add vpslldq, vpsrldq, vpsraw, vpsrad, vpsraq, vpsllw, vpslld, vpsllq, vpsrlw, vpsrld, vpsrlq
diff --git a/gen/gen_avx512.cpp b/gen/gen_avx512.cpp
index 07c6865..8ea8490 100644
--- a/gen/gen_avx512.cpp
+++ b/gen/gen_avx512.cpp
@@ -186,6 +186,7 @@
{ 0x3F, "vpmaxuq", T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1 | T_B64, false },
{ 0x39, "vpminsq", T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1 | T_B64, false },
{ 0x3B, "vpminuq", T_66 | T_0F38 | T_MUST_EVEX | T_YMM | T_EW1 | T_B64, false },
+ { 0xE2, "vpsraq", T_66 | T_0F | T_MUST_EVEX | T_YMM | T_EW1, false },
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl *p = &tbl[i];
@@ -195,6 +196,23 @@
}
}
+void putShift()
+{
+ const struct Tbl {
+ const char *name;
+ uint8 code;
+ int idx;
+ int ext_type;
+ } tbl[] = {
+ { "vpsraq", 0x72, 4, T_EW1 | T_B64 },
+ };
+ for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
+ const Tbl& p = tbl[i];
+ std::string type = type2String(T_0F | T_66 | T_YMM | T_EVEX | p.ext_type);
+ printf("void %s(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(%d), x, op, %s, 0x%02X, imm); }\n", p.name, p.idx, type.c_str(), p.code);
+ }
+}
+
int main()
{
puts("#ifndef XBYAK_DISABLE_AVX512");
@@ -203,5 +221,6 @@
putX_XM();
putM_X();
putX_X_XM_IMM();
+ putShift();
puts("#endif");
}
diff --git a/gen/gen_code.cpp b/gen/gen_code.cpp
index 973053c..44edb67 100644
--- a/gen/gen_code.cpp
+++ b/gen/gen_code.cpp
@@ -1123,15 +1123,15 @@
{ 0x09, "psignw", T_0F38 | T_66 | T_YMM, false, true },
{ 0x0A, "psignd", T_0F38 | T_66 | T_YMM, false, true },
- { 0xF1, "psllw", T_0F | T_66 | T_YMM, false, true },
- { 0xF2, "pslld", T_0F | T_66 | T_YMM, false, true },
- { 0xF3, "psllq", T_0F | T_66 | T_YMM, false, true },
+ { 0xF1, "psllw", T_0F | T_66 | T_YMM | T_EVEX, false, true },
+ { 0xF2, "pslld", T_0F | T_66 | T_YMM | T_EVEX | T_EW0, false, true },
+ { 0xF3, "psllq", T_0F | T_66 | T_YMM | T_EVEX | T_EW1, false, true },
- { 0xE1, "psraw", T_0F | T_66 | T_YMM, false, true },
- { 0xE2, "psrad", T_0F | T_66 | T_YMM, false, true },
- { 0xD1, "psrlw", T_0F | T_66 | T_YMM, false, true },
- { 0xD2, "psrld", T_0F | T_66 | T_YMM, false, true },
- { 0xD3, "psrlq", T_0F | T_66 | T_YMM, false, true },
+ { 0xE1, "psraw", T_0F | T_66 | T_YMM | T_EVEX, false, true },
+ { 0xE2, "psrad", T_0F | T_66 | T_YMM | T_EVEX | T_EW0, false, true },
+ { 0xD1, "psrlw", T_0F | T_66 | T_YMM | T_EVEX, false, true },
+ { 0xD2, "psrld", T_0F | T_66 | T_YMM | T_EVEX | T_EW0, false, true },
+ { 0xD3, "psrlq", T_0F | T_66 | T_YMM | T_EVEX | T_EW1, false, true },
{ 0xF8, "psubb", T_0F | T_66 | T_YMM, false, true },
{ 0xF9, "psubw", T_0F | T_66 | T_YMM, false, true },
@@ -1483,22 +1483,24 @@
const char *name;
uint8 code;
int idx;
+ int ext_type;
} tbl[] = {
- { "pslldq", 0x73, 7 },
- { "psrldq", 0x73, 3 },
- { "psllw", 0x71, 6 },
- { "pslld", 0x72, 6 },
- { "psllq", 0x73, 6 },
- { "psraw", 0x71, 4 },
- { "psrad", 0x72, 4 },
- { "psrlw", 0x71, 2 },
- { "psrld", 0x72, 2 },
- { "psrlq", 0x73, 2 },
+ { "pslldq", 0x73, 7, 0 },
+ { "psrldq", 0x73, 3, 0 },
+ { "psllw", 0x71, 6, 0 },
+ { "pslld", 0x72, 6, T_EW0 | T_B32 },
+ { "psllq", 0x73, 6, T_EW1 | T_B64 },
+ { "psraw", 0x71, 4, 0 },
+ { "psrad", 0x72, 4, T_EW0 | T_B32 },
+ { "psrlw", 0x71, 2, 0 },
+ { "psrld", 0x72, 2, T_EW0 | T_B32 },
+ { "psrlq", 0x73, 2, T_EW1 | T_B64 },
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl& p = tbl[i];
- printf("void v%s(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym%d : xm%d, x1, x2, T_0F | T_66 | T_YMM, 0x%02X, imm); }\n", p.name, p.idx, p.idx, p.code);
- printf("void v%s(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym%d : xm%d, x, x, T_0F | T_66 | T_YMM, 0x%02X, imm); }\n", p.name, p.idx, p.idx, p.code);
+ std::string type = type2String(T_0F | T_66 | T_YMM | T_EVEX | p.ext_type);
+ printf("void v%s(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(%d), x, op, %s, 0x%02X, imm); }\n", p.name, p.idx, type.c_str(), p.code);
+ printf("void v%s(const Xmm& x, uint8 imm) { v%s(x, x, imm); }\n", p.name, p.name);
}
}
// 4-op
diff --git a/test/make_nm.cpp b/test/make_nm.cpp
index ad6d76c..676ca16 100644
--- a/test/make_nm.cpp
+++ b/test/make_nm.cpp
@@ -1797,6 +1797,11 @@
put(p.name, XMM, XMM, IMM);
put(p.name, YMM, YMM, IMM);
put(p.name, YMM, IMM);
+ put(p.name, _ZMM, _ZMM, IMM8);
+#ifdef XBYAK64
+ put(p.name, _XMM3, _XMM3, IMM8);
+ put(p.name, _YMM3, _YMM3, IMM8);
+#endif
if (p.support_Y_Y_X) {
put(p.name, YMM, YMM, XMM);
}
@@ -2808,6 +2813,36 @@
{ "vpminuw", ZMM_KZ, _ZMM, _ZMM | _MEM },
{ "vpminud", ZMM_KZ, _ZMM, _ZMM | _MEM | M_1to16 },
{ "vpminuq", ZMM_KZ, _ZMM, _ZMM | _MEM | M_1to8 },
+
+ { "vpslldq", _XMM3, _XMM3 | _MEM, IMM8 },
+ { "vpslldq", _YMM3, _YMM3 | _MEM, IMM8 },
+ { "vpslldq", _ZMM, _ZMM | _MEM, IMM8 },
+
+ { "vpsrldq", _XMM3, _XMM3 | _MEM, IMM8 },
+ { "vpsrldq", _YMM3, _YMM3 | _MEM, IMM8 },
+ { "vpsrldq", _ZMM, _ZMM | _MEM, IMM8 },
+
+ { "vpsraw", XMM_KZ, _XMM, IMM8 },
+ { "vpsraw", ZMM_KZ, _ZMM, IMM8 },
+
+ { "vpsrad", XMM_KZ, _XMM | M_1to4, IMM8 },
+ { "vpsrad", ZMM_KZ, _ZMM | M_1to16, IMM8 },
+
+ { "vpsraq", XMM_KZ, _XMM | M_1to2, IMM8 },
+ { "vpsraq", ZMM_KZ, _ZMM | M_1to8, IMM8 },
+
+ { "vpsllw", _XMM3, _XMM3 | _MEM, IMM8 },
+ { "vpslld", _XMM3, _XMM3 | _MEM | M_1to4, IMM8 },
+ { "vpsllq", _XMM3, _XMM3 | _MEM | M_1to2, IMM8 },
+
+ { "vpsrlw", XMM_KZ, _XMM, IMM8 },
+ { "vpsrlw", ZMM_KZ, _ZMM, IMM8 },
+
+ { "vpsrld", XMM_KZ, _XMM | M_1to4, IMM8 },
+ { "vpsrld", ZMM_KZ, _ZMM | M_1to16, IMM8 },
+
+ { "vpsrlq", _XMM3, _XMM3 | _MEM | M_1to2, IMM8 },
+ { "vpsrlq", _ZMM, _ZMM | _MEM | M_1to8, IMM8 },
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl& p = tbl[i];
diff --git a/xbyak/xbyak.h b/xbyak/xbyak.h
index 3ef90c8..011d4d7 100644
--- a/xbyak/xbyak.h
+++ b/xbyak/xbyak.h
@@ -348,6 +348,7 @@
unsigned int zero_:1;
unsigned int mask_:3;
unsigned int rounding_:3;
+ void setIdx(int idx) { idx_ = idx; }
public:
enum Kind {
NONE = 0,
@@ -529,6 +530,7 @@
struct Xmm : public Mmx {
explicit Xmm(int idx = 0, Kind kind = Operand::XMM, int bit = 128) : Mmx(idx, kind, bit) { }
Xmm operator|(const EvexModifierRounding& emr) const { Xmm r(*this); r.setRounding(emr.rounding); return r; }
+ Xmm copyAndSetIdx(int idx) const { Xmm ret(*this); ret.setIdx(idx); return ret; }
};
struct Ymm : public Xmm {
@@ -1830,6 +1832,11 @@
// assert(!x.isMEM()); // QQQ
return x.isZMM() ? zm0 : x.isYMM() ? ym0 : xm0;
}
+ // get op.type(idx)
+ Xmm getXmm(const Operand& op, int idx) const
+ {
+ return op.isZMM() ? Zmm(idx) : op.isYMM() ? Ymm(idx) : Xmm(idx);
+ }
// support (x, x/m, imm), (y, y/m, imm)
void opAVX_X_XM_IMM(const Xmm& x, const Operand& op, int type, int code, int imm8 = NONE)
{
diff --git a/xbyak/xbyak_avx512.h b/xbyak/xbyak_avx512.h
index e7b2cc2..6b97878 100644
--- a/xbyak/xbyak_avx512.h
+++ b/xbyak/xbyak_avx512.h
@@ -99,4 +99,6 @@
void vpmaxuq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_66 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x3F); }
void vpminsq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_66 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x39); }
void vpminuq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_66 | T_EW1 | T_YMM | T_MUST_EVEX | T_B64, 0x3B); }
+void vpsraq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_EW1 | T_YMM | T_MUST_EVEX, 0xE2); }
+void vpsraq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(4), x, op, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x72, imm); }
#endif
diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
index 5ad88d1..e01d34e 100644
--- a/xbyak/xbyak_mnemonic.h
+++ b/xbyak/xbyak_mnemonic.h
@@ -892,21 +892,21 @@
void vpsignw(const Xmm& x, const Operand& op) { vpsignw(x, x, op); }
void vpsignd(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F38 | T_66 | T_YMM, 0x0A); }
void vpsignd(const Xmm& x, const Operand& op) { vpsignd(x, x, op); }
-void vpsllw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xF1); }
+void vpsllw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM | T_EVEX, 0xF1); }
void vpsllw(const Xmm& x, const Operand& op) { vpsllw(x, x, op); }
-void vpslld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xF2); }
+void vpslld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_EW0 | T_YMM | T_EVEX, 0xF2); }
void vpslld(const Xmm& x, const Operand& op) { vpslld(x, x, op); }
-void vpsllq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xF3); }
+void vpsllq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0xF3); }
void vpsllq(const Xmm& x, const Operand& op) { vpsllq(x, x, op); }
-void vpsraw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xE1); }
+void vpsraw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM | T_EVEX, 0xE1); }
void vpsraw(const Xmm& x, const Operand& op) { vpsraw(x, x, op); }
-void vpsrad(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xE2); }
+void vpsrad(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_EW0 | T_YMM | T_EVEX, 0xE2); }
void vpsrad(const Xmm& x, const Operand& op) { vpsrad(x, x, op); }
-void vpsrlw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xD1); }
+void vpsrlw(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM | T_EVEX, 0xD1); }
void vpsrlw(const Xmm& x, const Operand& op) { vpsrlw(x, x, op); }
-void vpsrld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xD2); }
+void vpsrld(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_EW0 | T_YMM | T_EVEX, 0xD2); }
void vpsrld(const Xmm& x, const Operand& op) { vpsrld(x, x, op); }
-void vpsrlq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xD3); }
+void vpsrlq(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0xD3); }
void vpsrlq(const Xmm& x, const Operand& op) { vpsrlq(x, x, op); }
void vpsubb(const Xmm& x1, const Xmm& x2, const Operand& op) { opAVX_X_X_XM(x1, x2, op, T_0F | T_66 | T_YMM, 0xF8); }
void vpsubb(const Xmm& x, const Operand& op) { vpsubb(x, x, op); }
@@ -1427,26 +1427,26 @@
void vpinsrd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { if (!op.isREG(32) && !op.isMEM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XMcvt(x1, x2, op, !op.isMEM(), Operand::XMM, T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0, 0x22, imm); }
void vpinsrd(const Xmm& x, const Operand& op, uint8 imm) { vpinsrd(x, x, op, imm); }
void vpmovmskb(const Reg32e& r, const Xmm& x) { bool isYMM= x.isYMM(); opAVX_X_X_XM(isYMM ? Ymm(r.getIdx()) : Xmm(r.getIdx()), isYMM ? ym0 : xm0, x, T_0F | T_66 | T_YMM, 0xD7); }
-void vpslldq(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym7 : xm7, x1, x2, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpslldq(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym7 : xm7, x, x, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpsrldq(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym3 : xm3, x1, x2, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpsrldq(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym3 : xm3, x, x, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpsllw(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym6 : xm6, x1, x2, T_0F | T_66 | T_YMM, 0x71, imm); }
-void vpsllw(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym6 : xm6, x, x, T_0F | T_66 | T_YMM, 0x71, imm); }
-void vpslld(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym6 : xm6, x1, x2, T_0F | T_66 | T_YMM, 0x72, imm); }
-void vpslld(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym6 : xm6, x, x, T_0F | T_66 | T_YMM, 0x72, imm); }
-void vpsllq(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym6 : xm6, x1, x2, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpsllq(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym6 : xm6, x, x, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpsraw(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym4 : xm4, x1, x2, T_0F | T_66 | T_YMM, 0x71, imm); }
-void vpsraw(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym4 : xm4, x, x, T_0F | T_66 | T_YMM, 0x71, imm); }
-void vpsrad(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym4 : xm4, x1, x2, T_0F | T_66 | T_YMM, 0x72, imm); }
-void vpsrad(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym4 : xm4, x, x, T_0F | T_66 | T_YMM, 0x72, imm); }
-void vpsrlw(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym2 : xm2, x1, x2, T_0F | T_66 | T_YMM, 0x71, imm); }
-void vpsrlw(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym2 : xm2, x, x, T_0F | T_66 | T_YMM, 0x71, imm); }
-void vpsrld(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym2 : xm2, x1, x2, T_0F | T_66 | T_YMM, 0x72, imm); }
-void vpsrld(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym2 : xm2, x, x, T_0F | T_66 | T_YMM, 0x72, imm); }
-void vpsrlq(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym2 : xm2, x1, x2, T_0F | T_66 | T_YMM, 0x73, imm); }
-void vpsrlq(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym2 : xm2, x, x, T_0F | T_66 | T_YMM, 0x73, imm); }
+void vpslldq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(7), x, op, T_0F | T_66 | T_YMM | T_EVEX, 0x73, imm); }
+void vpslldq(const Xmm& x, uint8 imm) { vpslldq(x, x, imm); }
+void vpsrldq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(3), x, op, T_0F | T_66 | T_YMM | T_EVEX, 0x73, imm); }
+void vpsrldq(const Xmm& x, uint8 imm) { vpsrldq(x, x, imm); }
+void vpsllw(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(6), x, op, T_0F | T_66 | T_YMM | T_EVEX, 0x71, imm); }
+void vpsllw(const Xmm& x, uint8 imm) { vpsllw(x, x, imm); }
+void vpslld(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(6), x, op, T_0F | T_66 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x72, imm); }
+void vpslld(const Xmm& x, uint8 imm) { vpslld(x, x, imm); }
+void vpsllq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(6), x, op, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x73, imm); }
+void vpsllq(const Xmm& x, uint8 imm) { vpsllq(x, x, imm); }
+void vpsraw(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(4), x, op, T_0F | T_66 | T_YMM | T_EVEX, 0x71, imm); }
+void vpsraw(const Xmm& x, uint8 imm) { vpsraw(x, x, imm); }
+void vpsrad(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(4), x, op, T_0F | T_66 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x72, imm); }
+void vpsrad(const Xmm& x, uint8 imm) { vpsrad(x, x, imm); }
+void vpsrlw(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(2), x, op, T_0F | T_66 | T_YMM | T_EVEX, 0x71, imm); }
+void vpsrlw(const Xmm& x, uint8 imm) { vpsrlw(x, x, imm); }
+void vpsrld(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(2), x, op, T_0F | T_66 | T_EW0 | T_YMM | T_EVEX | T_B32, 0x72, imm); }
+void vpsrld(const Xmm& x, uint8 imm) { vpsrld(x, x, imm); }
+void vpsrlq(const Xmm& x, const Operand& op, uint8 imm) { opAVX_X_X_XM(x.copyAndSetIdx(2), x, op, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_B64, 0x73, imm); }
+void vpsrlq(const Xmm& x, uint8 imm) { vpsrlq(x, x, imm); }
void vblendvpd(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_YMM, 0x4B, x4.getIdx() << 4); }
void vblendvpd(const Xmm& x1, const Operand& op, const Xmm& x4) { vblendvpd(x1, x1, op, x4); }
void vblendvps(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_YMM, 0x4A, x4.getIdx() << 4); }
