vbroadcastsd, vbroadcastss, vpbroadcast{b,w,d,q}
diff --git a/gen/gen_code.cpp b/gen/gen_code.cpp
index 141f058..4b19075 100644
--- a/gen/gen_code.cpp
+++ b/gen/gen_code.cpp
@@ -1003,11 +1003,11 @@
};
for (int i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl *p = &tbl[i];
- printf("void v%spd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x%02X); }\n", p->name, p->code);
- printf("void v%sps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x%02X); }\n", p->name, p->code);
+ printf("void v%spd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x%02X); }\n", p->name, p->code);
+ printf("void v%sps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x%02X); }\n", p->name, p->code);
if (p->only_pd_ps) continue;
- printf("void v%ssd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x%02X); }\n", p->name, p->code);
- printf("void v%sss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x%02X); }\n", p->name, p->code);
+ printf("void v%ssd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x%02X); }\n", p->name, p->code);
+ printf("void v%sss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x%02X); }\n", p->name, p->code);
}
}
// (x, x, x/m[, imm]) or (y, y, y/m[, imm])
@@ -1431,20 +1431,21 @@
printf("void vbroadcastf128(const Ymm& y, const Address& addr) { opAVX_X_XM_IMM(y, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x1A); }\n");
printf("void vbroadcasti128(const Ymm& y, const Address& addr) { opAVX_X_XM_IMM(y, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x5A); }\n");
- printf("void vbroadcastsd(const Ymm& y, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(y, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x19); }\n");
+ printf("void vbroadcastsd(const Ymm& y, const Operand& op) { if (!op.isMEM() && !(y.isYMM() && op.isXMM()) && !(y.isZMM() && op.isXMM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(y, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX | T_EW1, 0x19); }\n");
const struct Tbl {
const char *name;
uint8 code;
+ bool ew1;
} tbl[] = {
- { "vbroadcastss", 0x18 },
- { "vpbroadcastb", 0x78 },
- { "vpbroadcastw", 0x79 },
- { "vpbroadcastd", 0x58 },
- { "vpbroadcastq", 0x59 },
+ { "vbroadcastss", 0x18, false },
+ { "vpbroadcastb", 0x78, false },
+ { "vpbroadcastw", 0x79, false },
+ { "vpbroadcastd", 0x58, false },
+ { "vpbroadcastq", 0x59, true },
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl& p = tbl[i];
- printf("void %s(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x%02X); }\n", p.name, p.code);
+ printf("void %s(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX%s, 0x%02X); }\n", p.name, p.ew1 ? "|T_EW1" : "", p.code);
}
printf("void vextractf128(const Operand& op, const Ymm& y, uint8 imm) { opAVX_X_X_XMcvt(y, y.isXMM() ? xm0 : ym0, op, op.isXMM(), Operand::YMM, T_0F3A | T_66 | T_W0 | T_YMM, 0x19, imm); }\n");
diff --git a/test/make_nm.cpp b/test/make_nm.cpp
index 71c257a..fdfd2c9 100644
--- a/test/make_nm.cpp
+++ b/test/make_nm.cpp
@@ -79,11 +79,13 @@
const uint64 YMM = _YMM | _YMM2;
const uint64 K = 1ULL << 43;
const uint64 _ZMM = 1ULL << 44;
-const uint64 _ZMM2 = 1ULL << 45; // max value
+const uint64 _ZMM2 = 1ULL << 45;
#ifdef XBYAK64
const uint64 ZMM = _ZMM | _ZMM2;
+const uint64 _YMM3 = 1ULL << 46; // max value
#else
const uint64 ZMM = _ZMM;
+const uint64 _YMM3 = 0;
#endif
const uint64 NOPARA = 1ULL << (bitEnd - 1);
@@ -195,6 +197,13 @@
};
return tbl[idx];
}
+ case _YMM3:
+ {
+ static const char tbl[][6] = {
+ "ymm16", "ymm17", "ymm18", "ymm19", "ymm20", "ymm21", "ymm22", "ymm23",
+ };
+ return tbl[idx];
+ }
case _ZMM2:
{
static const char tbl[][6] = {
@@ -1917,7 +1926,8 @@
put("vbroadcastf128", YMM, MEM);
put("vbroadcasti128", YMM, MEM);
- put("vbroadcastsd", YMM, XMM|MEM);
+ put("vbroadcastsd", YMM|_YMM3, XMM|MEM);
+ put("vbroadcastsd", ZMM, XMM|MEM);
{
const char *tbl[] = {
"vbroadcastss",
@@ -1927,7 +1937,7 @@
"vpbroadcastq",
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
- put(tbl[i], XMM | YMM, XMM|MEM);
+ put(tbl[i], XMM | YMM | ZMM, XMM|MEM);
}
}
diff --git a/xbyak/xbyak.h b/xbyak/xbyak.h
index aa5b642..324aa62 100644
--- a/xbyak/xbyak.h
+++ b/xbyak/xbyak.h
@@ -169,6 +169,8 @@
ERR_ROUNDING_IS_ALREADY_SET,
ERR_K0_IS_INVALID,
ERR_EVEX_IS_INVALID,
+ ERR_SAE_IS_INVALID,
+ ERR_ER_IS_INVALID,
ERR_INTERNAL
};
@@ -223,6 +225,8 @@
"rounding is already set",
"k0 is invalid",
"evex is invalid",
+ "sae(suppress all exceptions) is invalid",
+ "er(embedded rounding) is invalid",
"internal error",
};
assert((size_t)err_ < sizeof(errTbl) / sizeof(*errTbl));
@@ -524,11 +528,11 @@
namespace inner {
enum SAEtype {
- T_SAE,
T_RN_SAE = 1,
T_RD_SAE = 2,
T_RU_SAE = 3,
T_RZ_SAE = 4,
+ T_SAE = 5
};
} // inner
@@ -1364,6 +1368,8 @@
T_EW1 = 1 << 14,
T_YMM = 1 << 15,
T_EVEX = 1 << 16,
+ T_ER = 1 << 17,
+ T_SAE = 1 << 18
};
void vex(const Reg& reg, const Reg& base, const Operand *v, int type, int code, bool x = false)
{
@@ -1402,6 +1408,8 @@
if (reg.isZMM()) {
int rounding = base.getRounding();
if (rounding) {
+ if (rounding == inner::T_SAE && !(type & T_SAE)) throw Error(ERR_SAE_IS_INVALID);
+ if (rounding != inner::T_SAE && !(type & T_ER)) throw Error(ERR_ER_IS_INVALID);
LL = rounding - 1;
b = true;
}
@@ -1781,7 +1789,7 @@
// 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)
{
- opAVX_X_X_XM(x, x.isXMM() ? xm0 : ym0, op, type, code, imm8);
+ opAVX_X_X_XM(x, x.isXMM() ? xm0 : x.isYMM() ? ym0 : zm0, op, type, code, imm8);
}
// QQQ:need to refactor
void opSp1(const Reg& reg, const Operand& op, uint8 pref, uint8 code0, uint8 code1)
diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
index c1a0ffb..4e9ebba 100644
--- a/xbyak/xbyak_mnemonic.h
+++ b/xbyak/xbyak_mnemonic.h
@@ -704,38 +704,38 @@
void fucom(const Fpu& reg) { opFpu(reg, 0xDD, 0xE0); }
void fucomp(const Fpu& reg) { opFpu(reg, 0xDD, 0xE8); }
void fxch(const Fpu& reg) { opFpu(reg, 0xD9, 0xC8); }
-void vaddpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x58); }
-void vaddps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x58); }
-void vaddsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x58); }
-void vaddss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x58); }
-void vsubpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x5C); }
-void vsubps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x5C); }
-void vsubsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x5C); }
-void vsubss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x5C); }
-void vmulpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x59); }
-void vmulps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x59); }
-void vmulsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x59); }
-void vmulss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x59); }
-void vdivpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x5E); }
-void vdivps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x5E); }
-void vdivsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x5E); }
-void vdivss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x5E); }
-void vmaxpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x5F); }
-void vmaxps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x5F); }
-void vmaxsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x5F); }
-void vmaxss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x5F); }
-void vminpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x5D); }
-void vminps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x5D); }
-void vminsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX, 0x5D); }
-void vminss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX, 0x5D); }
-void vandpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x54); }
-void vandps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x54); }
-void vandnpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x55); }
-void vandnps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x55); }
-void vorpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x56); }
-void vorps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x56); }
-void vxorpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX, 0x57); }
-void vxorps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX, 0x57); }
+void vaddpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x58); }
+void vaddps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x58); }
+void vaddsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x58); }
+void vaddss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x58); }
+void vsubpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x5C); }
+void vsubps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x5C); }
+void vsubsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x5C); }
+void vsubss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x5C); }
+void vmulpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x59); }
+void vmulps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x59); }
+void vmulsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x59); }
+void vmulss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x59); }
+void vdivpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x5E); }
+void vdivps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x5E); }
+void vdivsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x5E); }
+void vdivss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x5E); }
+void vmaxpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x5F); }
+void vmaxps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x5F); }
+void vmaxsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x5F); }
+void vmaxss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x5F); }
+void vminpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x5D); }
+void vminps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x5D); }
+void vminsd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F2 | T_EW1 | T_EVEX | T_ER, 0x5D); }
+void vminss(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_F3 | T_EW0 | T_EVEX | T_ER, 0x5D); }
+void vandpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x54); }
+void vandps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x54); }
+void vandnpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x55); }
+void vandnps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x55); }
+void vorpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x56); }
+void vorps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x56); }
+void vxorpd(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_66 | T_EW1 | T_YMM | T_EVEX | T_ER, 0x57); }
+void vxorps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_0F | T_EW0 | T_YMM | T_EVEX | T_ER, 0x57); }
void vblendpd(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x0D, imm); }
void vblendpd(const Xmm& x, const Operand& op, uint8 imm) { vblendpd(x, x, op, imm); }
void vblendps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x0C, imm); }
@@ -1400,12 +1400,12 @@
void vaesimc(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0, 0xDB); }
void vbroadcastf128(const Ymm& y, const Address& addr) { opAVX_X_XM_IMM(y, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x1A); }
void vbroadcasti128(const Ymm& y, const Address& addr) { opAVX_X_XM_IMM(y, addr, T_0F38 | T_66 | T_W0 | T_YMM, 0x5A); }
-void vbroadcastsd(const Ymm& y, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(y, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x19); }
-void vbroadcastss(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x18); }
-void vpbroadcastb(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x78); }
-void vpbroadcastw(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x79); }
-void vpbroadcastd(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x58); }
-void vpbroadcastq(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM, 0x59); }
+void vbroadcastsd(const Ymm& y, const Operand& op) { if (!op.isMEM() && !(y.isYMM() && op.isXMM()) && !(y.isZMM() && op.isXMM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(y, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX | T_EW1, 0x19); }
+void vbroadcastss(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX, 0x18); }
+void vpbroadcastb(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX, 0x78); }
+void vpbroadcastw(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX, 0x79); }
+void vpbroadcastd(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX, 0x58); }
+void vpbroadcastq(const Xmm& x, const Operand& op) { if (!(op.isXMM() || op.isMEM())) throw Error(ERR_BAD_COMBINATION); opAVX_X_XM_IMM(x, op, T_0F38 | T_66 | T_W0 | T_YMM | T_EVEX|T_EW1, 0x59); }
void vextractf128(const Operand& op, const Ymm& y, uint8 imm) { opAVX_X_X_XMcvt(y, y.isXMM() ? xm0 : ym0, op, op.isXMM(), Operand::YMM, T_0F3A | T_66 | T_W0 | T_YMM, 0x19, imm); }
void vextracti128(const Operand& op, const Ymm& y, uint8 imm) { opAVX_X_X_XMcvt(y, y.isXMM() ? xm0 : ym0, op, op.isXMM(), Operand::YMM, T_0F3A | T_66 | T_W0 | T_YMM, 0x39, imm); }
void vextractps(const Operand& op, const Xmm& x, uint8 imm) { if (!(op.isREG(32) || op.isMEM()) || x.isYMM()) throw Error(ERR_BAD_COMBINATION); opAVX_X_X_XMcvt(x, x.isXMM() ? xm0 : ym0, op, op.isREG(), Operand::XMM, T_0F3A | T_66 | T_W0, 0x17, imm); }
