_aarch64_sme_fmla_vg1x2>; |
| defm FMLA_VG4_M4Z4Z_S : sme2_dot_mla_add_sub_array_vg4_multi<"fmla", 0b011000, MatrixOp32, ZZZZ_s_mul_r, nxv4f32, int_aarch64_sme_fmla_vg1x4>; |
| defm FMLA_VG2_M2ZZI_S : sme2_multi_vec_array_vg2_index_32b<"fmla", 0b0000, ZZ_s_mul_r, ZPR4b32, nxv4f32, int_aarch64_sme_fmla_lane_vg1x2>; |
| defm FMLA_VG4_M4ZZI_S : sme2_multi_vec_array_vg4_index_32b<"fmla", 0b0000, ZZZZ_s_mul_r, ZPR4b32, nxv4f32, int_aarch64_sme_fmla_lane_vg1x4>; |
| |
| defm FMLS_VG2_M2ZZ_S : sme2_dot_mla_add_sub_array_vg2_single<"fmls", 0b0011001, MatrixOp32, ZZ_s, ZPR4b32, nxv4f32, int_aarch64_sme_fmls_single_vg1x2>; |
| defm FMLS_VG4_M4ZZ_S : sme2_dot_mla_add_sub_array_vg4_single<"fmls", 0b0111001, MatrixOp32, ZZZZ_s, ZPR4b32, nxv4f32, int_aarch64_sme_fmls_single_vg1x4>; |
| defm FMLS_VG2_M2Z2Z_S : sme2_dot_mla_add_sub_array_vg2_multi<"fmls", 0b011001, MatrixOp32, ZZ_s_mul_r, nxv4f32, int_aarch64_sme_fmls_vg1x2>; |
| defm FMLS_VG4_M4Z4Z_S : sme2_dot_mla_add_sub_array_vg4_multi<"fmls", 0b011001, MatrixOp32, ZZZZ_s_mul_r, nxv4f32, int_aarch64_sme_fmls_vg1x4>; |
| defm FMLS_VG2_M2ZZI_S : sme2_multi_vec_array_vg2_index_32b<"fmls", 0b0010, ZZ_s_mul_r, ZPR4b32, nxv4f32, int_aarch64_sme_fmls_lane_vg1x2>; |
| defm FMLS_VG4_M4ZZI_S : sme2_multi_vec_array_vg4_index_32b<"fmls", 0b0010, ZZZZ_s_mul_r, ZPR4b32, nxv4f32, int_aarch64_sme_fmls_lane_vg1x4>; |
| |
| defm ADD_VG2_M2Z_S : sme2_multivec_accum_add_sub_vg2<"add", 0b0010, MatrixOp32, ZZ_s_mul_r>; |
| defm ADD_VG4_M4Z_S : sme2_multivec_accum_add_sub_vg4<"add", 0b0010, MatrixOp32, ZZZZ_s_mul_r>; |
| |
| defm SUB_VG2_M2Z_S : sme2_multivec_accum_add_sub_vg2<"sub", 0b0011, MatrixOp32, ZZ_s_mul_r>; |
| defm SUB_VG4_M4Z_S : sme2_multivec_accum_add_sub_vg4<"sub", 0b0011, MatrixOp32, ZZZZ_s_mul_r>; |
| |
| defm FADD_VG2_M2Z_S : sme2_multivec_accum_add_sub_vg2<"fadd", 0b0000, MatrixOp32, ZZ_s_mul_r>; |
| defm FADD_VG4_M4Z_S : sme2_multivec_accum_add_sub_vg4<"fadd", 0b0000, MatrixOp32, ZZZZ_s_mul_r>; |
| |
| defm FSUB_VG2_M2Z_S : sme2_multivec_accum_add_sub_vg2<"fsub", 0b0001, MatrixOp32, ZZ_s_mul_r>; |
| defm FSUB_VG4_M4Z_S : sme2_multivec_accum_add_sub_vg4<"fsub", 0b0001, MatrixOp32, ZZZZ_s_mul_r>; |
| |
| defm SQDMULH_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"sqdmulh", 0b1000000>; |
| defm SQDMULH_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"sqdmulh", 0b1000000>; |
| defm SQDMULH_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"sqdmulh", 0b1000000>; |
| defm SQDMULH_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"sqdmulh", 0b1000000>; |
| |
| defm FMLAL_MZZI : sme2_mla_long_array_index<"fmlal", 0b10, 0b00, nxv8f16, int_aarch64_sme_fmlal_lane_vg2x1>; |
| defm FMLAL_VG2_M2ZZI : sme2_fp_mla_long_array_vg2_index<"fmlal", 0b00, nxv8f16, int_aarch64_sme_fmlal_lane_vg2x2>; |
| defm FMLAL_VG4_M4ZZI : sme2_fp_mla_long_array_vg4_index<"fmlal", 0b00, nxv8f16, int_aarch64_sme_fmlal_lane_vg2x4>; |
| defm FMLAL_MZZ : sme2_mla_long_array_single<"fmlal", 0b00, 0b00, nxv8f16, int_aarch64_sme_fmlal_single_vg2x1>; |
| defm FMLAL_VG2_M2ZZ : sme2_fp_mla_long_array_vg2_single<"fmlal", 0b00, nxv8f16, int_aarch64_sme_fmlal_single_vg2x2>; |
| defm FMLAL_VG4_M4ZZ : sme2_fp_mla_long_array_vg4_single<"fmlal", 0b00, nxv8f16, int_aarch64_sme_fmlal_single_vg2x4>; |
| defm FMLAL_VG2_M2Z2Z : sme2_fp_mla_long_array_vg2_multi<"fmlal", 0b00, nxv8f16, int_aarch64_sme_fmlal_vg2x2>; |
| defm FMLAL_VG4_M4Z4Z : sme2_fp_mla_long_array_vg4_multi<"fmlal", 0b00, nxv8f16, int_aarch64_sme_fmlal_vg2x4>; |
| |
| defm FMLSL_MZZI : sme2_mla_long_array_index<"fmlsl", 0b10, 0b01, nxv8f16, int_aarch64_sme_fmlsl_lane_vg2x1>; |
| defm FMLSL_VG2_M2ZZI : sme2_fp_mla_long_array_vg2_index<"fmlsl", 0b01, nxv8f16, int_aarch64_sme_fmlsl_lane_vg2x2>; |
| defm FMLSL_VG4_M4ZZI : sme2_fp_mla_long_array_vg4_index<"fmlsl", 0b01, nxv8f16, int_aarch64_sme_fmlsl_lane_vg2x4>; |
| defm FMLSL_MZZ : sme2_mla_long_array_single<"fmlsl", 0b00, 0b01, nxv8f16, int_aarch64_sme_fmlsl_single_vg2x1>; |
| defm FMLSL_VG2_M2ZZ : sme2_fp_mla_long_array_vg2_single<"fmlsl", 0b01, nxv8f16, int_aarch64_sme_fmlsl_single_vg2x2>; |
| defm FMLSL_VG4_M4ZZ : sme2_fp_mla_long_array_vg4_single<"fmlsl", 0b01, nxv8f16, int_aarch64_sme_fmlsl_single_vg2x4>; |
| defm FMLSL_VG2_M2Z2Z : sme2_fp_mla_long_array_vg2_multi<"fmlsl", 0b01, nxv8f16, int_aarch64_sme_fmlsl_vg2x2>; |
| defm FMLSL_VG4_M4Z4Z : sme2_fp_mla_long_array_vg4_multi<"fmlsl", 0b01, nxv8f16, int_aarch64_sme_fmlsl_vg2x4>; |
| |
| defm BFMLAL_MZZI : sme2_mla_long_array_index<"bfmlal", 0b10, 0b10, nxv8bf16, int_aarch64_sme_fmlal_lane_vg2x1>; |
| defm BFMLAL_VG2_M2ZZI : sme2_fp_mla_long_array_vg2_index<"bfmlal", 0b10, nxv8bf16, int_aarch64_sme_fmlal_lane_vg2x2>; |
| defm BFMLAL_VG4_M4ZZI : sme2_fp_mla_long_array_vg4_index<"bfmlal", 0b10, nxv8bf16, int_aarch64_sme_fmlal_lane_vg2x4>; |
| defm BFMLAL_MZZ : sme2_mla_long_array_single<"bfmlal", 0b00, 0b10, nxv8bf16, int_aarch64_sme_fmlal_single_vg2x1>; |
| defm BFMLAL_VG2_M2ZZ : sme2_fp_mla_long_array_vg2_single<"bfmlal", 0b10, nxv8bf16, int_aarch64_sme_fmlal_single_vg2x2>; |
| defm BFMLAL_VG4_M4ZZ : sme2_fp_mla_long_array_vg4_single<"bfmlal", 0b10, nxv8bf16, int_aarch64_sme_fmlal_single_vg2x4>; |
| defm BFMLAL_VG2_M2Z2Z : sme2_fp_mla_long_array_vg2_multi<"bfmlal", 0b10, nxv8bf16, int_aarch64_sme_fmlal_vg2x2>; |
| defm BFMLAL_VG4_M4Z4Z : sme2_fp_mla_long_array_vg4_multi<"bfmlal", 0b10, nxv8bf16, int_aarch64_sme_fmlal_vg2x4>; |
| |
| defm BFMLSL_MZZI : sme2_mla_long_array_index<"bfmlsl", 0b10, 0b11, nxv8bf16, int_aarch64_sme_fmlsl_lane_vg2x1>; |
| defm BFMLSL_VG2_M2ZZI : sme2_fp_mla_long_array_vg2_index<"bfmlsl", 0b11, nxv8bf16, int_aarch64_sme_fmlsl_lane_vg2x2>; |
| defm BFMLSL_VG4_M4ZZI : sme2_fp_mla_long_array_vg4_index<"bfmlsl", 0b11, nxv8bf16, int_aarch64_sme_fmlsl_lane_vg2x4>; |
| defm BFMLSL_MZZ : sme2_mla_long_array_single<"bfmlsl", 0b00, 0b11, nxv8bf16, int_aarch64_sme_fmlsl_single_vg2x1>; |
| defm BFMLSL_VG2_M2ZZ : sme2_fp_mla_long_array_vg2_single<"bfmlsl", 0b11, nxv8bf16, int_aarch64_sme_fmlsl_single_vg2x2>; |
| defm BFMLSL_VG4_M4ZZ : sme2_fp_mla_long_array_vg4_single<"bfmlsl", 0b11, nxv8bf16, int_aarch64_sme_fmlsl_single_vg2x4>; |
| defm BFMLSL_VG2_M2Z2Z : sme2_fp_mla_long_array_vg2_multi<"bfmlsl", 0b11, nxv8bf16, int_aarch64_sme_fmlsl_vg2x2>; |
| defm BFMLSL_VG4_M4Z4Z : sme2_fp_mla_long_array_vg4_multi<"bfmlsl", 0b11, nxv8bf16, int_aarch64_sme_fmlsl_vg2x4>; |
| |
| defm SMLAL_MZZI : sme2_mla_long_array_index<"smlal", 0b11, 0b00, nxv8i16, int_aarch64_sme_smlal_lane_vg2x1>; |
| defm SMLAL_VG2_M2ZZI : sme2_int_mla_long_array_vg2_index<"smlal", 0b00, int_aarch64_sme_smlal_lane_vg2x2>; |
| defm SMLAL_VG4_M4ZZI : sme2_int_mla_long_array_vg4_index<"smlal", 0b00, int_aarch64_sme_smlal_lane_vg2x4>; |
| defm SMLAL_MZZ : sme2_mla_long_array_single<"smlal",0b01, 0b00, nxv8i16, int_aarch64_sme_smlal_single_vg2x1>; |
| defm SMLAL_VG2_M2ZZ : sme2_int_mla_long_array_vg2_single<"smlal", 0b00, int_aarch64_sme_smlal_single_vg2x2>; |
| defm SMLAL_VG4_M4ZZ : sme2_int_mla_long_array_vg4_single<"smlal", 0b00, int_aarch64_sme_smlal_single_vg2x4>; |
| defm SMLAL_VG2_M2Z2Z : sme2_int_mla_long_array_vg2_multi<"smlal", 0b00, int_aarch64_sme_smlal_vg2x2>; |
| defm SMLAL_VG4_M4Z4Z : sme2_int_mla_long_array_vg4_multi<"smlal", 0b00, int_aarch64_sme_smlal_vg2x4>; |
| |
| defm SMLSL_MZZI : sme2_mla_long_array_index<"smlsl", 0b11, 0b01, nxv8i16, int_aarch64_sme_smlsl_lane_vg2x1>; |
| defm SMLSL_VG2_M2ZZI : sme2_int_mla_long_array_vg2_index<"smlsl", 0b01, int_aarch64_sme_smlsl_lane_vg2x2>; |
| defm SMLSL_VG4_M4ZZI : sme2_int_mla_long_array_vg4_index<"smlsl", 0b01, int_aarch64_sme_smlsl_lane_vg2x4>; |
| defm SMLSL_MZZ : sme2_mla_long_array_single<"smlsl",0b01, 0b01, nxv8i16, int_aarch64_sme_smlsl_single_vg2x1>; |
| defm SMLSL_VG2_M2ZZ : sme2_int_mla_long_array_vg2_single<"smlsl", 0b01, int_aarch64_sme_smlsl_single_vg2x2>; |
| defm SMLSL_VG4_M4ZZ : sme2_int_mla_long_array_vg4_single<"smlsl", 0b01, int_aarch64_sme_smlsl_single_vg2x4>; |
| defm SMLSL_VG2_M2Z2Z : sme2_int_mla_long_array_vg2_multi<"smlsl", 0b01, int_aarch64_sme_smlsl_vg2x2>; |
| defm SMLSL_VG4_M4Z4Z : sme2_int_mla_long_array_vg4_multi<"smlsl", 0b01, int_aarch64_sme_smlsl_vg2x4>; |
| |
| defm UMLAL_MZZI : sme2_mla_long_array_index<"umlal", 0b11, 0b10, nxv8i16, int_aarch64_sme_umlal_lane_vg2x1>; |
| defm UMLAL_VG2_M2ZZI : sme2_int_mla_long_array_vg2_index<"umlal", 0b10, int_aarch64_sme_umlal_lane_vg2x2>; |
| defm UMLAL_VG4_M4ZZI : sme2_int_mla_long_array_vg4_index<"umlal", 0b10, int_aarch64_sme_umlal_lane_vg2x4>; |
| defm UMLAL_MZZ : sme2_mla_long_array_single<"umlal",0b01, 0b10, nxv8i16, int_aarch64_sme_umlal_single_vg2x1>; |
| defm UMLAL_VG2_M2ZZ : sme2_int_mla_long_array_vg2_single<"umlal", 0b10, int_aarch64_sme_umlal_single_vg2x2>; |
| defm UMLAL_VG4_M4ZZ : sme2_int_mla_long_array_vg4_single<"umlal", 0b10, int_aarch64_sme_umlal_single_vg2x4>; |
| defm UMLAL_VG2_M2Z2Z : sme2_int_mla_long_array_vg2_multi<"umlal", 0b10, int_aarch64_sme_umlal_vg2x2>; |
| defm UMLAL_VG4_M4Z4Z : sme2_int_mla_long_array_vg4_multi<"umlal", 0b10, int_aarch64_sme_umlal_vg2x4>; |
| |
| defm UMLSL_MZZI : sme2_mla_long_array_index<"umlsl", 0b11, 0b11, nxv8i16, int_aarch64_sme_umlsl_lane_vg2x1>; |
| defm UMLSL_VG2_M2ZZI : sme2_int_mla_long_array_vg2_index<"umlsl", 0b11, int_aarch64_sme_umlsl_lane_vg2x2>; |
| defm UMLSL_VG4_M4ZZI : sme2_int_mla_long_array_vg4_index<"umlsl", 0b11, int_aarch64_sme_umlsl_lane_vg2x4>; |
| defm UMLSL_MZZ : sme2_mla_long_array_single<"umlsl",0b01, 0b11, nxv8i16, int_aarch64_sme_umlsl_single_vg2x1>; |
| defm UMLSL_VG2_M2ZZ : sme2_int_mla_long_array_vg2_single<"umlsl", 0b11, int_aarch64_sme_umlsl_single_vg2x2>; |
| defm UMLSL_VG4_M4ZZ : sme2_int_mla_long_array_vg4_single<"umlsl", 0b11, int_aarch64_sme_umlsl_single_vg2x4>; |
| defm UMLSL_VG2_M2Z2Z : sme2_int_mla_long_array_vg2_multi<"umlsl", 0b11, int_aarch64_sme_umlsl_vg2x2>; |
| defm UMLSL_VG4_M4Z4Z : sme2_int_mla_long_array_vg4_multi<"umlsl", 0b11, int_aarch64_sme_umlsl_vg2x4>; |
| |
| defm FCVT_Z2Z_StoH : sme2_cvt_vg2_single<"fcvt", 0b0000, nxv8f16, nxv4f32, int_aarch64_sve_fcvt_x2>; |
| defm FCVTN_Z2Z_StoH : sme2_cvt_vg2_single<"fcvtn", 0b0001, nxv8f16, nxv4f32, int_aarch64_sve_fcvtn_x2>; |
| defm BFCVT_Z2Z_StoH : sme2_cvt_vg2_single<"bfcvt", 0b1000, nxv8bf16, nxv4f32, int_aarch64_sve_bfcvt_x2>; |
| defm BFCVTN_Z2Z_StoH : sme2_cvt_vg2_single<"bfcvtn", 0b1001, nxv8bf16, nxv4f32, int_aarch64_sve_bfcvtn_x2>; |
| |
| defm SQCVT_Z2Z_StoH : sme2_cvt_vg2_single<"sqcvt", 0b0110, nxv8i16, nxv4i32, int_aarch64_sve_sqcvt_x2>; |
| defm UQCVT_Z2Z_StoH : sme2_cvt_vg2_single<"uqcvt", 0b0111, nxv8i16, nxv4i32, int_aarch64_sve_uqcvt_x2>; |
| defm SQCVTU_Z2Z_StoH : sme2_cvt_vg2_single<"sqcvtu", 0b1110, nxv8i16, nxv4i32, int_aarch64_sve_sqcvtu_x2>; |
| defm SQCVT_Z4Z : sme2_int_cvt_vg4_single<"sqcvt", 0b000, int_aarch64_sve_sqcvt_x4>; |
| defm UQCVT_Z4Z : sme2_int_cvt_vg4_single<"uqcvt", 0b001, int_aarch64_sve_uqcvt_x4>; |
| defm SQCVTU_Z4Z : sme2_int_cvt_vg4_single<"sqcvtu", 0b100, int_aarch64_sve_sqcvtu_x4>; |
| defm SQCVTN_Z4Z : sme2_int_cvt_vg4_single<"sqcvtn", 0b010, int_aarch64_sve_sqcvtn_x4>; |
| defm SQCVTUN_Z4Z : sme2_int_cvt_vg4_single<"sqcvtun", 0b110, int_aarch64_sve_sqcvtun_x4>; |
| defm UQCVTN_Z4Z : sme2_int_cvt_vg4_single<"uqcvtn", 0b011, int_aarch64_sve_uqcvtn_x4>; |
| |
| defm FCVTZS_2Z2Z_StoS : sme2_fp_cvt_vg2_multi<"fcvtzs", 0b00010>; |
| defm FCVTZS_4Z4Z_StoS : sme2_fp_cvt_vg4_multi<"fcvtzs", 0b0001000>; |
| defm FCVTZU_2Z2Z_StoS : sme2_fp_cvt_vg2_multi<"fcvtzu", 0b00011>; |
| defm FCVTZU_4Z4Z_StoS : sme2_fp_cvt_vg4_multi<"fcvtzu", 0b0001010>; |
| defm SCVTF_2Z2Z_StoS : sme2_fp_cvt_vg2_multi<"scvtf", 0b00100>; |
| defm SCVTF_4Z4Z_StoS : sme2_fp_cvt_vg4_multi<"scvtf", 0b0010000>; |
| defm UCVTF_2Z2Z_StoS : sme2_fp_cvt_vg2_multi<"ucvtf", 0b00101>; |
| defm UCVTF_4Z4Z_StoS : sme2_fp_cvt_vg4_multi<"ucvtf", 0b0010010>; |
| |
| defm SMAX_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"smax", 0b0000000>; |
| defm SMAX_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"smax", 0b0000000>; |
| defm SMAX_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"smax", 0b0000000>; |
| defm SMAX_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"smax", 0b0000000>; |
| |
| defm UMAX_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"umax", 0b0000001>; |
| defm UMAX_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"umax", 0b0000001>; |
| defm UMAX_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"umax", 0b0000001>; |
| defm UMAX_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"umax", 0b0000001>; |
| |
| defm SMIN_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"smin", 0b0000010>; |
| defm SMIN_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"smin", 0b0000010>; |
| defm SMIN_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"smin", 0b0000010>; |
| defm SMIN_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"smin", 0b0000010>; |
| |
| defm UMIN_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"umin", 0b0000011>; |
| defm UMIN_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"umin", 0b0000011>; |
| defm UMIN_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"umin", 0b0000011>; |
| defm UMIN_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"umin", 0b0000011>; |
| |
| defm FMAX_VG2_2ZZ : sme2_fp_sve_destructive_vector_vg2_single<"fmax", 0b0010000>; |
| defm FMAX_VG4_4ZZ : sme2_fp_sve_destructive_vector_vg4_single<"fmax", 0b0010000>; |
| defm FMAX_VG2_2Z2Z : sme2_fp_sve_destructive_vector_vg2_multi<"fmax", 0b0010000>; |
| defm FMAX_VG4_4Z4Z : sme2_fp_sve_destructive_vector_vg4_multi<"fmax", 0b0010000>; |
| |
| defm FMIN_VG2_2ZZ : sme2_fp_sve_destructive_vector_vg2_single<"fmin", 0b0010001>; |
| defm FMIN_VG4_4ZZ : sme2_fp_sve_destructive_vector_vg4_single<"fmin", 0b0010001>; |
| defm FMIN_VG2_2Z2Z : sme2_fp_sve_destructive_vector_vg2_multi<"fmin", 0b0010001>; |
| defm FMIN_VG4_4Z4Z : sme2_fp_sve_destructive_vector_vg4_multi<"fmin", 0b0010001>; |
| |
| defm FMAXNM_VG2_2ZZ : sme2_fp_sve_destructive_vector_vg2_single<"fmaxnm", 0b0010010>; |
| defm FMAXNM_VG4_4ZZ : sme2_fp_sve_destructive_vector_vg4_single<"fmaxnm", 0b0010010>; |
| defm FMAXNM_VG2_2Z2Z : sme2_fp_sve_destructive_vector_vg2_multi<"fmaxnm", 0b0010010>; |
| defm FMAXNM_VG4_4Z4Z : sme2_fp_sve_destructive_vector_vg4_multi<"fmaxnm", 0b0010010>; |
| |
| defm FMINNM_VG2_2ZZ : sme2_fp_sve_destructive_vector_vg2_single<"fminnm", 0b0010011>; |
| defm FMINNM_VG4_4ZZ : sme2_fp_sve_destructive_vector_vg4_single<"fminnm", 0b0010011>; |
| defm FMINNM_VG2_2Z2Z : sme2_fp_sve_destructive_vector_vg2_multi<"fminnm", 0b0010011>; |
| defm FMINNM_VG4_4Z4Z : sme2_fp_sve_destructive_vector_vg4_multi<"fminnm", 0b0010011>; |
| |
| defm SRSHL_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"srshl", 0b0100010>; |
| defm SRSHL_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"srshl", 0b0100010>; |
| defm SRSHL_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"srshl", 0b0100010>; |
| defm SRSHL_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"srshl", 0b0100010>; |
| |
| defm URSHL_VG2_2ZZ : sme2_int_sve_destructive_vector_vg2_single<"urshl", 0b0100011>; |
| defm URSHL_VG4_4ZZ : sme2_int_sve_destructive_vector_vg4_single<"urshl", 0b0100011>; |
| defm URSHL_VG2_2Z2Z : sme2_int_sve_destructive_vector_vg2_multi<"urshl", 0b0100011>; |
| defm URSHL_VG4_4Z4Z : sme2_int_sve_destructive_vector_vg4_multi<"urshl", 0b0100011>; |
| |
| defm FCLAMP_VG2_2Z2Z : sme2_fp_clamp_vector_vg2_multi<"fclamp">; |
| defm FCLAMP_VG4_4Z4Z : sme2_fp_clamp_vector_vg4_multi<"fclamp">; |
| |
| defm SCLAMP_VG2_2Z2Z : sme2_int_clamp_vector_vg2_multi<"sclamp", 0b0>; |
| defm SCLAMP_VG4_4Z4Z : sme2_int_clamp_vector_vg4_multi<"sclamp", 0b0>; |
| |
| defm UCLAMP_VG2_2Z2Z : sme2_int_clamp_vector_vg2_multi<"uclamp", 0b1>; |
| defm UCLAMP_VG4_4Z4Z : sme2_int_clamp_vector_vg4_multi<"uclamp", 0b1>; |
| |
| defm FDOT_VG2_M2ZZI_HtoS : sme2_multi_vec_array_vg2_index_32b<"fdot", 0b1001, ZZ_h_mul_r, ZPR4b16, nxv8f16, null_frag>; |
| defm FDOT_VG4_M4ZZI_HtoS : sme2_multi_vec_array_vg4_index_32b<"fdot", 0b1001, ZZZZ_h_mul_r, ZPR4b16, nxv8f16, null_frag>; |
| defm FDOT_VG2_M2ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"fdot", 0b0010000, MatrixOp32, ZZ_h, ZPR4b16>; |
| defm FDOT_VG4_M4ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"fdot", 0b0110000, MatrixOp32, ZZZZ_h, ZPR4b16>; |
| defm FDOT_VG2_M2Z2Z_HtoS : sme2_dot_mla_add_sub_array_vg2_multi<"fdot", 0b010000, MatrixOp32, ZZ_h_mul_r, nxv8f16, null_frag>; |
| defm FDOT_VG4_M4Z4Z_HtoS : sme2_dot_mla_add_sub_array_vg4_multi<"fdot", 0b010000, MatrixOp32, ZZZZ_h_mul_r, nxv8f16, null_frag>; |
| |
| defm BFDOT_VG2_M2ZZI_HtoS : sme2_multi_vec_array_vg2_index_32b<"bfdot", 0b1011, ZZ_h_mul_r, ZPR4b16, nxv8bf16, null_frag>; |
| defm BFDOT_VG4_M4ZZI_HtoS : sme2_multi_vec_array_vg4_index_32b<"bfdot", 0b1011, ZZZZ_h_mul_r, ZPR4b16, nxv8bf16, null_frag>; |
| defm BFDOT_VG2_M2ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"bfdot", 0b0010010, MatrixOp32, ZZ_h, ZPR4b16>; |
| defm BFDOT_VG4_M4ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"bfdot", 0b0110010, MatrixOp32, ZZZZ_h, ZPR4b16>; |
| defm BFDOT_VG2_M2Z2Z_HtoS : sme2_dot_mla_add_sub_array_vg2_multi<"bfdot", 0b010010, MatrixOp32, ZZ_h_mul_r, nxv8bf16, null_frag>; |
| defm BFDOT_VG4_M4Z4Z_HtoS : sme2_dot_mla_add_sub_array_vg4_multi<"bfdot", 0b010010, MatrixOp32, ZZZZ_h_mul_r, nxv8bf16, null_frag>; |
| |
| defm BFVDOT_VG2_M2ZZI_HtoS : sme2_multi_vec_array_vg2_index_32b<"bfvdot", 0b0011, ZZ_h_mul_r, ZPR4b16, nxv8bf16, int_aarch64_sme_fvdot_lane_za32_vg1x2>; |
| |
| defm FVDOT_VG2_M2ZZI_HtoS : sme2_multi_vec_array_vg2_index_32b<"fvdot", 0b0001, ZZ_h_mul_r, ZPR4b16, nxv8f16, int_aarch64_sme_fvdot_lane_za32_vg1x2>; |
| |
| defm SDOT_VG2_M2ZZI_HToS : sme2_multi_vec_array_vg2_index_32b<"sdot", 0b1000, ZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm SDOT_VG2_M2ZZI_BToS : sme2_multi_vec_array_vg2_index_32b<"sdot", 0b1100, ZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm SDOT_VG4_M4ZZI_HToS : sme2_multi_vec_array_vg4_index_32b<"sdot", 0b1000, ZZZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm SDOT_VG4_M4ZZI_BToS : sme2_multi_vec_array_vg4_index_32b<"sdot", 0b1100, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm SDOT_VG2_M2ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"sdot", 0b1010101, MatrixOp32, ZZ_h, ZPR4b16>; |
| defm SDOT_VG4_M4ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"sdot", 0b1110101, MatrixOp32, ZZZZ_h, ZPR4b16>; |
| defm SDOT_VG2_M2Z2Z_HtoS : sme2_dot_mla_add_sub_array_vg2_multi<"sdot", 0b110101, MatrixOp32, ZZ_h_mul_r, nxv8i16, null_frag>; |
| defm SDOT_VG4_M4Z4Z_HtoS : sme2_dot_mla_add_sub_array_vg4_multi<"sdot", 0b110101, MatrixOp32, ZZZZ_h_mul_r, nxv8i16, null_frag>; |
| defm SDOT_VG2_M2ZZ_BtoS : sme2_dot_mla_add_sub_array_vg24_single<"sdot", 0b0010100, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm SDOT_VG4_M4ZZ_BtoS : sme2_dot_mla_add_sub_array_vg24_single<"sdot", 0b0110100, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm SDOT_VG2_M2Z2Z_BtoS : sme2_dot_mla_add_sub_array_vg2_multi<"sdot", 0b010100, MatrixOp32, ZZ_b_mul_r, nxv16i8, null_frag>; |
| defm SDOT_VG4_M4Z4Z_BtoS : sme2_dot_mla_add_sub_array_vg4_multi<"sdot", 0b010100, MatrixOp32, ZZZZ_b_mul_r, nxv16i8, null_frag>; |
| |
| defm SUDOT_VG2_M2ZZI_BToS : sme2_multi_vec_array_vg2_index_32b<"sudot", 0b1111, ZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm SUDOT_VG4_M4ZZI_BToS : sme2_multi_vec_array_vg4_index_32b<"sudot", 0b1111, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm SUDOT_VG2_M2ZZ_BToS : sme2_dot_mla_add_sub_array_vg24_single<"sudot", 0b0010111, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm SUDOT_VG4_M4ZZ_BToS : sme2_dot_mla_add_sub_array_vg24_single<"sudot", 0b0110111, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| |
| defm SVDOT_VG2_M2ZZI_HtoS : sme2_multi_vec_array_vg2_index_32b<"svdot", 0b0100, ZZ_h_mul_r, ZPR4b16, nxv8i16, int_aarch64_sme_svdot_lane_za32_vg1x2>; |
| defm SVDOT_VG4_M4ZZI_BtoS : sme2_multi_vec_array_vg4_index_32b<"svdot", 0b0100, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, int_aarch64_sme_svdot_lane_za32_vg1x4>; |
| |
| defm SUVDOT_VG4_M4ZZI_BToS : sme2_multi_vec_array_vg4_index_32b<"suvdot", 0b0111, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, int_aarch64_sme_suvdot_lane_za32_vg1x4>; |
| |
| defm UDOT_VG2_M2ZZI_HToS : sme2_multi_vec_array_vg2_index_32b<"udot", 0b1010, ZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm UDOT_VG2_M2ZZI_BToS : sme2_multi_vec_array_vg2_index_32b<"udot", 0b1110, ZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm UDOT_VG4_M4ZZI_BtoS : sme2_multi_vec_array_vg4_index_32b<"udot", 0b1110, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm UDOT_VG4_M4ZZI_HToS : sme2_multi_vec_array_vg4_index_32b<"udot", 0b1010, ZZZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm UDOT_VG2_M2ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"udot", 0b1010111, MatrixOp32, ZZ_h, ZPR4b16>; |
| defm UDOT_VG4_M4ZZ_HtoS : sme2_dot_mla_add_sub_array_vg24_single<"udot", 0b1110111, MatrixOp32, ZZZZ_h, ZPR4b16>; |
| defm UDOT_VG2_M2Z2Z_HtoS : sme2_dot_mla_add_sub_array_vg2_multi<"udot", 0b110111, MatrixOp32, ZZ_h_mul_r, nxv8i16, null_frag>; |
| defm UDOT_VG4_M4Z4Z_HtoS : sme2_dot_mla_add_sub_array_vg4_multi<"udot", 0b110111, MatrixOp32, ZZZZ_h_mul_r, nxv8i16, null_frag>; |
| defm UDOT_VG2_M2ZZ_BtoS : sme2_dot_mla_add_sub_array_vg24_single<"udot", 0b0010110, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm UDOT_VG4_M4ZZ_BtoS : sme2_dot_mla_add_sub_array_vg24_single<"udot", 0b0110110, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm UDOT_VG2_M2Z2Z_BtoS : sme2_dot_mla_add_sub_array_vg2_multi<"udot", 0b010110, MatrixOp32, ZZ_b_mul_r, nxv16i8, null_frag>; |
| defm UDOT_VG4_M4Z4Z_BtoS : sme2_dot_mla_add_sub_array_vg4_multi<"udot", 0b010110, MatrixOp32, ZZZZ_b_mul_r, nxv16i8, null_frag>; |
| |
| defm USDOT_VG2_M2ZZI_BToS: sme2_multi_vec_array_vg2_index_32b<"usdot", 0b1101, ZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm USDOT_VG4_M4ZZI_BToS: sme2_multi_vec_array_vg4_index_32b<"usdot", 0b1101, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, null_frag>; |
| defm USDOT_VG2_M2ZZ_BToS : sme2_dot_mla_add_sub_array_vg24_single<"usdot", 0b0010101, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm USDOT_VG4_M4ZZ_BToS : sme2_dot_mla_add_sub_array_vg24_single<"usdot", 0b0110101, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm USDOT_VG2_M2Z2Z_BToS : sme2_dot_mla_add_sub_array_vg2_multi<"usdot", 0b010101, MatrixOp32, ZZ_b_mul_r, nxv16i8, null_frag>; |
| defm USDOT_VG4_M4Z4Z_BToS : sme2_dot_mla_add_sub_array_vg4_multi<"usdot", 0b010101, MatrixOp32, ZZZZ_b_mul_r, nxv16i8, null_frag>; |
| |
| defm USVDOT_VG4_M4ZZI_BToS : sme2_multi_vec_array_vg4_index_32b<"usvdot", 0b0101, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, int_aarch64_sme_usvdot_lane_za32_vg1x4>; |
| |
| defm UVDOT_VG2_M2ZZI_HtoS : sme2_multi_vec_array_vg2_index_32b<"uvdot", 0b0110, ZZ_h_mul_r, ZPR4b16, nxv8i16, int_aarch64_sme_uvdot_lane_za32_vg1x2>; |
| defm UVDOT_VG4_M4ZZI_BtoS : sme2_multi_vec_array_vg4_index_32b<"uvdot", 0b0110, ZZZZ_b_mul_r, ZPR4b8, nxv16i8, int_aarch64_sme_uvdot_lane_za32_vg1x4>; |
| |
| def SMLALL_MZZI_BtoS : sme2_mla_ll_array_index_32b<"smlall", 0b000>; |
| defm SMLALL_VG2_M2ZZI_BtoS : sme2_mla_ll_array_vg2_index_32b<"smlall", 0b000>; |
| defm SMLALL_VG4_M4ZZI_BtoS : sme2_mla_ll_array_vg4_index_32b<"smlall", 0b000>; |
| def SMLALL_MZZ_BtoS : sme2_mla_ll_array_single<"smlall", 0b0000, MatrixOp32, ZPR8, ZPR4b8>; |
| defm SMLALL_VG2_M2ZZ_BtoS : sme2_mla_ll_array_vg24_single<"smlall", 0b00000, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm SMLALL_VG4_M4ZZ_BtoS : sme2_mla_ll_array_vg24_single<"smlall", 0b01000, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm SMLALL_VG2_M2Z2Z_BtoS : sme2_mla_ll_array_vg2_multi<"smlall", 0b0000, MatrixOp32, ZZ_b_mul_r>; |
| defm SMLALL_VG4_M4Z4Z_BtoS : sme2_mla_ll_array_vg4_multi<"smlall", 0b0000, MatrixOp32, ZZZZ_b_mul_r>; |
| |
| def USMLALL_MZZI_BtoS : sme2_mla_ll_array_index_32b<"usmlall", 0b001>; |
| defm USMLALL_VG2_M2ZZI_BtoS : sme2_mla_ll_array_vg2_index_32b<"usmlall", 0b100>; |
| defm USMLALL_VG4_M4ZZI_BtoS : sme2_mla_ll_array_vg4_index_32b<"usmlall", 0b100>; |
| def USMLALL_MZZ_BtoS : sme2_mla_ll_array_single<"usmlall", 0b0001, MatrixOp32, ZPR8, ZPR4b8>; |
| defm USMLALL_VG2_M2ZZ_BtoS : sme2_mla_ll_array_vg24_single<"usmlall", 0b00001, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm USMLALL_VG4_M4ZZ_BtoS : sme2_mla_ll_array_vg24_single<"usmlall", 0b01001, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm USMLALL_VG2_M2Z2Z_BtoS : sme2_mla_ll_array_vg2_multi<"usmlall", 0b0001, MatrixOp32, ZZ_b_mul_r>; |
| defm USMLALL_VG4_M4Z4Z_BtoS : sme2_mla_ll_array_vg4_multi<"usmlall", 0b0001, MatrixOp32, ZZZZ_b_mul_r>; |
| |
| def SMLSLL_MZZI_BtoS : sme2_mla_ll_array_index_32b<"smlsll", 0b010>; |
| defm SMLSLL_VG2_M2ZZI_BtoS : sme2_mla_ll_array_vg2_index_32b<"smlsll", 0b001>; |
| defm SMLSLL_VG4_M4ZZI_BtoS : sme2_mla_ll_array_vg4_index_32b<"smlsll", 0b001>; |
| def SMLSLL_MZZ_BtoS : sme2_mla_ll_array_single<"smlsll", 0b0010, MatrixOp32, ZPR8, ZPR4b8>; |
| defm SMLSLL_VG2_M2ZZ_BtoS : sme2_mla_ll_array_vg24_single<"smlsll", 0b00010, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm SMLSLL_VG4_M4ZZ_BtoS : sme2_mla_ll_array_vg24_single<"smlsll", 0b01010, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm SMLSLL_VG2_M2Z2Z_BtoS : sme2_mla_ll_array_vg2_multi<"smlsll", 0b0010, MatrixOp32, ZZ_b_mul_r>; |
| defm SMLSLL_VG4_M4Z4Z_BtoS : sme2_mla_ll_array_vg4_multi<"smlsll", 0b0010, MatrixOp32, ZZZZ_b_mul_r>; |
| |
| def UMLALL_MZZI_BtoS : sme2_mla_ll_array_index_32b<"umlall", 0b100>; |
| defm UMLALL_VG2_M2ZZI_BtoS : sme2_mla_ll_array_vg2_index_32b<"umlall", 0b010>; |
| defm UMLALL_VG4_M4ZZI_BtoS : sme2_mla_ll_array_vg4_index_32b<"umlall", 0b010>; |
| def UMLALL_MZZ_BtoS : sme2_mla_ll_array_single<"umlall", 0b0100, MatrixOp32, ZPR8, ZPR4b8>; |
| defm UMLALL_VG2_M2ZZ_BtoS : sme2_mla_ll_array_vg24_single<"umlall", 0b00100, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm UMLALL_VG4_M4ZZ_BtoS : sme2_mla_ll_array_vg24_single<"umlall", 0b01100, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm UMLALL_VG2_M2Z2Z_BtoS : sme2_mla_ll_array_vg2_multi<"umlall", 0b0100, MatrixOp32, ZZ_b_mul_r>; |
| defm UMLALL_VG4_M4Z4Z_BtoS : sme2_mla_ll_array_vg4_multi<"umlall", 0b0100, MatrixOp32, ZZZZ_b_mul_r>; |
| |
| def SUMLALL_MZZI_BtoS : sme2_mla_ll_array_index_32b<"sumlall", 0b101>; |
| defm SUMLALL_VG2_M2ZZI_BtoS : sme2_mla_ll_array_vg2_index_32b<"sumlall", 0b110>; |
| defm SUMLALL_VG4_M4ZZI_BtoS : sme2_mla_ll_array_vg4_index_32b<"sumlall", 0b110>; |
| defm SUMLALL_VG2_M2ZZ_BtoS : sme2_mla_ll_array_vg24_single<"sumlall", 0b00101, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm SUMLALL_VG4_M4ZZ_BtoS : sme2_mla_ll_array_vg24_single<"sumlall", 0b01101, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| |
| def UMLSLL_MZZI_BtoS : sme2_mla_ll_array_index_32b<"umlsll", 0b110>; |
| defm UMLSLL_VG2_M2ZZI_BtoS : sme2_mla_ll_array_vg2_index_32b<"umlsll", 0b011>; |
| defm UMLSLL_VG4_M4ZZI_BtoS : sme2_mla_ll_array_vg4_index_32b<"umlsll", 0b011>; |
| def UMLSLL_MZZ_BtoS : sme2_mla_ll_array_single<"umlsll", 0b0110, MatrixOp32, ZPR8, ZPR4b8>; |
| defm UMLSLL_VG2_M2ZZ_BtoS : sme2_mla_ll_array_vg24_single<"umlsll", 0b00110, MatrixOp32, ZZ_b, ZPR4b8>; |
| defm UMLSLL_VG4_M4ZZ_BtoS : sme2_mla_ll_array_vg24_single<"umlsll", 0b01110, MatrixOp32, ZZZZ_b, ZPR4b8>; |
| defm UMLSLL_VG2_M2Z2Z_BtoS : sme2_mla_ll_array_vg2_multi<"umlsll", 0b0110, MatrixOp32, ZZ_b_mul_r>; |
| defm UMLSLL_VG4_M4Z4Z_BtoS : sme2_mla_ll_array_vg4_multi<"umlsll", 0b0110, MatrixOp32, ZZZZ_b_mul_r>; |
| |
| defm BMOPA_MPPZZ_S : sme2_bfp_mopx_tile<"bmopa", 0b100>; |
| defm BMOPS_MPPZZ_S : sme2_bfp_mopx_tile<"bmops", 0b101>; |
| |
| defm SMOPA_MPPZZ_HtoS : sme2_int_mopx_tile<"smopa", 0b000>; |
| defm SMOPS_MPPZZ_HtoS : sme2_int_mopx_tile<"smops", 0b001>; |
| |
| defm UMOPA_MPPZZ_HtoS : sme2_int_mopx_tile<"umopa", 0b100>; |
| defm UMOPS_MPPZZ_HtoS : sme2_int_mopx_tile<"umops", 0b101>; |
| |
| def ZERO_T : sme2_zero_zt<"zero", 0b0001>; |
| |
| def LDR_TX : sme2_spill_fill_vector<"ldr", 0b01111100>; |
| def STR_TX : sme2_spill_fill_vector<"str", 0b11111100>; |
| |
| def MOVT_XTI : sme2_movt_zt_to_scalar<"movt", 0b0011111>; |
| def MOVT_TIX : sme2_movt_scalar_to_zt<"movt", 0b0011111>; |
| |
| defm LUTI2_ZTZI : sme2_luti2_vector_index<"luti2">; |
| defm LUTI2_2ZTZI : sme2_luti2_vector_vg2_index<"luti2">; |
| defm LUTI2_4ZTZI : sme2_luti2_vector_vg4_index<"luti2">; |
| |
| defm LUTI4_ZTZI : sme2_luti4_vector_index<"luti4">; |
| defm LUTI4_2ZTZI : sme2_luti4_vector_vg2_index<"luti4">; |
| defm LUTI4_4ZTZI : sme2_luti4_vector_vg4_index<"luti4">; |
| |
| defm SUNPK_VG2_2ZZ : sme2_unpk_vector_vg2<"sunpk", 0b0>; |
| defm SUNPK_VG4_4Z2Z : sme2_unpk_vector_vg4<"sunpk", 0b0>; |
| defm UUNPK_VG2_2ZZ : sme2_unpk_vector_vg2<"uunpk", 0b1>; |
| defm UUNPK_VG4_4Z2Z : sme2_unpk_vector_vg4<"uunpk", 0b1>; |
| |
| defm ZIP_VG2_2ZZZ : sme2_zip_vector_vg2<"zip", 0b0>; |
| defm UZP_VG2_2ZZZ : sme2_zip_vector_vg2<"uzp", 0b1>; |
| defm ZIP_VG4_4Z4Z : sme2_zip_vector_vg4<"zip", 0b0110000>; |
| defm UZP_VG4_4Z4Z : sme2_zip_vector_vg4<"uzp", 0b0110001>; |
| defm ZIP_VG4_4Z4Z_Q : sme2_zip_vector_vg4_Q<"zip", 0b0111000>; |
| defm UZP_VG4_4Z4Z_Q : sme2_zip_vector_vg4_Q<"uzp", 0b0111001>; |
| |
| defm FRINTA_2Z2Z: sme2_frint_vector_vg2_multi<"frinta", 0b11000>; |
| defm FRINTA_4Z4Z: sme2_frint_vector_vg4_multi<"frinta", 0b1100000>; |
| defm FRINTM_2Z2Z: sme2_frint_vector_vg2_multi<"frintm", 0b10100>; |
| defm FRINTM_4Z4Z: sme2_frint_vector_vg4_multi<"frintm", 0b1010000>; |
| defm FRINTN_2Z2Z: sme2_frint_vector_vg2_multi<"frintn", 0b10000>; |
| defm FRINTN_4Z4Z: sme2_frint_vector_vg4_multi<"frintn", 0b1000000>; |
| defm FRINTP_2Z2Z: sme2_frint_vector_vg2_multi<"frintp", 0b10010>; |
| defm FRINTP_4Z4Z: sme2_frint_vector_vg4_multi<"frintp", 0b1001000>; |
| |
| defm MOVA_MXI2Z : sme2_mova_vec_to_tile_vg2_multi<"mova">; |
| defm MOVA_MXI4Z : sme2_mova_vec_to_tile_vg4_multi<"mova">; |
| defm MOVA_2ZMXI : sme2_mova_tile_to_vec_vg2_multi<"mova">; |
| defm MOVA_4ZMXI : sme2_mova_tile_to_vec_vg4_multi<"mova">; |
| |
| defm MOVA_VG2_MXI2Z : sme2_mova_vec_to_array_vg2_multi<"mova">; |
| defm MOVA_VG4_MXI4Z : sme2_mova_vec_to_array_vg4_multi<"mova">; |
| defm MOVA_VG2_2ZMXI : sme2_mova_array_to_vec_vg2_multi<0b000, "mova">; |
| defm MOVA_VG4_4ZMXI : sme2_mova_array_to_vec_vg4_multi<0b1000, "mova">; |
| |
| defm SQRSHR_VG2_Z2ZI : sme2_sat_shift_vector_vg2<"sqrshr", 0b0, 0b0>; |
| defm SQRSHR_VG4_Z4ZI : sme2_sat_shift_vector_vg4<"sqrshr", 0b000>; |
| |
| defm UQRSHR_VG2_Z2ZI : sme2_sat_shift_vector_vg2<"uqrshr", 0b0, 0b1>; |
| defm UQRSHR_VG4_Z4ZI : sme2_sat_shift_vector_vg4<"uqrshr", 0b001>; |
| |
| defm SQRSHRU_VG2_Z2ZI : sme2_sat_shift_vector_vg2<"sqrshru", 0b1, 0b0>; |
| defm SQRSHRU_VG4_Z4ZI : sme2_sat_shift_vector_vg4<"sqrshru", 0b010>; |
| |
| defm SQRSHRN_VG4_Z4ZI : sme2_sat_shift_vector_vg4<"sqrshrn", 0b100>; |
| defm UQRSHRN_VG4_Z4ZI : sme2_sat_shift_vector_vg4<"uqrshrn", 0b101>; |
| defm SQRSHRUN_VG4_Z4ZI : sme2_sat_shift_vector_vg4<"sqrshrun", 0b110>; |
| |
| defm SEL_VG2_2ZP2Z2Z: sme2_sel_vector_vg2<"sel">; |
| defm SEL_VG4_4ZP4Z4Z: sme2_sel_vector_vg4<"sel">; |
| |
| def LD1B_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b00, 0b0, ZZ_b_strided, GPR64shifted8, "ld1b">; |
| def LD1B_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b00, 0b0, ZZZZ_b_strided, GPR64shifted8, "ld1b">; |
| defm LD1B_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b00, 0b0, ZZ_b_strided, simm4s2, "ld1b">; |
| defm LD1B_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b00, 0b0, ZZZZ_b_strided, simm4s4, "ld1b">; |
| def LD1H_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b01, 0b0, ZZ_h_strided, GPR64shifted16, "ld1h">; |
| def LD1H_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b01, 0b0, ZZZZ_h_strided, GPR64shifted16, "ld1h">; |
| defm LD1H_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b01, 0b0, ZZ_h_strided, simm4s2, "ld1h">; |
| defm LD1H_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b01, 0b0, ZZZZ_h_strided, simm4s4, "ld1h">; |
| def LD1W_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b10, 0b0, ZZ_s_strided, GPR64shifted32, "ld1w">; |
| def LD1W_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b10, 0b0, ZZZZ_s_strided, GPR64shifted32, "ld1w">; |
| defm LD1W_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b10, 0b0, ZZ_s_strided, simm4s2, "ld1w">; |
| defm LD1W_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b10, 0b0, ZZZZ_s_strided, simm4s4, "ld1w">; |
| def LD1D_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b11, 0b0, ZZ_d_strided, GPR64shifted64, "ld1d">; |
| def LD1D_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b11, 0b0, ZZZZ_d_strided, GPR64shifted64, "ld1d">; |
| defm LD1D_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b11, 0b0, ZZ_d_strided, simm4s2, "ld1d">; |
| defm LD1D_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b11, 0b0, ZZZZ_d_strided, simm4s4, "ld1d">; |
| |
| def LDNT1B_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b00, 0b1, ZZ_b_strided, GPR64shifted8, "ldnt1b">; |
| def LDNT1B_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b00, 0b1, ZZZZ_b_strided, GPR64shifted8, "ldnt1b">; |
| defm LDNT1B_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b00, 0b1, ZZ_b_strided, simm4s2, "ldnt1b">; |
| defm LDNT1B_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b00, 0b1, ZZZZ_b_strided, simm4s4, "ldnt1b">; |
| def LDNT1H_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b01, 0b1, ZZ_h_strided, GPR64shifted16, "ldnt1h">; |
| def LDNT1H_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b01, 0b1, ZZZZ_h_strided, GPR64shifted16, "ldnt1h">; |
| defm LDNT1H_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b01, 0b1, ZZ_h_strided, simm4s2, "ldnt1h">; |
| defm LDNT1H_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b01, 0b1, ZZZZ_h_strided, simm4s4, "ldnt1h">; |
| def LDNT1W_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b10, 0b1, ZZ_s_strided, GPR64shifted32, "ldnt1w">; |
| def LDNT1W_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b10, 0b1, ZZZZ_s_strided, GPR64shifted32, "ldnt1w">; |
| defm LDNT1W_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b10, 0b1, ZZ_s_strided, simm4s2, "ldnt1w">; |
| defm LDNT1W_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b10, 0b1, ZZZZ_s_strided, simm4s4, "ldnt1w">; |
| def LDNT1D_VG2_M2ZPXX : sme2_ld_vector_vg2_multi_scalar_scalar<0b11, 0b1, ZZ_d_strided, GPR64shifted64, "ldnt1d">; |
| def LDNT1D_VG4_M4ZPXX : sme2_ld_vector_vg4_multi_scalar_scalar<0b11, 0b1, ZZZZ_d_strided, GPR64shifted64, "ldnt1d">; |
| defm LDNT1D_VG2_M2ZPXI : sme2_ld_vector_vg2_multi_scalar_immediate<0b11, 0b1, ZZ_d_strided, simm4s2, "ldnt1d">; |
| defm LDNT1D_VG4_M4ZPXI : sme2_ld_vector_vg4_multi_scalar_immediate<0b11, 0b1, ZZZZ_d_strided, simm4s4, "ldnt1d">; |
| |
| def ST1B_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b00, 0b0, ZZ_b_strided, GPR64shifted8, "st1b">; |
| def ST1B_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b00, 0b0, ZZZZ_b_strided, GPR64shifted8, "st1b">; |
| defm ST1B_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b00, 0b0, ZZ_b_strided, simm4s2, "st1b">; |
| defm ST1B_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b00, 0b0, ZZZZ_b_strided, simm4s4, "st1b">; |
| def ST1H_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b01, 0b0, ZZ_h_strided, GPR64shifted16, "st1h">; |
| def ST1H_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b01, 0b0, ZZZZ_h_strided, GPR64shifted16, "st1h">; |
| defm ST1H_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b01, 0b0, ZZ_h_strided, simm4s2, "st1h">; |
| defm ST1H_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b01, 0b0, ZZZZ_h_strided, simm4s4, "st1h">; |
| def ST1W_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b10, 0b0, ZZ_s_strided, GPR64shifted32, "st1w">; |
| def ST1W_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b10, 0b0, ZZZZ_s_strided, GPR64shifted32, "st1w">; |
| defm ST1W_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b10, 0b0, ZZ_s_strided, simm4s2, "st1w">; |
| defm ST1W_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b10, 0b0, ZZZZ_s_strided, simm4s4, "st1w">; |
| def ST1D_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b11, 0b0, ZZ_d_strided, GPR64shifted64, "st1d">; |
| def ST1D_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b11, 0b0, ZZZZ_d_strided, GPR64shifted64, "st1d">; |
| defm ST1D_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b11, 0b0, ZZ_d_strided, simm4s2, "st1d">; |
| defm ST1D_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b11, 0b0, ZZZZ_d_strided, simm4s4, "st1d">; |
| |
| def STNT1B_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b00, 0b1, ZZ_b_strided, GPR64shifted8, "stnt1b">; |
| def STNT1B_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b00, 0b1, ZZZZ_b_strided, GPR64shifted8, "stnt1b">; |
| defm STNT1B_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b00, 0b1, ZZ_b_strided, simm4s2, "stnt1b">; |
| defm STNT1B_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b00, 0b1, ZZZZ_b_strided, simm4s4, "stnt1b">; |
| def STNT1H_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b01, 0b1, ZZ_h_strided, GPR64shifted16, "stnt1h">; |
| def STNT1H_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b01, 0b1, ZZZZ_h_strided, GPR64shifted16, "stnt1h">; |
| defm STNT1H_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b01, 0b1, ZZ_h_strided, simm4s2, "stnt1h">; |
| defm STNT1H_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b01, 0b1, ZZZZ_h_strided, simm4s4, "stnt1h">; |
| def STNT1W_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b10, 0b1, ZZ_s_strided, GPR64shifted32, "stnt1w">; |
| def STNT1W_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b10, 0b1, ZZZZ_s_strided, GPR64shifted32, "stnt1w">; |
| defm STNT1W_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b10, 0b1, ZZ_s_strided, simm4s2, "stnt1w">; |
| defm STNT1W_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b10, 0b1, ZZZZ_s_strided, simm4s4, "stnt1w">; |
| def STNT1D_VG2_M2ZPXX : sme2_st_vector_vg2_multi_scalar_scalar<0b11, 0b1, ZZ_d_strided, GPR64shifted64, "stnt1d">; |
| def STNT1D_VG4_M4ZPXX : sme2_st_vector_vg4_multi_scalar_scalar<0b11, 0b1, ZZZZ_d_strided, GPR64shifted64, "stnt1d">; |
| defm STNT1D_VG2_M2ZPXI : sme2_st_vector_vg2_multi_scalar_immediate<0b11, 0b1, ZZ_d_strided, simm4s2, "stnt1d">; |
| defm STNT1D_VG4_M4ZPXI : sme2_st_vector_vg4_multi_scalar_immediate<0b11, 0b1, ZZZZ_d_strided, simm4s4, "stnt1d">; |
| } |
| |
| let Predicates = [HasSME2, HasSMEI16I64] in { |
| defm ADD_VG2_M2ZZ_D : sme2_dot_mla_add_sub_array_vg2_single<"add", 0b1011010, MatrixOp64, ZZ_d, ZPR4b64, nxv2i64, int_aarch64_sme_add_write_single_za_vg1x2>; |
| defm ADD_VG4_M4ZZ_D : sme2_dot_mla_add_sub_array_vg4_single<"add", 0b1111010, MatrixOp64, ZZZZ_d, ZPR4b64, nxv2i64, int_aarch64_sme_add_write_single_za_vg1x4>; |
| defm ADD_VG2_M2Z2Z_D : sme2_dot_mla_add_sub_array_vg2_multi<"add", 0b111010, MatrixOp64, ZZ_d_mul_r, nxv2i64, int_aarch64_sme_add_write_za_vg1x2>; |
| defm ADD_VG4_M4Z4Z_D : sme2_dot_mla_add_sub_array_vg4_multi<"add", 0b111010, MatrixOp64, ZZZZ_d_mul_r, nxv2i64, int_aarch64_sme_add_write_za_vg1x4>; |
| |
| defm SUB_VG2_M2ZZ_D : sme2_dot_mla_add_sub_array_vg2_single<"sub", 0b1011011, MatrixOp64, ZZ_d, ZPR4b64, nxv2i64, int_aarch64_sme_sub_write_single_za_vg1x2>; |
| defm SUB_VG4_M4ZZ_D : sme2_dot_mla_add_sub_array_vg4_single<"sub", 0b1111011, MatrixOp64, ZZZZ_d, ZPR4b64, nxv2i64, int_aarch64_sme_sub_write_single_za_vg1x4>; |
| defm SUB_VG2_M2Z2Z_D : sme2_dot_mla_add_sub_array_vg2_multi<"sub", 0b111011, MatrixOp64, ZZ_d_mul_r, nxv2i64, int_aarch64_sme_sub_write_za_vg1x2>; |
| defm SUB_VG4_M4Z4Z_D : sme2_dot_mla_add_sub_array_vg4_multi<"sub", 0b111011, MatrixOp64, ZZZZ_d_mul_r, nxv2i64, int_aarch64_sme_sub_write_za_vg1x4>; |
| |
| defm ADD_VG2_M2Z_D : sme2_multivec_accum_add_sub_vg2<"add", 0b1010, MatrixOp64, ZZ_d_mul_r>; |
| defm ADD_VG4_M4Z_D : sme2_multivec_accum_add_sub_vg4<"add", 0b1010, MatrixOp64, ZZZZ_d_mul_r>; |
| |
| defm SUB_VG2_M2Z_D : sme2_multivec_accum_add_sub_vg2<"sub", 0b1011, MatrixOp64, ZZ_d_mul_r>; |
| defm SUB_VG4_M4Z_D : sme2_multivec_accum_add_sub_vg4<"sub", 0b1011, MatrixOp64, ZZZZ_d_mul_r>; |
| |
| defm SDOT_VG2_M2ZZI_HtoD : sme2_multi_vec_array_vg2_index_64b<"sdot", 0b01, ZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm SDOT_VG4_M4ZZI_HtoD : sme2_multi_vec_array_vg4_index_64b<"sdot", 0b001, ZZZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm SDOT_VG2_M2ZZ_HtoD : sme2_dot_mla_add_sub_array_vg24_single<"sdot", 0b1010100, MatrixOp64, ZZ_h, ZPR4b16>; |
| defm SDOT_VG4_M4ZZ_HtoD : sme2_dot_mla_add_sub_array_vg24_single<"sdot", 0b1110100, MatrixOp64, ZZZZ_h, ZPR4b16>; |
| defm SDOT_VG2_M2Z2Z_HtoD : sme2_dot_mla_add_sub_array_vg2_multi<"sdot", 0b110100, MatrixOp64, ZZ_h_mul_r, nxv8i16, null_frag>; |
| defm SDOT_VG4_M4Z4Z_HtoD : sme2_dot_mla_add_sub_array_vg4_multi<"sdot", 0b110100, MatrixOp64, ZZZZ_h_mul_r, nxv8i16, null_frag>; |
| |
| defm SVDOT_VG4_M4ZZI_HtoD : sme2_multi_vec_array_vg4_index_64b<"svdot", 0b101, ZZZZ_h_mul_r, ZPR4b16, nxv8i16, int_aarch64_sme_svdot_lane_za64_vg1x4>; |
| |
| defm UDOT_VG2_M2ZZI_HtoD : sme2_multi_vec_array_vg2_index_64b<"udot", 0b11, ZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm UDOT_VG4_M4ZZI_HtoD : sme2_multi_vec_array_vg4_index_64b<"udot", 0b011, ZZZZ_h_mul_r, ZPR4b16, nxv8i16, null_frag>; |
| defm UDOT_VG2_M2ZZ_HtoD : sme2_dot_mla_add_sub_array_vg24_single<"udot", 0b1010110, MatrixOp64, ZZ_h, ZPR4b16>; |
| defm UDOT_VG4_M4ZZ_HtoD : sme2_dot_mla_add_sub_array_vg24_single<"udot", 0b1110110, MatrixOp64, ZZZZ_h, ZPR4b16>; |
| defm UDOT_VG2_M2Z2Z_HtoD : sme2_dot_mla_add_sub_array_vg2_multi<"udot", 0b110110, MatrixOp64, ZZ_h_mul_r, nxv8i16, null_frag>; |
| defm UDOT_VG4_M4Z4Z_HtoD : sme2_dot_mla_add_sub_array_vg4_multi<"udot", 0b110110, MatrixOp64, ZZZZ_h_mul_r, nxv8i16, null_frag>; |
| |
| defm UVDOT_VG4_M4ZZI_HtoD : sme2_multi_vec_array_vg4_index_64b<"uvdot", 0b111, ZZZZ_h_mul_r, ZPR4b16, nxv8i16, int_aarch64_sme_uvdot_lane_za64_vg1x4>; |
| |
| def SMLALL_MZZI_HtoD : sme2_mla_ll_array_index_64b<"smlall", 0b00>; |
| defm SMLALL_VG2_M2ZZI_HtoD : sme2_mla_ll_array_vg2_index_64b<"smlall", 0b00>; |
| defm SMLALL_VG4_M4ZZI_HtoD : sme2_mla_ll_array_vg4_index_64b<"smlall", 0b00>; |
| def SMLALL_MZZ_HtoD : sme2_mla_ll_array_single<"smlall", 0b1000, MatrixOp64, ZPR16, ZPR4b16>; |
| defm SMLALL_VG2_M2ZZ_HtoD : sme2_mla_ll_array_vg24_single<"smlall", 0b10000, MatrixOp64, ZZ_h, ZPR4b16>; |
| defm SMLALL_VG4_M4ZZ_HtoD : sme2_mla_ll_array_vg24_single<"smlall", 0b11000, MatrixOp64, ZZZZ_h, ZPR4b16>; |
| defm SMLALL_VG2_M2Z2Z_HtoD : sme2_mla_ll_array_vg2_multi<"smlall", 0b1000, MatrixOp64, ZZ_h_mul_r>; |
| defm SMLALL_VG4_M4Z4Z_HtoD : sme2_mla_ll_array_vg4_multi<"smlall", 0b1000, MatrixOp64, ZZZZ_h_mul_r>; |
| |
| def SMLSLL_MZZI_HtoD : sme2_mla_ll_array_index_64b<"smlsll", 0b01>; |
| defm SMLSLL_VG2_M2ZZI_HtoD : sme2_mla_ll_array_vg2_index_64b<"smlsll", 0b01>; |
| defm SMLSLL_VG4_M4ZZI_HtoD : sme2_mla_ll_array_vg4_index_64b<"smlsll", 0b01>; |
| def SMLSLL_MZZ_HtoD : sme2_mla_ll_array_single<"smlsll", 0b1010, MatrixOp64, ZPR16, ZPR4b16>; |
| defm SMLSLL_VG2_M2ZZ_HtoD : sme2_mla_ll_array_vg24_single<"smlsll", 0b10010, MatrixOp64, ZZ_h, ZPR4b16>; |
| defm SMLSLL_VG4_M4ZZ_HtoD : sme2_mla_ll_array_vg24_single<"smlsll", 0b11010, MatrixOp64, ZZZZ_h, ZPR4b16>; |
| defm SMLSLL_VG2_M2Z2Z_HtoD : sme2_mla_ll_array_vg2_multi<"smlsll", 0b1010, MatrixOp64, ZZ_h_mul_r>; |
| defm SMLSLL_VG4_M4Z4Z_HtoD : sme2_mla_ll_array_vg4_multi<"smlsll", 0b1010, MatrixOp64, ZZZZ_h_mul_r>; |
| |
| def UMLALL_MZZI_HtoD : sme2_mla_ll_array_index_64b<"umlall", 0b10>; |
| defm UMLALL_VG2_M2ZZI_HtoD : sme2_mla_ll_array_vg2_index_64b<"umlall", 0b10>; |
| defm UMLALL_VG4_M4ZZI_HtoD : sme2_mla_ll_array_vg4_index_64b<"umlall", 0b10>; |
| def UMLALL_MZZ_HtoD : sme2_mla_ll_array_single<"umlall", 0b1100, MatrixOp64, ZPR16, ZPR4b16>; |
| defm UMLALL_VG2_M2ZZ_HtoD : sme2_mla_ll_array_vg24_single<"umlall", 0b10100, MatrixOp64, ZZ_h, ZPR4b16>; |
| defm UMLALL_VG4_M4ZZ_HtoD : sme2_mla_ll_array_vg24_single<"umlall", 0b11100, MatrixOp64, ZZZZ_h, ZPR4b16>; |
| defm UMLALL_VG2_M2Z2Z_HtoD : sme2_mla_ll_array_vg2_multi<"umlall", 0b1100, MatrixOp64, ZZ_h_mul_r>; |
| defm UMLALL_VG4_M4Z4Z_HtoD : sme2_mla_ll_array_vg4_multi<"umlall", 0b1100, MatrixOp64, ZZZZ_h_mul_r>; |
| |
| def UMLSLL_MZZI_HtoD : sme2_mla_ll_array_index_64b<"umlsll", 0b11>; |
| defm UMLSLL_VG2_M2ZZI_HtoD : sme2_mla_ll_array_vg2_index_64b<"umlsll", 0b11>; |
| defm UMLSLL_VG4_M4ZZI_HtoD : sme2_mla_ll_array_vg4_index_64b<"umlsll", 0b11>; |
| def UMLSLL_MZZ_HtoD : sme2_mla_ll_array_single<"umlsll", 0b1110, MatrixOp64, ZPR16, ZPR4b16>; |
| defm UMLSLL_VG2_M2ZZ_HtoD : sme2_mla_ll_array_vg24_single<"umlsll", 0b10110, MatrixOp64, ZZ_h, ZPR4b16>; |
| defm UMLSLL_VG4_M4ZZ_HtoD : sme2_mla_ll_array_vg24_single<"umlsll", 0b11110, MatrixOp64, ZZZZ_h, ZPR4b16>; |
| defm UMLSLL_VG2_M2Z2Z_HtoD : sme2_mla_ll_array_vg2_multi<"umlsll", 0b1110, MatrixOp64, ZZ_h_mul_r>; |
| defm UMLSLL_VG4_M4Z4Z_HtoD : sme2_mla_ll_array_vg4_multi<"umlsll", 0b1110, MatrixOp64, ZZZZ_h_mul_r>; |
| } |
| |
| let Predicates = [HasSME2, HasSMEF64F64] in { |
| defm FMLA_VG2_M2ZZI_D : sme2_multi_vec_array_vg2_index_64b<"fmla", 0b00, ZZ_d_mul_r, ZPR4b64, nxv2f64, int_aarch64_sme_fmla_lane_vg1x2>; |
| defm FMLA_VG4_M4ZZI_D : sme2_multi_vec_array_vg4_index_64b<"fmla", 0b000, ZZZZ_d_mul_r, ZPR4b64, nxv2f64, int_aarch64_sme_fmla_lane_vg1x4>; |
| defm FMLA_VG2_M2ZZ_D : sme2_dot_mla_add_sub_array_vg2_single<"fmla", 0b1011000, MatrixOp64, ZZ_d, ZPR4b64, nxv2f64, int_aarch64_sme_fmla_single_vg1x2>; |
| defm FMLA_VG4_M4ZZ_D : sme2_dot_mla_add_sub_array_vg4_single<"fmla", 0b1111000, MatrixOp64, ZZZZ_d, ZPR4b64, nxv2f64, int_aarch64_sme_fmla_single_vg1x4>; |
| defm FMLA_VG2_M2Z2Z_D : sme2_dot_mla_add_sub_array_vg2_multi<"fmla", 0b111000, MatrixOp64, ZZ_d_mul_r, nxv2f64, int_aarch64_sme_fmla_vg1x2>; |
| defm FMLA_VG4_M4Z4Z_D : sme2_dot_mla_add_sub_array_vg4_multi<"fmla", 0b111000, MatrixOp64, ZZZZ_d_mul_r, nxv2f64, int_aarch64_sme_fmla_vg1x4>; |
| |
| defm FMLS_VG2_M2ZZI_D : sme2_multi_vec_array_vg2_index_64b<"fmls", 0b10, ZZ_d_mul_r, ZPR4b64, nxv2f64, int_aarch64_sme_fmls_lane_vg1x2>; |
| defm FMLS_VG4_M4ZZI_D : sme2_multi_vec_array_vg4_index_64b<"fmls", 0b010, ZZZZ_d_mul_r, ZPR4b64, nxv2f64, int_aarch64_sme_fmls_lane_vg1x4>; |
| defm FMLS_VG2_M2ZZ_D : sme2_dot_mla_add_sub_array_vg2_single<"fmls", 0b1011001, MatrixOp64, ZZ_d, ZPR4b64, nxv2f64, int_aarch64_sme_fmls_single_vg1x2>; |
| defm FMLS_VG4_M4ZZ_D : sme2_dot_mla_add_sub_array_vg4_single<"fmls", 0b1111001, MatrixOp64, ZZZZ_d, ZPR4b64, nxv2f64, int_aarch64_sme_fmls_single_vg1x4>; |
| defm FMLS_VG2_M2Z2Z_D : sme2_dot_mla_add_sub_array_vg2_multi<"fmls", 0b111001, MatrixOp64, ZZ_d_mul_r, nxv2f64, int_aarch64_sme_fmls_vg1x2>; |
| defm FMLS_VG4_M4Z4Z_D : sme2_dot_mla_add_sub_array_vg4_multi<"fmls", 0b111001, MatrixOp64, ZZZZ_d_mul_r, nxv2f64, int_aarch64_sme_fmls_vg1x4>; |
| |
| defm FADD_VG2_M2Z_D : sme2_multivec_accum_add_sub_vg2<"fadd", 0b1000, MatrixOp64, ZZ_d_mul_r>; |
| defm FADD_VG4_M4Z_D : sme2_multivec_accum_add_sub_vg4<"fadd", 0b1000, MatrixOp64, ZZZZ_d_mul_r>; |
| |
| defm FSUB_VG2_M2Z_D : sme2_multivec_accum_add_sub_vg2<"fsub", 0b1001, MatrixOp64, ZZ_d_mul_r>; |
| defm FSUB_VG4_M4Z_D : sme2_multivec_accum_add_sub_vg4<"fsub", 0b1001, MatrixOp64, ZZZZ_d_mul_r>; |
| } |
| |
| let Predicates = [HasSME2p1] in { |
| defm MOVAZ_ZMI : sme2p1_movaz_tile_to_vec<"movaz">; |
| defm MOVAZ_2ZMI : sme2p1_movaz_tile_to_vec_vg2<"movaz">; |
| defm MOVAZ_4ZMI : sme2p1_movaz_tile_to_vec_vg4<"movaz">; |
| defm MOVAZ_VG2_2ZM : sme2_mova_array_to_vec_vg2_multi<0b010, "movaz">; |
| defm MOVAZ_VG4_4ZM : sme2_mova_array_to_vec_vg4_multi<0b1100, "movaz">; |
| |
| defm ZERO_MXI : sme2p1_zero_matrix<"zero">; |
| |
| defm LUTI2_S_2ZTZI : sme2p1_luti2_vector_vg2_index<"luti2">; |
| defm LUTI2_S_4ZTZI : sme2p1_luti2_vector_vg4_index<"luti2">; |
| |
| defm LUTI4_S_2ZTZI : sme2p1_luti4_vector_vg2_index<"luti4">; |
| defm LUTI4_S_4ZTZI : sme2p1_luti4_vector_vg4_index<"luti4">; |
| } |
| |
| let Predicates = [HasSME2p1, HasSMEF16F16] in { |
| defm FADD_VG2_M2Z_H : sme2_multivec_accum_add_sub_vg2<"fadd", 0b0100, MatrixOp16, ZZ_h_mul_r>; |
| defm FADD_VG4_M4Z_H : sme2_multivec_accum_add_sub_vg4<"fadd", 0b0100, MatrixOp16, ZZZZ_h_mul_r>; |
| defm FSUB_VG2_M2Z_H : sme2_multivec_accum_add_sub_vg2<"fsub", 0b0101, MatrixOp16, ZZ_h_mul_r>; |
| defm FSUB_VG4_M4Z_H : sme2_multivec_accum_add_sub_vg4<"fsub", 0b0101, MatrixOp16, ZZZZ_h_mul_r>; |
| |
| defm FMLA_VG2_M2ZZI_H : sme2p1_multi_vec_array_vg2_index_16b<"fmla", 0b00>; |
| defm FMLA_VG4_M4ZZI_H : sme2p1_multi_vec_array_vg4_index_16b<"fmla", 0b00>; |
| defm FMLA_VG2_M2ZZ_H : sme2_dot_mla_add_sub_array_vg24_single<"fmla", 0b0011100, MatrixOp16, ZZ_h, ZPR4b16>; |
| defm FMLA_VG4_M4ZZ_H : sme2_dot_mla_add_sub_array_vg24_single<"fmla", 0b0111100, MatrixOp16, ZZZZ_h, ZPR4b16>; |
| defm FMLA_VG2_M2Z4Z_H : sme2_dot_mla_add_sub_array_vg2_multi<"fmla", 0b010001, MatrixOp16, ZZ_h_mul_r, nxv8f16, null_frag>; |
| defm FMLA_VG4_M4Z4Z_H : sme2_dot_mla_add_sub_array_vg4_multi<"fmla", 0b010001, MatrixOp16, ZZZZ_h_mul_r, nxv8f16, null_frag>; |
| |
| defm FMLS_VG2_M2ZZI_H : sme2p1_multi_vec_array_vg2_index_16b<"fmls", 0b01>; |
| defm FMLS_VG4_M4ZZI_H : sme2p1_multi_vec_array_vg4_index_16b<"fmls", 0b01>; |
| defm FMLS_VG2_M2ZZ_H : sme2_dot_mla_add_sub_array_vg24_single<"fmls", 0b0011101, MatrixOp16, ZZ_h, ZPR4b16>; |
| defm FMLS_VG4_M4ZZ_H : sme2_dot_mla_add_sub_array_vg24_single<"fmls", 0b0111101, MatrixOp16, ZZZZ_h, ZPR4b16>; |
| defm FMLS_VG2_M2Z2Z_H : sme2_dot_mla_add_sub_array_vg2_multi<"fmls", 0b010011, MatrixOp16, ZZ_h_mul_r, nxv8f16, null_frag>; |
| defm FMLS_VG4_M4Z2Z_H : sme2_dot_mla_add_sub_array_vg4_multi<"fmls", 0b010011, MatrixOp16, ZZZZ_h_mul_r, nxv8f16, null_frag>; |
| |
| defm FCVT_2ZZ_H : sme2p1_fp_cvt_vector_vg2_single<"fcvt", 0b0>; |
| defm FCVTL_2ZZ_H : sme2p1_fp_cvt_vector_vg2_single<"fcvtl", 0b1>; |
| |
| defm FMOPA_MPPZZ_H : sme2p1_fmop_tile_fp16<"fmopa", 0b0, 0b0>; |
| defm FMOPS_MPPZZ_H : sme2p1_fmop_tile_fp16<"fmops", 0b0, 0b1>; |
| } |
| |
| let Predicates = [HasSME2p1, HasB16B16] in { |
| defm BFADD_VG2_M2Z_H : sme2_multivec_accum_add_sub_vg2<"bfadd", 0b1100, MatrixOp16, ZZ_h_mul_r>; |
| defm BFADD_VG4_M4Z_H : sme2_multivec_accum_add_sub_vg4<"bfadd", 0b1100, MatrixOp16, ZZZZ_h_mul_r>; |
| defm BFSUB_VG2_M2Z_H : sme2_multivec_accum_add_sub_vg2<"bfsub", 0b1101, MatrixOp16, ZZ_h_mul_r>; |
| defm BFSUB_VG4_M4Z_H : sme2_multivec_accum_add_sub_vg4<"bfsub", 0b1101, MatrixOp16, ZZZZ_h_mul_r>; |
| |
| defm BFMLA_VG2_M2ZZI : sme2p1_multi_vec_array_vg2_index_16b<"bfmla", 0b10>; |
| defm BFMLA_VG4_M4ZZI : sme2p1_multi_vec_array_vg4_index_16b<"bfmla", 0b10>; |
| defm BFMLA_VG2_M2ZZ : sme2_dot_mla_add_sub_array_vg24_single<"bfmla", 0b1011100, MatrixOp16, ZZ_h, ZPR4b16>; |
| defm BFMLA_VG4_M4ZZ : sme2_dot_mla_add_sub_array_vg24_single<"bfmla", 0b1111100, MatrixOp16, ZZZZ_h, ZPR4b16>; |
| defm BFMLA_VG2_M2Z2Z : sme2_dot_mla_add_sub_array_vg2_multi<"bfmla", 0b110001, MatrixOp16, ZZ_h_mul_r, nxv8bf16, null_frag>; |
| defm BFMLA_VG4_M4Z4Z : sme2_dot_mla_add_sub_array_vg4_multi<"bfmla", 0b110001, MatrixOp16, ZZZZ_h_mul_r, nxv8bf16, null_frag>; |
| |
| defm BFMLS_VG2_M2ZZI : sme2p1_multi_vec_array_vg2_index_16b<"bfmls", 0b11>; |
| defm BFMLS_VG4_M4ZZI : sme2p1_multi_vec_array_vg4_index_16b<"bfmls", 0b11>; |
| defm BFMLS_VG2_M2ZZ : sme2_dot_mla_add_sub_array_vg24_single<"bfmls", 0b1011101, MatrixOp16, ZZ_h, ZPR4b16>; |
| defm BFMLS_VG4_M4ZZ : sme2_dot_mla_add_sub_array_vg24_single<"bfmls", 0b1111101, MatrixOp16, ZZZZ_h, ZPR4b16>; |
| defm BFMLS_VG2_M2Z2Z : sme2_dot_mla_add_sub_array_vg2_multi<"bfmls", 0b110011, MatrixOp16, ZZ_h_mul_r, nxv8bf16, null_frag>; |
| defm BFMLS_VG4_M4Z4Z : sme2_dot_mla_add_sub_array_vg4_multi<"bfmls", 0b110011, MatrixOp16, ZZZZ_h_mul_r, nxv8bf16, null_frag>; |
| |
| |
| defm BFMAX_VG2_2ZZ : sme2p1_bf_max_min_vector_vg2_single<"bfmax", 0b0010000>; |
| defm BFMAX_VG4_4ZZ : sme2p1_bf_max_min_vector_vg4_single<"bfmax", 0b0010000>; |
| defm BFMAX_VG2_2Z2Z : sme2p1_bf_max_min_vector_vg2_multi<"bfmax", 0b0010000>; |
| defm BFMAX_VG4_4Z2Z : sme2p1_bf_max_min_vector_vg4_multi<"bfmax", 0b0010000>; |
| |
| defm BFMIN_VG2_2ZZ : sme2p1_bf_max_min_vector_vg2_single<"bfmin", 0b0010001>; |
| defm BFMIN_VG4_4ZZ : sme2p1_bf_max_min_vector_vg4_single<"bfmin", 0b0010001>; |
| defm BFMIN_VG2_2Z2Z : sme2p1_bf_max_min_vector_vg2_multi<"bfmin", 0b0010001>; |
| defm BFMIN_VG4_4Z2Z : sme2p1_bf_max_min_vector_vg4_multi<"bfmin", 0b0010001>; |
| |
| defm BFMAXNM_VG2_2ZZ : sme2p1_bf_max_min_vector_vg2_single<"bfmaxnm", 0b0010010>; |
| defm BFMAXNM_VG4_4ZZ : sme2p1_bf_max_min_vector_vg4_single<"bfmaxnm", 0b0010010>; |
| defm BFMAXNM_VG2_2Z2Z : sme2p1_bf_max_min_vector_vg2_multi<"bfmaxnm", 0b0010010>; |
| defm BFMAXNM_VG4_4Z2Z : sme2p1_bf_max_min_vector_vg4_multi<"bfmaxnm", 0b0010010>; |
| |
| defm BFMINNM_VG2_2ZZ : sme2p1_bf_max_min_vector_vg2_single<"bfminnm", 0b0010011>; |
| defm BFMINNM_VG4_4ZZ : sme2p1_bf_max_min_vector_vg4_single<"bfminnm", 0b0010011>; |
| defm BFMINNM_VG2_2Z2Z : sme2p1_bf_max_min_vector_vg2_multi<"bfminnm", 0b0010011>; |
| defm BFMINNM_VG4_4Z2Z : sme2p1_bf_max_min_vector_vg4_multi<"bfminnm", 0b0010011>; |
| |
| defm BFCLAMP_VG2_2ZZZ: sme2p1_bfclamp_vector_vg2_multi<"bfclamp">; |
| defm BFCLAMP_VG4_4ZZZ: sme2p1_bfclamp_vector_vg4_multi<"bfclamp">; |
| |
| defm BFMOPA_MPPZZ_H : sme2p1_fmop_tile_fp16<"bfmopa", 0b1, 0b0>; |
| defm BFMOPS_MPPZZ_H : sme2p1_fmop_tile_fp16<"bfmops", 0b1, 0b1>; |
| } |
