string OpcodeStr, string PackTy, string Suff,
PatFrag MemFrag128, PatFrag MemFrag256,
SDNode Op, ValueType OpTy128, ValueType OpTy256,
- X86FoldableSchedWrite sched128,
- X86FoldableSchedWrite sched256> {
+ X86SchedWriteWidths sched> {
defm NAME#213#Suff : fma3p_rm_213<opc213, !strconcat(OpcodeStr, "213", PackTy),
- VR128, OpTy128, f128mem, MemFrag128, Op, sched128>;
+ VR128, OpTy128, f128mem, MemFrag128, Op, sched.XMM>;
defm NAME#231#Suff : fma3p_rm_231<opc231, !strconcat(OpcodeStr, "231", PackTy),
- VR128, OpTy128, f128mem, MemFrag128, Op, sched128>;
+ VR128, OpTy128, f128mem, MemFrag128, Op, sched.XMM>;
defm NAME#132#Suff : fma3p_rm_132<opc132, !strconcat(OpcodeStr, "132", PackTy),
- VR128, OpTy128, f128mem, MemFrag128, Op, sched128>;
+ VR128, OpTy128, f128mem, MemFrag128, Op, sched.XMM>;
defm NAME#213#Suff#Y : fma3p_rm_213<opc213, !strconcat(OpcodeStr, "213", PackTy),
- VR256, OpTy256, f256mem, MemFrag256, Op, sched256>,
+ VR256, OpTy256, f256mem, MemFrag256, Op, sched.YMM>,
VEX_L;
defm NAME#231#Suff#Y : fma3p_rm_231<opc231, !strconcat(OpcodeStr, "231", PackTy),
- VR256, OpTy256, f256mem, MemFrag256, Op, sched256>,
+ VR256, OpTy256, f256mem, MemFrag256, Op, sched.YMM>,
VEX_L;
defm NAME#132#Suff#Y : fma3p_rm_132<opc132, !strconcat(OpcodeStr, "132", PackTy),
- VR256, OpTy256, f256mem, MemFrag256, Op, sched256>,
+ VR256, OpTy256, f256mem, MemFrag256, Op, sched.YMM>,
VEX_L;
}
let ExeDomain = SSEPackedSingle in {
defm VFMADD : fma3p_forms<0x98, 0xA8, 0xB8, "vfmadd", "ps", "PS",
loadv4f32, loadv8f32, X86Fmadd, v4f32, v8f32,
- WriteFMA, WriteFMAY>;
+ SchedWriteFMA>;
defm VFMSUB : fma3p_forms<0x9A, 0xAA, 0xBA, "vfmsub", "ps", "PS",
loadv4f32, loadv8f32, X86Fmsub, v4f32, v8f32,
- WriteFMA, WriteFMAY>;
+ SchedWriteFMA>;
defm VFMADDSUB : fma3p_forms<0x96, 0xA6, 0xB6, "vfmaddsub", "ps", "PS",
loadv4f32, loadv8f32, X86Fmaddsub, v4f32, v8f32,
- WriteFMA, WriteFMAY>;
+ SchedWriteFMA>;
defm VFMSUBADD : fma3p_forms<0x97, 0xA7, 0xB7, "vfmsubadd", "ps", "PS",
loadv4f32, loadv8f32, X86Fmsubadd, v4f32, v8f32,
- WriteFMA, WriteFMAY>;
+ SchedWriteFMA>;
}
let ExeDomain = SSEPackedDouble in {
defm VFMADD : fma3p_forms<0x98, 0xA8, 0xB8, "vfmadd", "pd", "PD",
loadv2f64, loadv4f64, X86Fmadd, v2f64,
- v4f64, WriteFMA, WriteFMAY>, VEX_W;
+ v4f64, SchedWriteFMA>, VEX_W;
defm VFMSUB : fma3p_forms<0x9A, 0xAA, 0xBA, "vfmsub", "pd", "PD",
loadv2f64, loadv4f64, X86Fmsub, v2f64,
- v4f64, WriteFMA, WriteFMAY>, VEX_W;
+ v4f64, SchedWriteFMA>, VEX_W;
defm VFMADDSUB : fma3p_forms<0x96, 0xA6, 0xB6, "vfmaddsub", "pd", "PD",
loadv2f64, loadv4f64, X86Fmaddsub,
- v2f64, v4f64, WriteFMA, WriteFMAY>, VEX_W;
+ v2f64, v4f64, SchedWriteFMA>, VEX_W;
defm VFMSUBADD : fma3p_forms<0x97, 0xA7, 0xB7, "vfmsubadd", "pd", "PD",
loadv2f64, loadv4f64, X86Fmsubadd,
- v2f64, v4f64, WriteFMA, WriteFMAY>, VEX_W;
+ v2f64, v4f64, SchedWriteFMA>, VEX_W;
}
// Fused Negative Multiply-Add
let ExeDomain = SSEPackedSingle in {
defm VFNMADD : fma3p_forms<0x9C, 0xAC, 0xBC, "vfnmadd", "ps", "PS", loadv4f32,
- loadv8f32, X86Fnmadd, v4f32, v8f32, WriteFMA, WriteFMAY>;
+ loadv8f32, X86Fnmadd, v4f32, v8f32, SchedWriteFMA>;
defm VFNMSUB : fma3p_forms<0x9E, 0xAE, 0xBE, "vfnmsub", "ps", "PS", loadv4f32,
- loadv8f32, X86Fnmsub, v4f32, v8f32, WriteFMA, WriteFMAY>;
+ loadv8f32, X86Fnmsub, v4f32, v8f32, SchedWriteFMA>;
}
let ExeDomain = SSEPackedDouble in {
defm VFNMADD : fma3p_forms<0x9C, 0xAC, 0xBC, "vfnmadd", "pd", "PD", loadv2f64,
- loadv4f64, X86Fnmadd, v2f64, v4f64, WriteFMA, WriteFMAY>, VEX_W;
+ loadv4f64, X86Fnmadd, v2f64, v4f64, SchedWriteFMA>, VEX_W;
defm VFNMSUB : fma3p_forms<0x9E, 0xAE, 0xBE, "vfnmsub", "pd", "PD", loadv2f64,
- loadv4f64, X86Fnmsub, v2f64, v4f64, WriteFMA, WriteFMAY>, VEX_W;
+ loadv4f64, X86Fnmsub, v2f64, v4f64, SchedWriteFMA>, VEX_W;
}
// All source register operands of FMA opcodes defined in fma3s_rm multiclass
}
defm VFMADD : fma3s<0x99, 0xA9, 0xB9, "vfmadd", X86Fmadds1, X86Fmadd,
- WriteFMAS>, VEX_LIG;
+ SchedWriteFMA.Scl>, VEX_LIG;
defm VFMSUB : fma3s<0x9B, 0xAB, 0xBB, "vfmsub", X86Fmsubs1, X86Fmsub,
- WriteFMAS>, VEX_LIG;
+ SchedWriteFMA.Scl>, VEX_LIG;
defm VFNMADD : fma3s<0x9D, 0xAD, 0xBD, "vfnmadd", X86Fnmadds1, X86Fnmadd,
- WriteFMAS>, VEX_LIG;
+ SchedWriteFMA.Scl>, VEX_LIG;
defm VFNMSUB : fma3s<0x9F, 0xAF, 0xBF, "vfnmsub", X86Fnmsubs1, X86Fnmsub,
- WriteFMAS>, VEX_LIG;
+ SchedWriteFMA.Scl>, VEX_LIG;
//===----------------------------------------------------------------------===//
// FMA4 - AMD 4 operand Fused Multiply-Add instructions
multiclass fma4p<bits<8> opc, string OpcodeStr, SDNode OpNode,
ValueType OpVT128, ValueType OpVT256,
PatFrag ld_frag128, PatFrag ld_frag256,
- X86FoldableSchedWrite sched128,
- X86FoldableSchedWrite sched256> {
+ X86SchedWriteWidths sched> {
let isCommutable = 1 in
def rr : FMA4<opc, MRMSrcRegOp4, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2, VR128:$src3),
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[(set VR128:$dst,
(OpVT128 (OpNode VR128:$src1, VR128:$src2, VR128:$src3)))]>,
- VEX_W, Sched<[sched128]>;
+ VEX_W, Sched<[sched.XMM]>;
def rm : FMA4<opc, MRMSrcMemOp4, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2, f128mem:$src3),
!strconcat(OpcodeStr,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[(set VR128:$dst, (OpNode VR128:$src1, VR128:$src2,
(ld_frag128 addr:$src3)))]>, VEX_W,
- Sched<[sched128.Folded, ReadAfterLd, ReadAfterLd]>;
+ Sched<[sched.XMM.Folded, ReadAfterLd, ReadAfterLd]>;
def mr : FMA4<opc, MRMSrcMem, (outs VR128:$dst),
(ins VR128:$src1, f128mem:$src2, VR128:$src3),
!strconcat(OpcodeStr,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[(set VR128:$dst,
(OpNode VR128:$src1, (ld_frag128 addr:$src2), VR128:$src3))]>,
- Sched<[sched128.Folded, ReadAfterLd,
+ Sched<[sched.XMM.Folded, ReadAfterLd,
// f128mem:$src2
ReadDefault, ReadDefault, ReadDefault, ReadDefault,
ReadDefault,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[(set VR256:$dst,
(OpVT256 (OpNode VR256:$src1, VR256:$src2, VR256:$src3)))]>,
- VEX_W, VEX_L, Sched<[sched256]>;
+ VEX_W, VEX_L, Sched<[sched.YMM]>;
def Yrm : FMA4<opc, MRMSrcMemOp4, (outs VR256:$dst),
(ins VR256:$src1, VR256:$src2, f256mem:$src3),
!strconcat(OpcodeStr,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[(set VR256:$dst, (OpNode VR256:$src1, VR256:$src2,
(ld_frag256 addr:$src3)))]>, VEX_W, VEX_L,
- Sched<[sched256.Folded, ReadAfterLd, ReadAfterLd]>;
+ Sched<[sched.YMM.Folded, ReadAfterLd, ReadAfterLd]>;
def Ymr : FMA4<opc, MRMSrcMem, (outs VR256:$dst),
(ins VR256:$src1, f256mem:$src2, VR256:$src3),
!strconcat(OpcodeStr,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[(set VR256:$dst, (OpNode VR256:$src1,
(ld_frag256 addr:$src2), VR256:$src3))]>, VEX_L,
- Sched<[sched256.Folded, ReadAfterLd,
+ Sched<[sched.YMM.Folded, ReadAfterLd,
// f256mem:$src2
ReadDefault, ReadDefault, ReadDefault, ReadDefault,
ReadDefault,
(ins VR128:$src1, VR128:$src2, VR128:$src3),
!strconcat(OpcodeStr,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"), []>,
- Sched<[sched128]>, FoldGenData<NAME#rr>;
+ Sched<[sched.XMM]>, FoldGenData<NAME#rr>;
def Yrr_REV : FMA4<opc, MRMSrcReg, (outs VR256:$dst),
(ins VR256:$src1, VR256:$src2, VR256:$src3),
!strconcat(OpcodeStr,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"), []>,
- VEX_L, Sched<[sched256]>, FoldGenData<NAME#Yrr>;
+ VEX_L, Sched<[sched.YMM]>, FoldGenData<NAME#Yrr>;
} // isCodeGenOnly = 1
}
let ExeDomain = SSEPackedSingle in {
// Scalar Instructions
defm VFMADDSS4 : fma4s<0x6A, "vfmaddss", FR32, f32mem, f32, X86Fmadd, loadf32,
- WriteFMAS>,
+ SchedWriteFMA.Scl>,
fma4s_int<0x6A, "vfmaddss", ssmem, v4f32, sse_load_f32,
- X86Fmadd4s, WriteFMAS>;
+ X86Fmadd4s, SchedWriteFMA.Scl>;
defm VFMSUBSS4 : fma4s<0x6E, "vfmsubss", FR32, f32mem, f32, X86Fmsub, loadf32,
- WriteFMAS>,
+ SchedWriteFMA.Scl>,
fma4s_int<0x6E, "vfmsubss", ssmem, v4f32, sse_load_f32,
- X86Fmsub4s, WriteFMAS>;
+ X86Fmsub4s, SchedWriteFMA.Scl>;
defm VFNMADDSS4 : fma4s<0x7A, "vfnmaddss", FR32, f32mem, f32,
- X86Fnmadd, loadf32, WriteFMAS>,
+ X86Fnmadd, loadf32, SchedWriteFMA.Scl>,
fma4s_int<0x7A, "vfnmaddss", ssmem, v4f32, sse_load_f32,
- X86Fnmadd4s, WriteFMAS>;
+ X86Fnmadd4s, SchedWriteFMA.Scl>;
defm VFNMSUBSS4 : fma4s<0x7E, "vfnmsubss", FR32, f32mem, f32,
- X86Fnmsub, loadf32, WriteFMAS>,
+ X86Fnmsub, loadf32, SchedWriteFMA.Scl>,
fma4s_int<0x7E, "vfnmsubss", ssmem, v4f32, sse_load_f32,
- X86Fnmsub4s, WriteFMAS>;
+ X86Fnmsub4s, SchedWriteFMA.Scl>;
// Packed Instructions
defm VFMADDPS4 : fma4p<0x68, "vfmaddps", X86Fmadd, v4f32, v8f32,
- loadv4f32, loadv8f32, WriteFMA, WriteFMAY>;
+ loadv4f32, loadv8f32, SchedWriteFMA>;
defm VFMSUBPS4 : fma4p<0x6C, "vfmsubps", X86Fmsub, v4f32, v8f32,
- loadv4f32, loadv8f32, WriteFMA, WriteFMAY>;
+ loadv4f32, loadv8f32, SchedWriteFMA>;
defm VFNMADDPS4 : fma4p<0x78, "vfnmaddps", X86Fnmadd, v4f32, v8f32,
- loadv4f32, loadv8f32, WriteFMA, WriteFMAY>;
+ loadv4f32, loadv8f32, SchedWriteFMA>;
defm VFNMSUBPS4 : fma4p<0x7C, "vfnmsubps", X86Fnmsub, v4f32, v8f32,
- loadv4f32, loadv8f32, WriteFMA, WriteFMAY>;
+ loadv4f32, loadv8f32, SchedWriteFMA>;
defm VFMADDSUBPS4 : fma4p<0x5C, "vfmaddsubps", X86Fmaddsub, v4f32, v8f32,
- loadv4f32, loadv8f32, WriteFMA, WriteFMAY>;
+ loadv4f32, loadv8f32, SchedWriteFMA>;
defm VFMSUBADDPS4 : fma4p<0x5E, "vfmsubaddps", X86Fmsubadd, v4f32, v8f32,
- loadv4f32, loadv8f32, WriteFMA, WriteFMAY>;
+ loadv4f32, loadv8f32, SchedWriteFMA>;
}
let ExeDomain = SSEPackedDouble in {
// Scalar Instructions
defm VFMADDSD4 : fma4s<0x6B, "vfmaddsd", FR64, f64mem, f64, X86Fmadd, loadf64,
- WriteFMAS>,
+ SchedWriteFMA.Scl>,
fma4s_int<0x6B, "vfmaddsd", sdmem, v2f64, sse_load_f64,
- X86Fmadd4s, WriteFMAS>;
+ X86Fmadd4s, SchedWriteFMA.Scl>;
defm VFMSUBSD4 : fma4s<0x6F, "vfmsubsd", FR64, f64mem, f64, X86Fmsub, loadf64,
- WriteFMAS>,
+ SchedWriteFMA.Scl>,
fma4s_int<0x6F, "vfmsubsd", sdmem, v2f64, sse_load_f64,
- X86Fmsub4s, WriteFMAS>;
+ X86Fmsub4s, SchedWriteFMA.Scl>;
defm VFNMADDSD4 : fma4s<0x7B, "vfnmaddsd", FR64, f64mem, f64,
- X86Fnmadd, loadf64, WriteFMAS>,
+ X86Fnmadd, loadf64, SchedWriteFMA.Scl>,
fma4s_int<0x7B, "vfnmaddsd", sdmem, v2f64, sse_load_f64,
- X86Fnmadd4s, WriteFMAS>;
+ X86Fnmadd4s, SchedWriteFMA.Scl>;
defm VFNMSUBSD4 : fma4s<0x7F, "vfnmsubsd", FR64, f64mem, f64,
- X86Fnmsub, loadf64, WriteFMAS>,
+ X86Fnmsub, loadf64, SchedWriteFMA.Scl>,
fma4s_int<0x7F, "vfnmsubsd", sdmem, v2f64, sse_load_f64,
- X86Fnmsub4s, WriteFMAS>;
+ X86Fnmsub4s, SchedWriteFMA.Scl>;
// Packed Instructions
defm VFMADDPD4 : fma4p<0x69, "vfmaddpd", X86Fmadd, v2f64, v4f64,
- loadv2f64, loadv4f64, WriteFMA, WriteFMAY>;
+ loadv2f64, loadv4f64, SchedWriteFMA>;
defm VFMSUBPD4 : fma4p<0x6D, "vfmsubpd", X86Fmsub, v2f64, v4f64,
- loadv2f64, loadv4f64, WriteFMA, WriteFMAY>;
+ loadv2f64, loadv4f64, SchedWriteFMA>;
defm VFNMADDPD4 : fma4p<0x79, "vfnmaddpd", X86Fnmadd, v2f64, v4f64,
- loadv2f64, loadv4f64, WriteFMA, WriteFMAY>;
+ loadv2f64, loadv4f64, SchedWriteFMA>;
defm VFNMSUBPD4 : fma4p<0x7D, "vfnmsubpd", X86Fnmsub, v2f64, v4f64,
- loadv2f64, loadv4f64, WriteFMA, WriteFMAY>;
+ loadv2f64, loadv4f64, SchedWriteFMA>;
defm VFMADDSUBPD4 : fma4p<0x5D, "vfmaddsubpd", X86Fmaddsub, v2f64, v4f64,
- loadv2f64, loadv4f64, WriteFMA, WriteFMAY>;
+ loadv2f64, loadv4f64, SchedWriteFMA>;
defm VFMSUBADDPD4 : fma4p<0x5F, "vfmsubaddpd", X86Fmsubadd, v2f64, v4f64,
- loadv2f64, loadv4f64, WriteFMA, WriteFMAY>;
+ loadv2f64, loadv4f64, SchedWriteFMA>;
}