diff options
Diffstat (limited to 'lib/Target/ARM/ARMInstrNEON.td')
| -rw-r--r-- | lib/Target/ARM/ARMInstrNEON.td | 217 |
1 files changed, 119 insertions, 98 deletions
diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td index ed9d31d80f36..d5ce2b8468df 100644 --- a/lib/Target/ARM/ARMInstrNEON.td +++ b/lib/Target/ARM/ARMInstrNEON.td @@ -1621,12 +1621,13 @@ multiclass N3VNInt_HSD<bit op24, bit op23, bits<4> op11_8, bit op4, // First with only element sizes of 16 and 32 bits: multiclass N3VLInt_HS<bit op24, bit op23, bits<4> op11_8, bit op4, - InstrItinClass itin, string OpcodeStr, string Dt, + InstrItinClass itin16, InstrItinClass itin32, + string OpcodeStr, string Dt, Intrinsic IntOp, bit Commutable = 0> { - def v4i32 : N3VLInt<op24, op23, 0b01, op11_8, op4, itin, + def v4i32 : N3VLInt<op24, op23, 0b01, op11_8, op4, itin16, OpcodeStr, !strconcat(Dt, "16"), v4i32, v4i16, IntOp, Commutable>; - def v2i64 : N3VLInt<op24, op23, 0b10, op11_8, op4, itin, + def v2i64 : N3VLInt<op24, op23, 0b10, op11_8, op4, itin32, OpcodeStr, !strconcat(Dt, "32"), v2i64, v2i32, IntOp, Commutable>; } @@ -1642,11 +1643,12 @@ multiclass N3VLIntSL_HS<bit op24, bits<4> op11_8, // ....then also with element size of 8 bits: multiclass N3VLInt_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, - InstrItinClass itin, string OpcodeStr, string Dt, + InstrItinClass itin16, InstrItinClass itin32, + string OpcodeStr, string Dt, Intrinsic IntOp, bit Commutable = 0> - : N3VLInt_HS<op24, op23, op11_8, op4, itin, OpcodeStr, Dt, + : N3VLInt_HS<op24, op23, op11_8, op4, itin16, itin32, OpcodeStr, Dt, IntOp, Commutable> { - def v8i16 : N3VLInt<op24, op23, 0b00, op11_8, op4, itin, + def v8i16 : N3VLInt<op24, op23, 0b00, op11_8, op4, itin16, OpcodeStr, !strconcat(Dt, "8"), v8i16, v8i8, IntOp, Commutable>; } @@ -1711,21 +1713,22 @@ multiclass N3VMulOpSL_HS<bits<4> op11_8, // Neon 3-argument intrinsics, // element sizes of 8, 16 and 32 bits: multiclass N3VInt3_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, + InstrItinClass itinD, InstrItinClass itinQ, string OpcodeStr, string Dt, Intrinsic IntOp> { // 64-bit vector types. - def v8i8 : N3VDInt3<op24, op23, 0b00, op11_8, op4, IIC_VMACi16D, + def v8i8 : N3VDInt3<op24, op23, 0b00, op11_8, op4, itinD, OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, IntOp>; - def v4i16 : N3VDInt3<op24, op23, 0b01, op11_8, op4, IIC_VMACi16D, + def v4i16 : N3VDInt3<op24, op23, 0b01, op11_8, op4, itinD, OpcodeStr, !strconcat(Dt, "16"), v4i16, v4i16, IntOp>; - def v2i32 : N3VDInt3<op24, op23, 0b10, op11_8, op4, IIC_VMACi32D, + def v2i32 : N3VDInt3<op24, op23, 0b10, op11_8, op4, itinD, OpcodeStr, !strconcat(Dt, "32"), v2i32, v2i32, IntOp>; // 128-bit vector types. - def v16i8 : N3VQInt3<op24, op23, 0b00, op11_8, op4, IIC_VMACi16Q, + def v16i8 : N3VQInt3<op24, op23, 0b00, op11_8, op4, itinQ, OpcodeStr, !strconcat(Dt, "8"), v16i8, v16i8, IntOp>; - def v8i16 : N3VQInt3<op24, op23, 0b01, op11_8, op4, IIC_VMACi16Q, + def v8i16 : N3VQInt3<op24, op23, 0b01, op11_8, op4, itinQ, OpcodeStr, !strconcat(Dt, "16"), v8i16, v8i16, IntOp>; - def v4i32 : N3VQInt3<op24, op23, 0b10, op11_8, op4, IIC_VMACi32Q, + def v4i32 : N3VQInt3<op24, op23, 0b10, op11_8, op4, itinQ, OpcodeStr, !strconcat(Dt, "32"), v4i32, v4i32, IntOp>; } @@ -1734,10 +1737,11 @@ multiclass N3VInt3_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, // First with only element sizes of 16 and 32 bits: multiclass N3VLInt3_HS<bit op24, bit op23, bits<4> op11_8, bit op4, + InstrItinClass itin16, InstrItinClass itin32, string OpcodeStr, string Dt, Intrinsic IntOp> { - def v4i32 : N3VLInt3<op24, op23, 0b01, op11_8, op4, IIC_VMACi16D, + def v4i32 : N3VLInt3<op24, op23, 0b01, op11_8, op4, itin16, OpcodeStr, !strconcat(Dt, "16"), v4i32, v4i16, IntOp>; - def v2i64 : N3VLInt3<op24, op23, 0b10, op11_8, op4, IIC_VMACi16D, + def v2i64 : N3VLInt3<op24, op23, 0b10, op11_8, op4, itin32, OpcodeStr, !strconcat(Dt, "32"), v2i64, v2i32, IntOp>; } @@ -1751,9 +1755,10 @@ multiclass N3VLInt3SL_HS<bit op24, bits<4> op11_8, // ....then also with element size of 8 bits: multiclass N3VLInt3_QHS<bit op24, bit op23, bits<4> op11_8, bit op4, + InstrItinClass itin16, InstrItinClass itin32, string OpcodeStr, string Dt, Intrinsic IntOp> - : N3VLInt3_HS<op24, op23, op11_8, op4, OpcodeStr, Dt, IntOp> { - def v8i16 : N3VLInt3<op24, op23, 0b00, op11_8, op4, IIC_VMACi16D, + : N3VLInt3_HS<op24, op23, op11_8, op4, itin16, itin32, OpcodeStr, Dt, IntOp> { + def v8i16 : N3VLInt3<op24, op23, 0b00, op11_8, op4, itin16, OpcodeStr, !strconcat(Dt, "8"), v8i16, v8i8, IntOp>; } @@ -2001,10 +2006,10 @@ def VADDfd : N3VD<0, 0, 0b00, 0b1101, 0, IIC_VBIND, "vadd", "f32", def VADDfq : N3VQ<0, 0, 0b00, 0b1101, 0, IIC_VBINQ, "vadd", "f32", v4f32, v4f32, fadd, 1>; // VADDL : Vector Add Long (Q = D + D) -defm VADDLs : N3VLInt_QHS<0,1,0b0000,0, IIC_VSHLiD, "vaddl", "s", - int_arm_neon_vaddls, 1>; -defm VADDLu : N3VLInt_QHS<1,1,0b0000,0, IIC_VSHLiD, "vaddl", "u", - int_arm_neon_vaddlu, 1>; +defm VADDLs : N3VLInt_QHS<0,1,0b0000,0, IIC_VSHLiD, IIC_VSHLiD, + "vaddl", "s", int_arm_neon_vaddls, 1>; +defm VADDLu : N3VLInt_QHS<1,1,0b0000,0, IIC_VSHLiD, IIC_VSHLiD, + "vaddl", "u", int_arm_neon_vaddlu, 1>; // VADDW : Vector Add Wide (Q = Q + D) defm VADDWs : N3VWInt_QHS<0,1,0b0001,0, "vaddw", "s", int_arm_neon_vaddws, 0>; defm VADDWu : N3VWInt_QHS<1,1,0b0001,0, "vaddw", "u", int_arm_neon_vaddwu, 0>; @@ -2118,10 +2123,10 @@ def : Pat<(v4i32 (int_arm_neon_vqrdmulh (v4i32 QPR:$src1), (SubReg_i32_lane imm:$lane)))>; // VMULL : Vector Multiply Long (integer and polynomial) (Q = D * D) -defm VMULLs : N3VLInt_QHS<0,1,0b1100,0, IIC_VMULi16D, "vmull", "s", - int_arm_neon_vmulls, 1>; -defm VMULLu : N3VLInt_QHS<1,1,0b1100,0, IIC_VMULi16D, "vmull", "u", - int_arm_neon_vmullu, 1>; +defm VMULLs : N3VLInt_QHS<0,1,0b1100,0, IIC_VMULi16D, IIC_VMULi32D, + "vmull", "s", int_arm_neon_vmulls, 1>; +defm VMULLu : N3VLInt_QHS<1,1,0b1100,0, IIC_VMULi16D, IIC_VMULi32D, + "vmull", "u", int_arm_neon_vmullu, 1>; def VMULLp : N3VLInt<0, 1, 0b00, 0b1110, 0, IIC_VMULi16D, "vmull", "p8", v8i16, v8i8, int_arm_neon_vmullp, 1>; defm VMULLsls : N3VLIntSL_HS<0, 0b1010, IIC_VMULi16D, "vmull", "s", @@ -2130,10 +2135,10 @@ defm VMULLslu : N3VLIntSL_HS<1, 0b1010, IIC_VMULi16D, "vmull", "u", int_arm_neon_vmullu>; // VQDMULL : Vector Saturating Doubling Multiply Long (Q = D * D) -defm VQDMULL : N3VLInt_HS<0,1,0b1101,0, IIC_VMULi16D, "vqdmull", "s", - int_arm_neon_vqdmull, 1>; -defm VQDMULLsl: N3VLIntSL_HS<0, 0b1011, IIC_VMULi16D, "vqdmull", "s", - int_arm_neon_vqdmull>; +defm VQDMULL : N3VLInt_HS<0,1,0b1101,0, IIC_VMULi16D, IIC_VMULi32D, + "vqdmull", "s", int_arm_neon_vqdmull, 1>; +defm VQDMULLsl: N3VLIntSL_HS<0, 0b1011, IIC_VMULi16D, + "vqdmull", "s", int_arm_neon_vqdmull>; // Vector Multiply-Accumulate and Multiply-Subtract Operations. @@ -2177,15 +2182,17 @@ def : Pat<(v4f32 (fadd (v4f32 QPR:$src1), (SubReg_i32_lane imm:$lane)))>; // VMLAL : Vector Multiply Accumulate Long (Q += D * D) -defm VMLALs : N3VLInt3_QHS<0,1,0b1000,0, "vmlal", "s", int_arm_neon_vmlals>; -defm VMLALu : N3VLInt3_QHS<1,1,0b1000,0, "vmlal", "u", int_arm_neon_vmlalu>; +defm VMLALs : N3VLInt3_QHS<0,1,0b1000,0, IIC_VMACi16D, IIC_VMACi32D, + "vmlal", "s", int_arm_neon_vmlals>; +defm VMLALu : N3VLInt3_QHS<1,1,0b1000,0, IIC_VMACi16D, IIC_VMACi32D, + "vmlal", "u", int_arm_neon_vmlalu>; defm VMLALsls : N3VLInt3SL_HS<0, 0b0010, "vmlal", "s", int_arm_neon_vmlals>; defm VMLALslu : N3VLInt3SL_HS<1, 0b0010, "vmlal", "u", int_arm_neon_vmlalu>; // VQDMLAL : Vector Saturating Doubling Multiply Accumulate Long (Q += D * D) -defm VQDMLAL : N3VLInt3_HS<0, 1, 0b1001, 0, "vqdmlal", "s", - int_arm_neon_vqdmlal>; +defm VQDMLAL : N3VLInt3_HS<0, 1, 0b1001, 0, IIC_VMACi16D, IIC_VMACi32D, + "vqdmlal", "s", int_arm_neon_vqdmlal>; defm VQDMLALsl: N3VLInt3SL_HS<0, 0b0011, "vqdmlal", "s", int_arm_neon_vqdmlal>; // VMLS : Vector Multiply Subtract (integer and floating-point) @@ -2227,15 +2234,17 @@ def : Pat<(v4f32 (fsub (v4f32 QPR:$src1), (SubReg_i32_lane imm:$lane)))>; // VMLSL : Vector Multiply Subtract Long (Q -= D * D) -defm VMLSLs : N3VLInt3_QHS<0,1,0b1010,0, "vmlsl", "s", int_arm_neon_vmlsls>; -defm VMLSLu : N3VLInt3_QHS<1,1,0b1010,0, "vmlsl", "u", int_arm_neon_vmlslu>; +defm VMLSLs : N3VLInt3_QHS<0,1,0b1010,0, IIC_VMACi16D, IIC_VMACi32D, + "vmlsl", "s", int_arm_neon_vmlsls>; +defm VMLSLu : N3VLInt3_QHS<1,1,0b1010,0, IIC_VMACi16D, IIC_VMACi32D, + "vmlsl", "u", int_arm_neon_vmlslu>; defm VMLSLsls : N3VLInt3SL_HS<0, 0b0110, "vmlsl", "s", int_arm_neon_vmlsls>; defm VMLSLslu : N3VLInt3SL_HS<1, 0b0110, "vmlsl", "u", int_arm_neon_vmlslu>; // VQDMLSL : Vector Saturating Doubling Multiply Subtract Long (Q -= D * D) -defm VQDMLSL : N3VLInt3_HS<0, 1, 0b1011, 0, "vqdmlsl", "s", - int_arm_neon_vqdmlsl>; +defm VQDMLSL : N3VLInt3_HS<0, 1, 0b1011, 0, IIC_VMACi16D, IIC_VMACi32D, + "vqdmlsl", "s", int_arm_neon_vqdmlsl>; defm VQDMLSLsl: N3VLInt3SL_HS<0, 0b111, "vqdmlsl", "s", int_arm_neon_vqdmlsl>; // Vector Subtract Operations. @@ -2248,26 +2257,26 @@ def VSUBfd : N3VD<0, 0, 0b10, 0b1101, 0, IIC_VBIND, "vsub", "f32", def VSUBfq : N3VQ<0, 0, 0b10, 0b1101, 0, IIC_VBINQ, "vsub", "f32", v4f32, v4f32, fsub, 0>; // VSUBL : Vector Subtract Long (Q = D - D) -defm VSUBLs : N3VLInt_QHS<0,1,0b0010,0, IIC_VSHLiD, "vsubl", "s", - int_arm_neon_vsubls, 1>; -defm VSUBLu : N3VLInt_QHS<1,1,0b0010,0, IIC_VSHLiD, "vsubl", "u", - int_arm_neon_vsublu, 1>; +defm VSUBLs : N3VLInt_QHS<0,1,0b0010,0, IIC_VSHLiD, IIC_VSHLiD, + "vsubl", "s", int_arm_neon_vsubls, 1>; +defm VSUBLu : N3VLInt_QHS<1,1,0b0010,0, IIC_VSHLiD, IIC_VSHLiD, + "vsubl", "u", int_arm_neon_vsublu, 1>; // VSUBW : Vector Subtract Wide (Q = Q - D) defm VSUBWs : N3VWInt_QHS<0,1,0b0011,0, "vsubw", "s", int_arm_neon_vsubws, 0>; defm VSUBWu : N3VWInt_QHS<1,1,0b0011,0, "vsubw", "u", int_arm_neon_vsubwu, 0>; // VHSUB : Vector Halving Subtract defm VHSUBs : N3VInt_QHS<0, 0, 0b0010, 0, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vhsub", "s", int_arm_neon_vhsubs, 0>; defm VHSUBu : N3VInt_QHS<1, 0, 0b0010, 0, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vhsub", "u", int_arm_neon_vhsubu, 0>; // VQSUB : Vector Saturing Subtract defm VQSUBs : N3VInt_QHSD<0, 0, 0b0010, 1, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vqsub", "s", int_arm_neon_vqsubs, 0>; defm VQSUBu : N3VInt_QHSD<1, 0, 0b0010, 1, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vqsub", "u", int_arm_neon_vqsubu, 0>; // VSUBHN : Vector Subtract and Narrow Returning High Half (D = Q - Q) defm VSUBHN : N3VNInt_HSD<0,1,0b0110,0, "vsubhn", "i", @@ -2279,8 +2288,8 @@ defm VRSUBHN : N3VNInt_HSD<1,1,0b0110,0, "vrsubhn", "i", // Vector Comparisons. // VCEQ : Vector Compare Equal -defm VCEQ : N3V_QHS<1, 0, 0b1000, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, - IIC_VBINi4Q, "vceq", "i", NEONvceq, 1>; +defm VCEQ : N3V_QHS<1, 0, 0b1000, 1, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, + IIC_VSUBi4Q, "vceq", "i", NEONvceq, 1>; def VCEQfd : N3VD<0,0,0b00,0b1110,0, IIC_VBIND, "vceq", "f32", v2i32, v2f32, NEONvceq, 1>; def VCEQfq : N3VQ<0,0,0b00,0b1110,0, IIC_VBINQ, "vceq", "f32", v4i32, v4f32, @@ -2290,10 +2299,10 @@ defm VCEQz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00010, 0, "vceq", "i", "$dst, $src, #0">; // VCGE : Vector Compare Greater Than or Equal -defm VCGEs : N3V_QHS<0, 0, 0b0011, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, - IIC_VBINi4Q, "vcge", "s", NEONvcge, 0>; -defm VCGEu : N3V_QHS<1, 0, 0b0011, 1, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, - IIC_VBINi4Q, "vcge", "u", NEONvcgeu, 0>; +defm VCGEs : N3V_QHS<0, 0, 0b0011, 1, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, + IIC_VSUBi4Q, "vcge", "s", NEONvcge, 0>; +defm VCGEu : N3V_QHS<1, 0, 0b0011, 1, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, + IIC_VSUBi4Q, "vcge", "u", NEONvcgeu, 0>; def VCGEfd : N3VD<1,0,0b00,0b1110,0, IIC_VBIND, "vcge", "f32", v2i32, v2f32, NEONvcge, 0>; def VCGEfq : N3VQ<1,0,0b00,0b1110,0, IIC_VBINQ, "vcge", "f32", v4i32, v4f32, @@ -2306,10 +2315,10 @@ defm VCLEz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00011, 0, "vcle", "s", "$dst, $src, #0">; // VCGT : Vector Compare Greater Than -defm VCGTs : N3V_QHS<0, 0, 0b0011, 0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, - IIC_VBINi4Q, "vcgt", "s", NEONvcgt, 0>; -defm VCGTu : N3V_QHS<1, 0, 0b0011, 0, IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, - IIC_VBINi4Q, "vcgt", "u", NEONvcgtu, 0>; +defm VCGTs : N3V_QHS<0, 0, 0b0011, 0, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, + IIC_VSUBi4Q, "vcgt", "s", NEONvcgt, 0>; +defm VCGTu : N3V_QHS<1, 0, 0b0011, 0, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, + IIC_VSUBi4Q, "vcgt", "u", NEONvcgtu, 0>; def VCGTfd : N3VD<1,0,0b10,0b1110,0, IIC_VBIND, "vcgt", "f32", v2i32, v2f32, NEONvcgt, 0>; def VCGTfq : N3VQ<1,0,0b10,0b1110,0, IIC_VBINQ, "vcgt", "f32", v4i32, v4f32, @@ -2387,11 +2396,11 @@ def VORNq : N3VX<0, 0, 0b11, 0b0001, 1, 1, (outs QPR:$dst), // VMVN : Vector Bitwise NOT def VMVNd : N2VX<0b11, 0b11, 0b00, 0b00, 0b01011, 0, 0, - (outs DPR:$dst), (ins DPR:$src), IIC_VSHLiD, + (outs DPR:$dst), (ins DPR:$src), IIC_VSUBiD, "vmvn", "$dst, $src", "", [(set DPR:$dst, (v2i32 (vnot8 DPR:$src)))]>; def VMVNq : N2VX<0b11, 0b11, 0b00, 0b00, 0b01011, 1, 0, - (outs QPR:$dst), (ins QPR:$src), IIC_VSHLiD, + (outs QPR:$dst), (ins QPR:$src), IIC_VSUBiD, "vmvn", "$dst, $src", "", [(set QPR:$dst, (v4i32 (vnot16 QPR:$src)))]>; def : Pat<(v2i32 (vnot8 DPR:$src)), (VMVNd DPR:$src)>; @@ -2447,10 +2456,10 @@ def VBITq : N3VX<1, 0, 0b10, 0b0001, 1, 1, // VABD : Vector Absolute Difference defm VABDs : N3VInt_QHS<0, 0, 0b0111, 0, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vabd", "s", int_arm_neon_vabds, 0>; defm VABDu : N3VInt_QHS<1, 0, 0b0111, 0, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vabd", "u", int_arm_neon_vabdu, 0>; def VABDfd : N3VDInt<1, 0, 0b10, 0b1101, 0, N3RegFrm, IIC_VBIND, "vabd", "f32", v2f32, v2f32, int_arm_neon_vabds, 0>; @@ -2458,56 +2467,68 @@ def VABDfq : N3VQInt<1, 0, 0b10, 0b1101, 0, N3RegFrm, IIC_VBINQ, "vabd", "f32", v4f32, v4f32, int_arm_neon_vabds, 0>; // VABDL : Vector Absolute Difference Long (Q = | D - D |) -defm VABDLs : N3VLInt_QHS<0,1,0b0111,0, IIC_VBINi4Q, +defm VABDLs : N3VLInt_QHS<0,1,0b0111,0, IIC_VSUBi4Q, IIC_VSUBi4Q, "vabdl", "s", int_arm_neon_vabdls, 0>; -defm VABDLu : N3VLInt_QHS<1,1,0b0111,0, IIC_VBINi4Q, +defm VABDLu : N3VLInt_QHS<1,1,0b0111,0, IIC_VSUBi4Q, IIC_VSUBi4Q, "vabdl", "u", int_arm_neon_vabdlu, 0>; // VABA : Vector Absolute Difference and Accumulate -defm VABAs : N3VInt3_QHS<0,0,0b0111,1, "vaba", "s", int_arm_neon_vabas>; -defm VABAu : N3VInt3_QHS<1,0,0b0111,1, "vaba", "u", int_arm_neon_vabau>; +defm VABAs : N3VInt3_QHS<0,0,0b0111,1, IIC_VABAD, IIC_VABAQ, + "vaba", "s", int_arm_neon_vabas>; +defm VABAu : N3VInt3_QHS<1,0,0b0111,1, IIC_VABAD, IIC_VABAQ, + "vaba", "u", int_arm_neon_vabau>; // VABAL : Vector Absolute Difference and Accumulate Long (Q += | D - D |) -defm VABALs : N3VLInt3_QHS<0,1,0b0101,0, "vabal", "s", int_arm_neon_vabals>; -defm VABALu : N3VLInt3_QHS<1,1,0b0101,0, "vabal", "u", int_arm_neon_vabalu>; +defm VABALs : N3VLInt3_QHS<0,1,0b0101,0, IIC_VABAD, IIC_VABAD, + "vabal", "s", int_arm_neon_vabals>; +defm VABALu : N3VLInt3_QHS<1,1,0b0101,0, IIC_VABAD, IIC_VABAD, + "vabal", "u", int_arm_neon_vabalu>; // Vector Maximum and Minimum. // VMAX : Vector Maximum defm VMAXs : N3VInt_QHS<0, 0, 0b0110, 0, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vmax", "s", int_arm_neon_vmaxs, 1>; defm VMAXu : N3VInt_QHS<1, 0, 0b0110, 0, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vmax", "u", int_arm_neon_vmaxu, 1>; -def VMAXfd : N3VDInt<0, 0, 0b00, 0b1111, 0, N3RegFrm, IIC_VBIND, "vmax", - "f32", v2f32, v2f32, int_arm_neon_vmaxs, 1>; -def VMAXfq : N3VQInt<0, 0, 0b00, 0b1111, 0, N3RegFrm, IIC_VBINQ, "vmax", - "f32", v4f32, v4f32, int_arm_neon_vmaxs, 1>; +def VMAXfd : N3VDInt<0, 0, 0b00, 0b1111, 0, N3RegFrm, IIC_VBIND, + "vmax", "f32", + v2f32, v2f32, int_arm_neon_vmaxs, 1>; +def VMAXfq : N3VQInt<0, 0, 0b00, 0b1111, 0, N3RegFrm, IIC_VBINQ, + "vmax", "f32", + v4f32, v4f32, int_arm_neon_vmaxs, 1>; // VMIN : Vector Minimum defm VMINs : N3VInt_QHS<0, 0, 0b0110, 1, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vmin", "s", int_arm_neon_vmins, 1>; defm VMINu : N3VInt_QHS<1, 0, 0b0110, 1, N3RegFrm, - IIC_VBINi4D, IIC_VBINi4D, IIC_VBINi4Q, IIC_VBINi4Q, + IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, IIC_VSUBi4Q, "vmin", "u", int_arm_neon_vminu, 1>; -def VMINfd : N3VDInt<0, 0, 0b10, 0b1111, 0, N3RegFrm, IIC_VBIND, "vmin", - "f32", v2f32, v2f32, int_arm_neon_vmins, 1>; -def VMINfq : N3VQInt<0, 0, 0b10, 0b1111, 0, N3RegFrm, IIC_VBINQ, "vmin", - "f32", v4f32, v4f32, int_arm_neon_vmins, 1>; +def VMINfd : N3VDInt<0, 0, 0b10, 0b1111, 0, N3RegFrm, IIC_VBIND, + "vmin", "f32", + v2f32, v2f32, int_arm_neon_vmins, 1>; +def VMINfq : N3VQInt<0, 0, 0b10, 0b1111, 0, N3RegFrm, IIC_VBINQ, + "vmin", "f32", + v4f32, v4f32, int_arm_neon_vmins, 1>; // Vector Pairwise Operations. // VPADD : Vector Pairwise Add -def VPADDi8 : N3VDInt<0, 0, 0b00, 0b1011, 1, N3RegFrm, IIC_VBINiD, "vpadd", - "i8", v8i8, v8i8, int_arm_neon_vpadd, 0>; -def VPADDi16 : N3VDInt<0, 0, 0b01, 0b1011, 1, N3RegFrm, IIC_VBINiD, "vpadd", - "i16", v4i16, v4i16, int_arm_neon_vpadd, 0>; -def VPADDi32 : N3VDInt<0, 0, 0b10, 0b1011, 1, N3RegFrm, IIC_VBINiD, "vpadd", - "i32", v2i32, v2i32, int_arm_neon_vpadd, 0>; -def VPADDf : N3VDInt<1, 0, 0b00, 0b1101, 0, N3RegFrm, IIC_VBIND, "vpadd", - "f32", v2f32, v2f32, int_arm_neon_vpadd, 0>; +def VPADDi8 : N3VDInt<0, 0, 0b00, 0b1011, 1, N3RegFrm, IIC_VSHLiD, + "vpadd", "i8", + v8i8, v8i8, int_arm_neon_vpadd, 0>; +def VPADDi16 : N3VDInt<0, 0, 0b01, 0b1011, 1, N3RegFrm, IIC_VSHLiD, + "vpadd", "i16", + v4i16, v4i16, int_arm_neon_vpadd, 0>; +def VPADDi32 : N3VDInt<0, 0, 0b10, 0b1011, 1, N3RegFrm, IIC_VSHLiD, + "vpadd", "i32", + v2i32, v2i32, int_arm_neon_vpadd, 0>; +def VPADDf : N3VDInt<1, 0, 0b00, 0b1101, 0, N3RegFrm, + IIC_VBIND, "vpadd", "f32", + v2f32, v2f32, int_arm_neon_vpadd, 0>; // VPADDL : Vector Pairwise Add Long defm VPADDLs : N2VPLInt_QHS<0b11, 0b11, 0b00, 0b00100, 0, "vpaddl", "s", @@ -2522,35 +2543,35 @@ defm VPADALu : N2VPLInt2_QHS<0b11, 0b11, 0b00, 0b01101, 0, "vpadal", "u", int_arm_neon_vpadalu>; // VPMAX : Vector Pairwise Maximum -def VPMAXs8 : N3VDInt<0, 0, 0b00, 0b1010, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXs8 : N3VDInt<0, 0, 0b00, 0b1010, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "s8", v8i8, v8i8, int_arm_neon_vpmaxs, 0>; -def VPMAXs16 : N3VDInt<0, 0, 0b01, 0b1010, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXs16 : N3VDInt<0, 0, 0b01, 0b1010, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "s16", v4i16, v4i16, int_arm_neon_vpmaxs, 0>; -def VPMAXs32 : N3VDInt<0, 0, 0b10, 0b1010, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXs32 : N3VDInt<0, 0, 0b10, 0b1010, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "s32", v2i32, v2i32, int_arm_neon_vpmaxs, 0>; -def VPMAXu8 : N3VDInt<1, 0, 0b00, 0b1010, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXu8 : N3VDInt<1, 0, 0b00, 0b1010, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "u8", v8i8, v8i8, int_arm_neon_vpmaxu, 0>; -def VPMAXu16 : N3VDInt<1, 0, 0b01, 0b1010, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXu16 : N3VDInt<1, 0, 0b01, 0b1010, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "u16", v4i16, v4i16, int_arm_neon_vpmaxu, 0>; -def VPMAXu32 : N3VDInt<1, 0, 0b10, 0b1010, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXu32 : N3VDInt<1, 0, 0b10, 0b1010, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "u32", v2i32, v2i32, int_arm_neon_vpmaxu, 0>; -def VPMAXf : N3VDInt<1, 0, 0b00, 0b1111, 0, N3RegFrm, IIC_VBINi4D, "vpmax", +def VPMAXf : N3VDInt<1, 0, 0b00, 0b1111, 0, N3RegFrm, IIC_VSUBi4D, "vpmax", "f32", v2f32, v2f32, int_arm_neon_vpmaxs, 0>; // VPMIN : Vector Pairwise Minimum -def VPMINs8 : N3VDInt<0, 0, 0b00, 0b1010, 1, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINs8 : N3VDInt<0, 0, 0b00, 0b1010, 1, N3RegFrm, IIC_VSUBi4D, "vpmin", "s8", v8i8, v8i8, int_arm_neon_vpmins, 0>; -def VPMINs16 : N3VDInt<0, 0, 0b01, 0b1010, 1, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINs16 : N3VDInt<0, 0, 0b01, 0b1010, 1, N3RegFrm, IIC_VSUBi4D, "vpmin", "s16", v4i16, v4i16, int_arm_neon_vpmins, 0>; -def VPMINs32 : N3VDInt<0, 0, 0b10, 0b1010, 1, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINs32 : N3VDInt<0, 0, 0b10, 0b1010, 1, N3RegFrm, IIC_VSUBi4D, "vpmin", "s32", v2i32, v2i32, int_arm_neon_vpmins, 0>; -def VPMINu8 : N3VDInt<1, 0, 0b00, 0b1010, 1, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINu8 : N3VDInt<1, 0, 0b00, 0b1010, 1, N3RegFrm, IIC_VSUBi4D, "vpmin", "u8", v8i8, v8i8, int_arm_neon_vpminu, 0>; -def VPMINu16 : N3VDInt<1, 0, 0b01, 0b1010, 1, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINu16 : N3VDInt<1, 0, 0b01, 0b1010, 1, N3RegFrm, IIC_VSUBi4D, "vpmin", "u16", v4i16, v4i16, int_arm_neon_vpminu, 0>; -def VPMINu32 : N3VDInt<1, 0, 0b10, 0b1010, 1, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINu32 : N3VDInt<1, 0, 0b10, 0b1010, 1, N3RegFrm, IIC_VSUBi4D, "vpmin", "u32", v2i32, v2i32, int_arm_neon_vpminu, 0>; -def VPMINf : N3VDInt<1, 0, 0b10, 0b1111, 0, N3RegFrm, IIC_VBINi4D, "vpmin", +def VPMINf : N3VDInt<1, 0, 0b10, 0b1111, 0, N3RegFrm, IIC_VSUBi4D, "vpmin", "f32", v2f32, v2f32, int_arm_neon_vpmins, 0>; // Vector Reciprocal and Reciprocal Square Root Estimate and Step. |