forked from OSchip/llvm-project
[AVX512] Implemented AVX512VL FP bnary packed instructions (VADDP*, VSUBP*, VMULP*, VDIVP*, VMAXP*, VMINP*)
Refactored through AVX512_maskable Added encoding tests for them. llvm-svn: 220858
This commit is contained in:
Robert Khasanov
parent
f51a34ec1f
commit
595e598869
|
|
@ -2984,118 +2984,58 @@ defm VDIV : avx512_binop_s<0x5E, "div", fdiv, SSE_ALU_ITINS_S>;
|
|||
}
|
||||
|
||||
multiclass avx512_fp_packed<bits<8> opc, string OpcodeStr, SDNode OpNode,
|
||||
RegisterClass KRC,
|
||||
RegisterClass RC, ValueType vt,
|
||||
X86MemOperand x86memop, PatFrag mem_frag,
|
||||
X86MemOperand x86scalar_mop, PatFrag scalar_mfrag,
|
||||
string BrdcstStr,
|
||||
Domain d, OpndItins itins, bit commutable> {
|
||||
let isCommutable = commutable in {
|
||||
def rr : PI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
|
||||
!strconcat(OpcodeStr, " \t{$src2, $src1, $dst|$dst, $src1, $src2}"),
|
||||
[(set RC:$dst, (vt (OpNode RC:$src1, RC:$src2)))], itins.rr, d>,
|
||||
EVEX_4V;
|
||||
|
||||
def rrk: PI<opc, MRMSrcReg, (outs RC:$dst), (ins KRC:$mask, RC:$src1, RC:$src2),
|
||||
!strconcat(OpcodeStr,
|
||||
" \t{$src2, $src1, $dst {${mask}} |$dst {${mask}}, $src1, $src2}"),
|
||||
[], itins.rr, d>, EVEX_4V, EVEX_K;
|
||||
|
||||
def rrkz: PI<opc, MRMSrcReg, (outs RC:$dst), (ins KRC:$mask, RC:$src1, RC:$src2),
|
||||
!strconcat(OpcodeStr,
|
||||
" \t{$src2, $src1, $dst {${mask}} {z}|$dst {${mask}} {z}, $src1, $src2}"),
|
||||
[], itins.rr, d>, EVEX_4V, EVEX_KZ;
|
||||
}
|
||||
|
||||
X86VectorVTInfo _, bit IsCommutable> {
|
||||
defm rr: AVX512_maskable<opc, MRMSrcReg, _, (outs _.RC:$dst),
|
||||
(ins _.RC:$src1, _.RC:$src2), OpcodeStr##_.Suffix,
|
||||
"$src2, $src1", "$src1, $src2",
|
||||
(_.VT (OpNode _.RC:$src1, _.RC:$src2))>, EVEX_4V;
|
||||
let mayLoad = 1 in {
|
||||
def rm : PI<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
|
||||
!strconcat(OpcodeStr, " \t{$src2, $src1, $dst|$dst, $src1, $src2}"),
|
||||
[(set RC:$dst, (OpNode RC:$src1, (mem_frag addr:$src2)))],
|
||||
itins.rm, d>, EVEX_4V;
|
||||
defm rm: AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
|
||||
(ins _.RC:$src1, _.MemOp:$src2), OpcodeStr##_.Suffix,
|
||||
"$src2, $src1", "$src1, $src2",
|
||||
(OpNode _.RC:$src1, (_.LdFrag addr:$src2))>, EVEX_4V;
|
||||
defm rmb: AVX512_maskable<opc, MRMSrcMem, _, (outs _.RC:$dst),
|
||||
(ins _.RC:$src1, _.ScalarMemOp:$src2), OpcodeStr##_.Suffix,
|
||||
"${src2}"##_.BroadcastStr##", $src1",
|
||||
"$src1, ${src2}"##_.BroadcastStr,
|
||||
(OpNode _.RC:$src1, (_.VT (X86VBroadcast
|
||||
(_.ScalarLdFrag addr:$src2))))>,
|
||||
EVEX_4V, EVEX_B;
|
||||
}//let mayLoad = 1
|
||||
}
|
||||
|
||||
def rmb : PI<opc, MRMSrcMem, (outs RC:$dst),
|
||||
(ins RC:$src1, x86scalar_mop:$src2),
|
||||
!strconcat(OpcodeStr, " \t{${src2}", BrdcstStr,
|
||||
", $src1, $dst|$dst, $src1, ${src2}", BrdcstStr, "}"),
|
||||
[(set RC:$dst, (OpNode RC:$src1,
|
||||
(vt (X86VBroadcast (scalar_mfrag addr:$src2)))))],
|
||||
itins.rm, d>, EVEX_4V, EVEX_B;
|
||||
multiclass avx512_fp_binop_p<bits<8> opc, string OpcodeStr, SDNode OpNode,
|
||||
bit IsCommutable = 0> {
|
||||
defm PSZ : avx512_fp_packed<opc, OpcodeStr, OpNode, v16f32_info,
|
||||
IsCommutable>, EVEX_V512, PS,
|
||||
EVEX_CD8<32, CD8VF>;
|
||||
defm PDZ : avx512_fp_packed<opc, OpcodeStr, OpNode, v8f64_info,
|
||||
IsCommutable>, EVEX_V512, PD, VEX_W,
|
||||
EVEX_CD8<64, CD8VF>;
|
||||
|
||||
def rmk : PI<opc, MRMSrcMem, (outs RC:$dst),
|
||||
(ins KRC:$mask, RC:$src1, x86memop:$src2), !strconcat(OpcodeStr,
|
||||
"\t{$src2, $src1, $dst {${mask}}|$dst {${mask}}, $src1, $src2}"),
|
||||
[], itins.rm, d>, EVEX_4V, EVEX_K;
|
||||
|
||||
def rmkz : PI<opc, MRMSrcMem, (outs RC:$dst),
|
||||
(ins KRC:$mask, RC:$src1, x86memop:$src2), !strconcat(OpcodeStr,
|
||||
"\t{$src2, $src1, $dst {${mask}} {z}|$dst {${mask}} {z}, $src1, $src2}"),
|
||||
[], itins.rm, d>, EVEX_4V, EVEX_KZ;
|
||||
|
||||
def rmbk : PI<opc, MRMSrcMem, (outs RC:$dst),
|
||||
(ins KRC:$mask, RC:$src1, x86scalar_mop:$src2), !strconcat(OpcodeStr,
|
||||
" \t{${src2}", BrdcstStr,
|
||||
", $src1, $dst {${mask}}|$dst {${mask}}, $src1, ${src2}", BrdcstStr, "}"),
|
||||
[], itins.rm, d>, EVEX_4V, EVEX_B, EVEX_K;
|
||||
|
||||
def rmbkz : PI<opc, MRMSrcMem, (outs RC:$dst),
|
||||
(ins KRC:$mask, RC:$src1, x86scalar_mop:$src2), !strconcat(OpcodeStr,
|
||||
" \t{${src2}", BrdcstStr,
|
||||
", $src1, $dst {${mask}} {z}|$dst {${mask}} {z}, $src1, ${src2}",
|
||||
BrdcstStr, "}"),
|
||||
[], itins.rm, d>, EVEX_4V, EVEX_B, EVEX_KZ;
|
||||
// Define only if AVX512VL feature is present.
|
||||
let Predicates = [HasVLX] in {
|
||||
defm PSZ128 : avx512_fp_packed<opc, OpcodeStr, OpNode, v4f32x_info,
|
||||
IsCommutable>, EVEX_V128, PS,
|
||||
EVEX_CD8<32, CD8VF>;
|
||||
defm PSZ256 : avx512_fp_packed<opc, OpcodeStr, OpNode, v8f32x_info,
|
||||
IsCommutable>, EVEX_V256, PS,
|
||||
EVEX_CD8<32, CD8VF>;
|
||||
defm PDZ128 : avx512_fp_packed<opc, OpcodeStr, OpNode, v2f64x_info,
|
||||
IsCommutable>, EVEX_V128, PD, VEX_W,
|
||||
EVEX_CD8<64, CD8VF>;
|
||||
defm PDZ256 : avx512_fp_packed<opc, OpcodeStr, OpNode, v4f64x_info,
|
||||
IsCommutable>, EVEX_V256, PD, VEX_W,
|
||||
EVEX_CD8<64, CD8VF>;
|
||||
}
|
||||
}
|
||||
|
||||
defm VADDPSZ : avx512_fp_packed<0x58, "addps", fadd, VK16WM, VR512, v16f32, f512mem,
|
||||
memopv16f32, f32mem, loadf32, "{1to16}", SSEPackedSingle,
|
||||
SSE_ALU_ITINS_P.s, 1>, EVEX_V512, PS, EVEX_CD8<32, CD8VF>;
|
||||
|
||||
defm VADDPDZ : avx512_fp_packed<0x58, "addpd", fadd, VK8WM, VR512, v8f64, f512mem,
|
||||
memopv8f64, f64mem, loadf64, "{1to8}", SSEPackedDouble,
|
||||
SSE_ALU_ITINS_P.d, 1>,
|
||||
EVEX_V512, PD, VEX_W, EVEX_CD8<64, CD8VF>;
|
||||
|
||||
defm VMULPSZ : avx512_fp_packed<0x59, "mulps", fmul, VK16WM, VR512, v16f32, f512mem,
|
||||
memopv16f32, f32mem, loadf32, "{1to16}", SSEPackedSingle,
|
||||
SSE_ALU_ITINS_P.s, 1>, EVEX_V512, PS, EVEX_CD8<32, CD8VF>;
|
||||
defm VMULPDZ : avx512_fp_packed<0x59, "mulpd", fmul, VK8WM, VR512, v8f64, f512mem,
|
||||
memopv8f64, f64mem, loadf64, "{1to8}", SSEPackedDouble,
|
||||
SSE_ALU_ITINS_P.d, 1>,
|
||||
EVEX_V512, PD, VEX_W, EVEX_CD8<64, CD8VF>;
|
||||
|
||||
defm VMINPSZ : avx512_fp_packed<0x5D, "minps", X86fmin, VK16WM, VR512, v16f32, f512mem,
|
||||
memopv16f32, f32mem, loadf32, "{1to16}", SSEPackedSingle,
|
||||
SSE_ALU_ITINS_P.s, 1>,
|
||||
EVEX_V512, PS, EVEX_CD8<32, CD8VF>;
|
||||
defm VMAXPSZ : avx512_fp_packed<0x5F, "maxps", X86fmax, VK16WM, VR512, v16f32, f512mem,
|
||||
memopv16f32, f32mem, loadf32, "{1to16}", SSEPackedSingle,
|
||||
SSE_ALU_ITINS_P.s, 1>,
|
||||
EVEX_V512, PS, EVEX_CD8<32, CD8VF>;
|
||||
|
||||
defm VMINPDZ : avx512_fp_packed<0x5D, "minpd", X86fmin, VK8WM, VR512, v8f64, f512mem,
|
||||
memopv8f64, f64mem, loadf64, "{1to8}", SSEPackedDouble,
|
||||
SSE_ALU_ITINS_P.d, 1>,
|
||||
EVEX_V512, PD, VEX_W, EVEX_CD8<64, CD8VF>;
|
||||
defm VMAXPDZ : avx512_fp_packed<0x5F, "maxpd", X86fmax, VK8WM, VR512, v8f64, f512mem,
|
||||
memopv8f64, f64mem, loadf64, "{1to8}", SSEPackedDouble,
|
||||
SSE_ALU_ITINS_P.d, 1>,
|
||||
EVEX_V512, PD, VEX_W, EVEX_CD8<64, CD8VF>;
|
||||
|
||||
defm VSUBPSZ : avx512_fp_packed<0x5C, "subps", fsub, VK16WM, VR512, v16f32, f512mem,
|
||||
memopv16f32, f32mem, loadf32, "{1to16}", SSEPackedSingle,
|
||||
SSE_ALU_ITINS_P.s, 0>, EVEX_V512, PS, EVEX_CD8<32, CD8VF>;
|
||||
defm VDIVPSZ : avx512_fp_packed<0x5E, "divps", fdiv, VK16WM, VR512, v16f32, f512mem,
|
||||
memopv16f32, f32mem, loadf32, "{1to16}", SSEPackedSingle,
|
||||
SSE_ALU_ITINS_P.s, 0>, EVEX_V512, PS, EVEX_CD8<32, CD8VF>;
|
||||
|
||||
defm VSUBPDZ : avx512_fp_packed<0x5C, "subpd", fsub, VK8WM, VR512, v8f64, f512mem,
|
||||
memopv8f64, f64mem, loadf64, "{1to8}", SSEPackedDouble,
|
||||
SSE_ALU_ITINS_P.d, 0>,
|
||||
EVEX_V512, PD, VEX_W, EVEX_CD8<64, CD8VF>;
|
||||
defm VDIVPDZ : avx512_fp_packed<0x5E, "divpd", fdiv, VK8WM, VR512, v8f64, f512mem,
|
||||
memopv8f64, f64mem, loadf64, "{1to8}", SSEPackedDouble,
|
||||
SSE_ALU_ITINS_P.d, 0>,
|
||||
EVEX_V512, PD, VEX_W, EVEX_CD8<64, CD8VF>;
|
||||
defm VADD : avx512_fp_binop_p<0x58, "vadd", fadd, 1>;
|
||||
defm VMUL : avx512_fp_binop_p<0x59, "vmul", fmul, 1>;
|
||||
defm VMIN : avx512_fp_binop_p<0x5D, "vmin", X86fmin, 1>;
|
||||
defm VMAX : avx512_fp_binop_p<0x5F, "vmax", X86fmax, 1>;
|
||||
defm VSUB : avx512_fp_binop_p<0x5C, "vsub", fsub>;
|
||||
defm VDIV : avx512_fp_binop_p<0x5E, "vdiv", fdiv>;
|
||||
|
||||
def : Pat<(v16f32 (int_x86_avx512_mask_max_ps_512 (v16f32 VR512:$src1),
|
||||
(v16f32 VR512:$src2), (bc_v16f32 (v16i32 immAllZerosV)),
|
||||
|
|
|
|||
File diff suppressed because it is too large
Load Diff
Loading…
Reference in New Issue