eden-emu/xbyak
19
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

add vsqrtph

Browse source
This commit is contained in:
MITSUNARI Shigeo 2021-09-07 14:48:33 +09:00
parent 52765d54f7
commit e00b508d5f
3 changed files with 15 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
gen/gen_avx512.cpp
View file

@ -662,6 +662,7 @@ void putX_XM_IMM()
{ 0x4E, "vrsqrt14ps", T_66 | T_0F38 | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, false },
{ 0x4E, "vrsqrtph", T_66 | T_MAP6 | T_YMM | T_MUST_EVEX | T_EW0 | T_B16, false },
{ 0x51, "vsqrtph", T_MAP5| T_YMM | T_MUST_EVEX | T_EW0 | T_ER_Z | T_B16, false },
{ 0x09, "vrndscalepd", T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, true },
{ 0x08, "vrndscaleps", T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, true },

13
test/misc.cpp
View file

@ -985,6 +985,12 @@ CYBOZU_TEST_AUTO(vaddph)
vrsqrtph(zmm2, ptr_b [rax+0x40]);
vrsqrtsh(xmm1|k5|T_z, xmm7, ptr [rax+0x40]);
vsqrtph(xmm1|k4|T_z, ptr [rax+0x40]);
vsqrtph(xmm1|k4|T_z, ptr_b [rax+0x40]);
vsqrtph(ymm1|k4|T_z, ptr_b [rax+0x40]);
vsqrtph(zmm1|k4|T_z, ptr [rax+0x40]);
vsqrtph(zmm1|k4|T_z, ptr_b [rax+0x40]);
}
} c;
const uint8_t tbl[] = {
@ -1126,6 +1132,13 @@ CYBOZU_TEST_AUTO(vaddph)
// vrsqrtsh
0x62, 0xf6, 0x45, 0x8d, 0x4f, 0x48, 0x20,
// vsqrtph
0x62, 0xf5, 0x7c, 0x8c, 0x51, 0x48, 0x04,
0x62, 0xf5, 0x7c, 0x9c, 0x51, 0x48, 0x20,
0x62, 0xf5, 0x7c, 0xbc, 0x51, 0x48, 0x20,
0x62, 0xf5, 0x7c, 0xcc, 0x51, 0x48, 0x01,
0x62, 0xf5, 0x7c, 0xdc, 0x51, 0x48, 0x20,
};
const size_t n = sizeof(tbl) / sizeof(tbl[0]);
CYBOZU_TEST_EQUAL(c.getSize(), n);

1
xbyak/xbyak_mnemonic.h
View file

@ -2223,6 +2223,7 @@ void vshuff32x4(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) {
void vshuff64x2(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, 0x23, imm); }
void vshufi32x4(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW0 | T_B32, 0x43, imm); }
void vshufi64x2(const Ymm& y1, const Ymm& y2, const Operand& op, uint8_t imm) { opAVX_X_X_XM(y1, y2, op, T_66 | T_0F3A | T_YMM | T_MUST_EVEX | T_EW1 | T_B64, 0x43, imm); }
void vsqrtph(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_EW0 | T_YMM | T_ER_Z | T_MUST_EVEX | T_B16, 0x51); }
void vsubph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5C); }
void vsubsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5C); }
void vucomish(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_MAP5 | T_MUST_EVEX | T_EW0 | T_SAE_X | T_N2, 0x2E); }