//===----------------------------------------------------------------------===//
multiclass sse12_mov_hilo_packed<bits<8>opc, RegisterClass RC,
- PatFrag mov_frag, string base_opc,
+ SDNode psnode, SDNode pdnode, string base_opc,
string asm_opr> {
def PSrm : PI<opc, MRMSrcMem,
(outs VR128:$dst), (ins VR128:$src1, f64mem:$src2),
!strconcat(base_opc, "s", asm_opr),
[(set RC:$dst,
- (mov_frag RC:$src1,
+ (psnode RC:$src1,
(bc_v4f32 (v2f64 (scalar_to_vector (loadf64 addr:$src2))))))],
IIC_DEFAULT, SSEPackedSingle>, TB;
def PDrm : PI<opc, MRMSrcMem,
(outs RC:$dst), (ins RC:$src1, f64mem:$src2),
!strconcat(base_opc, "d", asm_opr),
- [(set RC:$dst, (v2f64 (mov_frag RC:$src1,
+ [(set RC:$dst, (v2f64 (pdnode RC:$src1,
(scalar_to_vector (loadf64 addr:$src2)))))],
IIC_DEFAULT, SSEPackedDouble>, TB, OpSize;
}
let AddedComplexity = 20 in {
- defm VMOVL : sse12_mov_hilo_packed<0x12, VR128, movlp, "movlp",
+ defm VMOVL : sse12_mov_hilo_packed<0x12, VR128, X86Movlps, X86Movlpd, "movlp",
"\t{$src2, $src1, $dst|$dst, $src1, $src2}">, VEX_4V;
}
let Constraints = "$src1 = $dst", AddedComplexity = 20 in {
- defm MOVL : sse12_mov_hilo_packed<0x12, VR128, movlp, "movlp",
+ defm MOVL : sse12_mov_hilo_packed<0x12, VR128, X86Movlps, X86Movlpd, "movlp",
"\t{$src2, $dst|$dst, $src2}">;
}
(iPTR 0))), addr:$dst)]>;
let Predicates = [HasAVX] in {
- let AddedComplexity = 20 in {
- // vector_shuffle v1, (load v2) <4, 5, 2, 3> using MOVLPS
- def : Pat<(v4f32 (movlp VR128:$src1, (load addr:$src2))),
- (VMOVLPSrm VR128:$src1, addr:$src2)>;
- def : Pat<(v4i32 (movlp VR128:$src1, (load addr:$src2))),
- (VMOVLPSrm VR128:$src1, addr:$src2)>;
- // vector_shuffle v1, (load v2) <2, 1> using MOVLPS
- def : Pat<(v2f64 (movlp VR128:$src1, (load addr:$src2))),
- (VMOVLPDrm VR128:$src1, addr:$src2)>;
- def : Pat<(v2i64 (movlp VR128:$src1, (load addr:$src2))),
- (VMOVLPDrm VR128:$src1, addr:$src2)>;
- }
-
- // (store (vector_shuffle (load addr), v2, <4, 5, 2, 3>), addr) using MOVLPS
- def : Pat<(store (v4f32 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
- (VMOVLPSmr addr:$src1, VR128:$src2)>;
- def : Pat<(store (v4i32 (movlp (bc_v4i32 (loadv2i64 addr:$src1)),
- VR128:$src2)), addr:$src1),
- (VMOVLPSmr addr:$src1, VR128:$src2)>;
-
- // (store (vector_shuffle (load addr), v2, <2, 1>), addr) using MOVLPS
- def : Pat<(store (v2f64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
- (VMOVLPDmr addr:$src1, VR128:$src2)>;
- def : Pat<(store (v2i64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
- (VMOVLPDmr addr:$src1, VR128:$src2)>;
-
// Shuffle with VMOVLPS
def : Pat<(v4f32 (X86Movlps VR128:$src1, (load addr:$src2))),
(VMOVLPSrm VR128:$src1, addr:$src2)>;
def : Pat<(v4i32 (X86Movlps VR128:$src1, (load addr:$src2))),
(VMOVLPSrm VR128:$src1, addr:$src2)>;
- def : Pat<(X86Movlps VR128:$src1,
- (bc_v4f32 (v2f64 (scalar_to_vector (loadf64 addr:$src2))))),
- (VMOVLPSrm VR128:$src1, addr:$src2)>;
// Shuffle with VMOVLPD
def : Pat<(v2f64 (X86Movlpd VR128:$src1, (load addr:$src2))),
(VMOVLPDrm VR128:$src1, addr:$src2)>;
def : Pat<(v2i64 (X86Movlpd VR128:$src1, (load addr:$src2))),
(VMOVLPDrm VR128:$src1, addr:$src2)>;
- def : Pat<(v2f64 (X86Movlpd VR128:$src1,
- (scalar_to_vector (loadf64 addr:$src2)))),
- (VMOVLPDrm VR128:$src1, addr:$src2)>;
// Store patterns
def : Pat<(store (v4f32 (X86Movlps (load addr:$src1), VR128:$src2)),
}
let Predicates = [HasSSE1] in {
- let AddedComplexity = 20 in {
- // vector_shuffle v1, (load v2) <4, 5, 2, 3> using MOVLPS
- def : Pat<(v4f32 (movlp VR128:$src1, (load addr:$src2))),
- (MOVLPSrm VR128:$src1, addr:$src2)>;
- def : Pat<(v4i32 (movlp VR128:$src1, (load addr:$src2))),
- (MOVLPSrm VR128:$src1, addr:$src2)>;
- }
-
// (store (vector_shuffle (load addr), v2, <4, 5, 2, 3>), addr) using MOVLPS
def : Pat<(store (i64 (vector_extract (bc_v2i64 (v4f32 VR128:$src2)),
(iPTR 0))), addr:$src1),
(MOVLPSmr addr:$src1, VR128:$src2)>;
- def : Pat<(store (v4f32 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
- (MOVLPSmr addr:$src1, VR128:$src2)>;
- def : Pat<(store (v4i32 (movlp (bc_v4i32 (loadv2i64 addr:$src1)),
- VR128:$src2)), addr:$src1),
- (MOVLPSmr addr:$src1, VR128:$src2)>;
// Shuffle with MOVLPS
def : Pat<(v4f32 (X86Movlps VR128:$src1, (load addr:$src2))),
(MOVLPSrm VR128:$src1, addr:$src2)>;
def : Pat<(v4i32 (X86Movlps VR128:$src1, (load addr:$src2))),
(MOVLPSrm VR128:$src1, addr:$src2)>;
- def : Pat<(X86Movlps VR128:$src1,
- (bc_v4f32 (v2f64 (scalar_to_vector (loadf64 addr:$src2))))),
- (MOVLPSrm VR128:$src1, addr:$src2)>;
def : Pat<(X86Movlps VR128:$src1,
(bc_v4f32 (v2i64 (scalar_to_vector (loadi64 addr:$src2))))),
(MOVLPSrm VR128:$src1, addr:$src2)>;
}
let Predicates = [HasSSE2] in {
- let AddedComplexity = 20 in {
- // vector_shuffle v1, (load v2) <2, 1> using MOVLPS
- def : Pat<(v2f64 (movlp VR128:$src1, (load addr:$src2))),
- (MOVLPDrm VR128:$src1, addr:$src2)>;
- def : Pat<(v2i64 (movlp VR128:$src1, (load addr:$src2))),
- (MOVLPDrm VR128:$src1, addr:$src2)>;
- }
-
- // (store (vector_shuffle (load addr), v2, <2, 1>), addr) using MOVLPS
- def : Pat<(store (v2f64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
- (MOVLPDmr addr:$src1, VR128:$src2)>;
- def : Pat<(store (v2i64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
- (MOVLPDmr addr:$src1, VR128:$src2)>;
-
// Shuffle with MOVLPD
def : Pat<(v2f64 (X86Movlpd VR128:$src1, (load addr:$src2))),
(MOVLPDrm VR128:$src1, addr:$src2)>;
def : Pat<(v2i64 (X86Movlpd VR128:$src1, (load addr:$src2))),
(MOVLPDrm VR128:$src1, addr:$src2)>;
- def : Pat<(v2f64 (X86Movlpd VR128:$src1,
- (scalar_to_vector (loadf64 addr:$src2)))),
- (MOVLPDrm VR128:$src1, addr:$src2)>;
// Store patterns
def : Pat<(store (v2f64 (X86Movlpd (load addr:$src1), VR128:$src2)),
//===----------------------------------------------------------------------===//
let AddedComplexity = 20 in {
- defm VMOVH : sse12_mov_hilo_packed<0x16, VR128, movlhps, "movhp",
+ defm VMOVH : sse12_mov_hilo_packed<0x16, VR128, X86Movlhps, X86Movlhpd, "movhp",
"\t{$src2, $src1, $dst|$dst, $src1, $src2}">, VEX_4V;
}
let Constraints = "$src1 = $dst", AddedComplexity = 20 in {
- defm MOVH : sse12_mov_hilo_packed<0x16, VR128, movlhps, "movhp",
+ defm MOVH : sse12_mov_hilo_packed<0x16, VR128, X86Movlhps, X86Movlhpd, "movhp",
"\t{$src2, $dst|$dst, $src2}">;
}
let Predicates = [HasAVX] in {
// VMOVHPS patterns
- def : Pat<(movlhps VR128:$src1, (bc_v4i32 (v2i64 (X86vzload addr:$src2)))),
- (VMOVHPSrm (v4i32 VR128:$src1), addr:$src2)>;
- def : Pat<(X86Movlhps VR128:$src1,
- (bc_v4f32 (v2f64 (scalar_to_vector (loadf64 addr:$src2))))),
- (VMOVHPSrm VR128:$src1, addr:$src2)>;
def : Pat<(X86Movlhps VR128:$src1,
(bc_v4f32 (v2i64 (scalar_to_vector (loadi64 addr:$src2))))),
(VMOVHPSrm VR128:$src1, addr:$src2)>;
def : Pat<(v2f64 (X86Unpckl VR128:$src1,
(scalar_to_vector (loadf64 addr:$src2)))),
(VMOVHPDrm VR128:$src1, addr:$src2)>;
-
- // FIXME: This should be matched by a X86Movhpd instead. Same as above
- def : Pat<(v2f64 (X86Movlhpd VR128:$src1,
- (scalar_to_vector (loadf64 addr:$src2)))),
- (VMOVHPDrm VR128:$src1, addr:$src2)>;
}
let Predicates = [HasSSE1] in {
// MOVHPS patterns
- def : Pat<(movlhps VR128:$src1, (bc_v4i32 (v2i64 (X86vzload addr:$src2)))),
- (MOVHPSrm (v4i32 VR128:$src1), addr:$src2)>;
- def : Pat<(X86Movlhps VR128:$src1,
- (bc_v4f32 (v2f64 (scalar_to_vector (loadf64 addr:$src2))))),
- (MOVHPSrm VR128:$src1, addr:$src2)>;
def : Pat<(X86Movlhps VR128:$src1,
(bc_v4f32 (v2i64 (scalar_to_vector (loadi64 addr:$src2))))),
(MOVHPSrm VR128:$src1, addr:$src2)>;
def : Pat<(v2f64 (X86Unpckl VR128:$src1,
(scalar_to_vector (loadf64 addr:$src2)))),
(MOVHPDrm VR128:$src1, addr:$src2)>;
-
- // FIXME: This should be matched by a X86Movhpd instead. Same as above
- def : Pat<(v2f64 (X86Movlhpd VR128:$src1,
- (scalar_to_vector (loadf64 addr:$src2)))),
- (MOVHPDrm VR128:$src1, addr:$src2)>;
}
//===----------------------------------------------------------------------===//