603 lines
20 KiB
C++
603 lines
20 KiB
C++
/////////////////////////////////////////////////////////////////////////
|
|
// $Id$
|
|
/////////////////////////////////////////////////////////////////////////
|
|
//
|
|
// Copyright (c) 2011-2013 Stanislav Shwartsman
|
|
// Written by Stanislav Shwartsman [sshwarts at sourceforge net]
|
|
//
|
|
// This library is free software; you can redistribute it and/or
|
|
// modify it under the terms of the GNU Lesser General Public
|
|
// License as published by the Free Software Foundation; either
|
|
// version 2 of the License, or (at your option) any later version.
|
|
//
|
|
// This library is distributed in the hope that it will be useful,
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
// Lesser General Public License for more details.
|
|
//
|
|
// You should have received a copy of the GNU Lesser General Public
|
|
// License along with this library; if not, write to the Free Software
|
|
// Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA B 02110-1301 USA
|
|
//
|
|
/////////////////////////////////////////////////////////////////////////
|
|
|
|
#define NEED_CPU_REG_SHORTCUTS 1
|
|
#include "bochs.h"
|
|
#include "cpu.h"
|
|
#define LOG_THIS BX_CPU_THIS_PTR
|
|
|
|
#if BX_SUPPORT_AVX
|
|
|
|
#include "simd_int.h"
|
|
#include "simd_compare.h"
|
|
|
|
#define AVX_2OP(HANDLER, func) \
|
|
/* AVX instruction with two src operands */ \
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C :: HANDLER (bxInstruction_c *i) \
|
|
{ \
|
|
BxPackedAvxRegister op1 = BX_READ_AVX_REG(i->src1()), op2 = BX_READ_AVX_REG(i->src2()); \
|
|
unsigned len = i->getVL(); \
|
|
\
|
|
for (unsigned n=0; n < len; n++) \
|
|
(func)(&op1.vmm128(n), &op2.vmm128(n)); \
|
|
\
|
|
BX_WRITE_AVX_REGZ(i->dst(), op1, len); \
|
|
\
|
|
BX_NEXT_INSTR(i); \
|
|
}
|
|
|
|
AVX_2OP(VANDPS_VpsHpsWpsR, xmm_andps)
|
|
AVX_2OP(VANDNPS_VpsHpsWpsR, xmm_andnps)
|
|
AVX_2OP(VXORPS_VpsHpsWpsR, xmm_xorps)
|
|
AVX_2OP(VORPS_VpsHpsWpsR, xmm_orps)
|
|
|
|
AVX_2OP(VUNPCKLPS_VpsHpsWpsR, xmm_unpcklps)
|
|
AVX_2OP(VUNPCKLPD_VpdHpdWpdR, xmm_unpcklpd)
|
|
AVX_2OP(VUNPCKHPS_VpsHpsWpsR, xmm_unpckhps)
|
|
AVX_2OP(VUNPCKHPD_VpdHpdWpdR, xmm_unpckhpd)
|
|
|
|
AVX_2OP(VPADDB_VdqHdqWdqR, xmm_paddb)
|
|
AVX_2OP(VPADDW_VdqHdqWdqR, xmm_paddw)
|
|
AVX_2OP(VPADDD_VdqHdqWdqR, xmm_paddd)
|
|
AVX_2OP(VPADDQ_VdqHdqWdqR, xmm_paddq)
|
|
AVX_2OP(VPSUBB_VdqHdqWdqR, xmm_psubb)
|
|
AVX_2OP(VPSUBW_VdqHdqWdqR, xmm_psubw)
|
|
AVX_2OP(VPSUBD_VdqHdqWdqR, xmm_psubd)
|
|
AVX_2OP(VPSUBQ_VdqHdqWdqR, xmm_psubq)
|
|
|
|
AVX_2OP(VPCMPEQB_VdqHdqWdqR, xmm_pcmpeqb)
|
|
AVX_2OP(VPCMPEQW_VdqHdqWdqR, xmm_pcmpeqw)
|
|
AVX_2OP(VPCMPEQD_VdqHdqWdqR, xmm_pcmpeqd)
|
|
AVX_2OP(VPCMPEQQ_VdqHdqWdqR, xmm_pcmpeqq)
|
|
AVX_2OP(VPCMPGTB_VdqHdqWdqR, xmm_pcmpgtb)
|
|
AVX_2OP(VPCMPGTW_VdqHdqWdqR, xmm_pcmpgtw)
|
|
AVX_2OP(VPCMPGTD_VdqHdqWdqR, xmm_pcmpgtd)
|
|
AVX_2OP(VPCMPGTQ_VdqHdqWdqR, xmm_pcmpgtq)
|
|
|
|
AVX_2OP(VPMINSB_VdqHdqWdqR, xmm_pminsb)
|
|
AVX_2OP(VPMINSW_VdqHdqWdqR, xmm_pminsw)
|
|
AVX_2OP(VPMINSD_VdqHdqWdqR, xmm_pminsd)
|
|
AVX_2OP(VPMINSQ_VdqHdqWdqR, xmm_pminsq)
|
|
AVX_2OP(VPMINUB_VdqHdqWdqR, xmm_pminub)
|
|
AVX_2OP(VPMINUW_VdqHdqWdqR, xmm_pminuw)
|
|
AVX_2OP(VPMINUD_VdqHdqWdqR, xmm_pminud)
|
|
AVX_2OP(VPMINUQ_VdqHdqWdqR, xmm_pminuq)
|
|
AVX_2OP(VPMAXSB_VdqHdqWdqR, xmm_pmaxsb)
|
|
AVX_2OP(VPMAXSW_VdqHdqWdqR, xmm_pmaxsw)
|
|
AVX_2OP(VPMAXSD_VdqHdqWdqR, xmm_pmaxsd)
|
|
AVX_2OP(VPMAXSQ_VdqHdqWdqR, xmm_pmaxsq)
|
|
AVX_2OP(VPMAXUB_VdqHdqWdqR, xmm_pmaxub)
|
|
AVX_2OP(VPMAXUW_VdqHdqWdqR, xmm_pmaxuw)
|
|
AVX_2OP(VPMAXUD_VdqHdqWdqR, xmm_pmaxud)
|
|
AVX_2OP(VPMAXUQ_VdqHdqWdqR, xmm_pmaxuq)
|
|
|
|
AVX_2OP(VPSIGNB_VdqHdqWdqR, xmm_psignb)
|
|
AVX_2OP(VPSIGNW_VdqHdqWdqR, xmm_psignw)
|
|
AVX_2OP(VPSIGND_VdqHdqWdqR, xmm_psignd)
|
|
|
|
AVX_2OP(VPSUBSB_VdqHdqWdqR, xmm_psubsb)
|
|
AVX_2OP(VPSUBSW_VdqHdqWdqR, xmm_psubsw)
|
|
AVX_2OP(VPSUBUSB_VdqHdqWdqR, xmm_psubusb)
|
|
AVX_2OP(VPSUBUSW_VdqHdqWdqR, xmm_psubusw)
|
|
AVX_2OP(VPADDSB_VdqHdqWdqR, xmm_paddsb)
|
|
AVX_2OP(VPADDSW_VdqHdqWdqR, xmm_paddsw)
|
|
AVX_2OP(VPADDUSB_VdqHdqWdqR, xmm_paddusb)
|
|
AVX_2OP(VPADDUSW_VdqHdqWdqR, xmm_paddusw)
|
|
|
|
AVX_2OP(VPHADDW_VdqHdqWdqR, xmm_phaddw)
|
|
AVX_2OP(VPHADDD_VdqHdqWdqR, xmm_phaddd)
|
|
AVX_2OP(VPHADDSW_VdqHdqWdqR, xmm_phaddsw)
|
|
AVX_2OP(VPHSUBW_VdqHdqWdqR, xmm_phsubw)
|
|
AVX_2OP(VPHSUBD_VdqHdqWdqR, xmm_phsubd)
|
|
AVX_2OP(VPHSUBSW_VdqHdqWdqR, xmm_phsubsw)
|
|
|
|
AVX_2OP(VPAVGB_VdqHdqWdqR, xmm_pavgb)
|
|
AVX_2OP(VPAVGW_VdqHdqWdqR, xmm_pavgw)
|
|
|
|
AVX_2OP(VPACKUSWB_VdqHdqWdqR, xmm_packuswb)
|
|
AVX_2OP(VPACKSSWB_VdqHdqWdqR, xmm_packsswb)
|
|
AVX_2OP(VPACKUSDW_VdqHdqWdqR, xmm_packusdw)
|
|
AVX_2OP(VPACKSSDW_VdqHdqWdqR, xmm_packssdw)
|
|
|
|
AVX_2OP(VPUNPCKLBW_VdqHdqWdqR, xmm_punpcklbw)
|
|
AVX_2OP(VPUNPCKLWD_VdqHdqWdqR, xmm_punpcklwd)
|
|
AVX_2OP(VPUNPCKHBW_VdqHdqWdqR, xmm_punpckhbw)
|
|
AVX_2OP(VPUNPCKHWD_VdqHdqWdqR, xmm_punpckhwd)
|
|
|
|
AVX_2OP(VPMULLD_VdqHdqWdqR, xmm_pmulld)
|
|
AVX_2OP(VPMULLW_VdqHdqWdqR, xmm_pmullw)
|
|
AVX_2OP(VPMULHW_VdqHdqWdqR, xmm_pmulhw)
|
|
AVX_2OP(VPMULHUW_VdqHdqWdqR, xmm_pmulhuw)
|
|
AVX_2OP(VPMULDQ_VdqHdqWdqR, xmm_pmuldq)
|
|
AVX_2OP(VPMULUDQ_VdqHdqWdqR, xmm_pmuludq)
|
|
AVX_2OP(VPMULHRSW_VdqHdqWdqR, xmm_pmulhrsw)
|
|
|
|
AVX_2OP(VPMADDWD_VdqHdqWdqR, xmm_pmaddwd)
|
|
AVX_2OP(VPMADDUBSW_VdqHdqWdqR, xmm_pmaddubsw)
|
|
|
|
AVX_2OP(VPSADBW_VdqHdqWdqR, xmm_psadbw)
|
|
|
|
AVX_2OP(VPSRAVD_VdqHdqWdqR, xmm_psravd)
|
|
AVX_2OP(VPSLLVD_VdqHdqWdqR, xmm_psllvd)
|
|
AVX_2OP(VPSLLVQ_VdqHdqWdqR, xmm_psllvq)
|
|
AVX_2OP(VPSRLVD_VdqHdqWdqR, xmm_psrlvd)
|
|
AVX_2OP(VPSRLVQ_VdqHdqWdqR, xmm_psrlvq)
|
|
|
|
#define AVX_1OP(HANDLER, func) \
|
|
/* AVX instruction with single src operand */ \
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C :: HANDLER (bxInstruction_c *i) \
|
|
{ \
|
|
BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()); \
|
|
unsigned len = i->getVL(); \
|
|
\
|
|
for (unsigned n=0; n < len; n++) \
|
|
(func)(&op.vmm128(n)); \
|
|
\
|
|
BX_WRITE_AVX_REGZ(i->dst(), op, len); \
|
|
\
|
|
BX_NEXT_INSTR(i); \
|
|
}
|
|
|
|
AVX_1OP(VPABSB_VdqWdqR, xmm_pabsb)
|
|
AVX_1OP(VPABSW_VdqWdqR, xmm_pabsw)
|
|
AVX_1OP(VPABSD_VdqWdqR, xmm_pabsd)
|
|
|
|
#define AVX_PSHIFT(HANDLER, func) \
|
|
/* AVX packed shift instruction */ \
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C:: HANDLER (bxInstruction_c *i) \
|
|
{ \
|
|
BxPackedAvxRegister op = BX_READ_AVX_REG(i->src1()); \
|
|
unsigned len = i->getVL(); \
|
|
\
|
|
for (unsigned n=0; n < len; n++) \
|
|
(func)(&op.vmm128(n), BX_READ_XMM_REG_LO_QWORD(i->src2())); \
|
|
\
|
|
BX_WRITE_AVX_REGZ(i->dst(), op, len); \
|
|
\
|
|
BX_NEXT_INSTR(i); \
|
|
}
|
|
|
|
AVX_PSHIFT(VPSRLW_VdqHdqWdqR, xmm_psrlw);
|
|
AVX_PSHIFT(VPSRLD_VdqHdqWdqR, xmm_psrld);
|
|
AVX_PSHIFT(VPSRLQ_VdqHdqWdqR, xmm_psrlq);
|
|
AVX_PSHIFT(VPSRAW_VdqHdqWdqR, xmm_psraw);
|
|
AVX_PSHIFT(VPSRAD_VdqHdqWdqR, xmm_psrad);
|
|
AVX_PSHIFT(VPSLLW_VdqHdqWdqR, xmm_psllw);
|
|
AVX_PSHIFT(VPSLLD_VdqHdqWdqR, xmm_pslld);
|
|
AVX_PSHIFT(VPSLLQ_VdqHdqWdqR, xmm_psllq);
|
|
|
|
#define AVX_PSHIFT_IMM(HANDLER, func) \
|
|
/* AVX packed shift with imm8 instruction */ \
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C:: HANDLER (bxInstruction_c *i) \
|
|
{ \
|
|
BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()); \
|
|
unsigned len = i->getVL(); \
|
|
\
|
|
for (unsigned n=0; n < len; n++) \
|
|
(func)(&op.vmm128(n), i->Ib()); \
|
|
\
|
|
BX_WRITE_AVX_REGZ(i->dst(), op, len); \
|
|
\
|
|
BX_NEXT_INSTR(i); \
|
|
}
|
|
|
|
AVX_PSHIFT_IMM(VPSRLW_UdqIb, xmm_psrlw);
|
|
AVX_PSHIFT_IMM(VPSRLD_UdqIb, xmm_psrld);
|
|
AVX_PSHIFT_IMM(VPSRLQ_UdqIb, xmm_psrlq);
|
|
AVX_PSHIFT_IMM(VPSRAW_UdqIb, xmm_psraw);
|
|
AVX_PSHIFT_IMM(VPSRAD_UdqIb, xmm_psrad);
|
|
AVX_PSHIFT_IMM(VPSLLW_UdqIb, xmm_psllw);
|
|
AVX_PSHIFT_IMM(VPSLLD_UdqIb, xmm_pslld);
|
|
AVX_PSHIFT_IMM(VPSLLQ_UdqIb, xmm_psllq);
|
|
|
|
AVX_PSHIFT_IMM(VPSRLDQ_UdqIb, xmm_psrldq);
|
|
AVX_PSHIFT_IMM(VPSLLDQ_UdqIb, xmm_pslldq);
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPSHUFHW_VdqWdqIbR(bxInstruction_c *i)
|
|
{
|
|
BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()), result;
|
|
Bit8u order = i->Ib();
|
|
unsigned len = i->getVL();
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pshufhw(&result.vmm128(n), &op.vmm128(n), order);
|
|
|
|
BX_WRITE_AVX_REGZ(i->dst(), result, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPSHUFLW_VdqWdqIbR(bxInstruction_c *i)
|
|
{
|
|
BxPackedAvxRegister op = BX_READ_AVX_REG(i->src()), result;
|
|
Bit8u order = i->Ib();
|
|
unsigned len = i->getVL();
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pshuflw(&result.vmm128(n), &op.vmm128(n), order);
|
|
|
|
BX_WRITE_AVX_REGZ(i->dst(), result, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPSHUFB_VdqHdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedAvxRegister op1 = BX_READ_AVX_REG(i->src1());
|
|
BxPackedAvxRegister op2 = BX_READ_AVX_REG(i->src2()), result;
|
|
unsigned len = i->getVL();
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pshufb(&result.vmm128(n), &op1.vmm128(n), &op2.vmm128(n));
|
|
|
|
BX_WRITE_AVX_REGZ(i->dst(), result, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VMPSADBW_VdqHdqWdqIbR(bxInstruction_c *i)
|
|
{
|
|
BxPackedAvxRegister op1 = BX_READ_AVX_REG(i->src1());
|
|
BxPackedAvxRegister op2 = BX_READ_AVX_REG(i->src2()), result;
|
|
|
|
Bit8u control = i->Ib();
|
|
unsigned len = i->getVL();
|
|
|
|
for (unsigned n=0; n < len; n++) {
|
|
xmm_mpsadbw(&result.vmm128(n), &op1.vmm128(n), &op2.vmm128(n), control & 0x7);
|
|
control >>= 3;
|
|
}
|
|
|
|
BX_WRITE_AVX_REGZ(i->dst(), result, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPBLENDW_VdqHdqWdqIbR(bxInstruction_c *i)
|
|
{
|
|
BxPackedAvxRegister op1 = BX_READ_AVX_REG(i->src1()), op2 = BX_READ_AVX_REG(i->src2());
|
|
|
|
unsigned len = i->getVL();
|
|
Bit8u mask = i->Ib();
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pblendw(&op1.vmm128(n), &op2.vmm128(n), mask);
|
|
|
|
BX_WRITE_AVX_REGZ(i->dst(), op1, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPBROADCASTB_VdqWb(bxInstruction_c *i)
|
|
{
|
|
unsigned len = i->getVL();
|
|
BxPackedYmmRegister op;
|
|
|
|
Bit8u val_8 = BX_READ_XMM_REG_LO_BYTE(i->src());
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pbroadcastb(&op.ymm128(n), val_8);
|
|
|
|
BX_WRITE_YMM_REGZ_VLEN(i->dst(), op, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPBROADCASTW_VdqWw(bxInstruction_c *i)
|
|
{
|
|
unsigned len = i->getVL();
|
|
BxPackedYmmRegister op;
|
|
|
|
Bit16u val_16 = BX_READ_XMM_REG_LO_WORD(i->src());
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pbroadcastw(&op.ymm128(n), val_16);
|
|
|
|
BX_WRITE_YMM_REGZ_VLEN(i->dst(), op, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPBROADCASTD_VdqWd(bxInstruction_c *i)
|
|
{
|
|
unsigned len = i->getVL();
|
|
BxPackedYmmRegister op;
|
|
|
|
Bit32u val_32 = BX_READ_XMM_REG_LO_DWORD(i->src());
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pbroadcastd(&op.ymm128(n), val_32);
|
|
|
|
BX_WRITE_YMM_REGZ_VLEN(i->dst(), op, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPBROADCASTQ_VdqWq(bxInstruction_c *i)
|
|
{
|
|
unsigned len = i->getVL();
|
|
BxPackedYmmRegister op;
|
|
|
|
Bit64u val_64 = BX_READ_XMM_REG_LO_QWORD(i->src());
|
|
|
|
for (unsigned n=0; n < len; n++)
|
|
xmm_pbroadcastq(&op.ymm128(n), val_64);
|
|
|
|
BX_WRITE_YMM_REGZ_VLEN(i->dst(), op, len);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVSXBW256_VdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedXmmRegister op = BX_READ_XMM_REG(i->src());
|
|
BxPackedYmmRegister result;
|
|
|
|
for (int n=0; n<16; n++)
|
|
result.ymm16u(n) = (Bit8s) op.xmmsbyte(n);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVSXBD256_VdqWqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister result;
|
|
BxPackedMmxRegister op;
|
|
|
|
// use MMX register as 64-bit value with convinient accessors
|
|
MMXUQ(op) = BX_READ_XMM_REG_LO_QWORD(i->src());
|
|
|
|
result.ymm32u(0) = (Bit8s) MMXSB0(op);
|
|
result.ymm32u(1) = (Bit8s) MMXSB1(op);
|
|
result.ymm32u(2) = (Bit8s) MMXSB2(op);
|
|
result.ymm32u(3) = (Bit8s) MMXSB3(op);
|
|
result.ymm32u(4) = (Bit8s) MMXSB4(op);
|
|
result.ymm32u(5) = (Bit8s) MMXSB5(op);
|
|
result.ymm32u(6) = (Bit8s) MMXSB6(op);
|
|
result.ymm32u(7) = (Bit8s) MMXSB7(op);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVSXBQ256_VdqWdR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister result;
|
|
Bit32u val32 = BX_READ_XMM_REG_LO_DWORD(i->src());
|
|
|
|
result.ymm64u(0) = (Bit8s) (val32 & 0xFF);
|
|
result.ymm64u(1) = (Bit8s) ((val32 >> 8) & 0xFF);
|
|
result.ymm64u(2) = (Bit8s) ((val32 >> 16) & 0xFF);
|
|
result.ymm64u(3) = (Bit8s) (val32 >> 24);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVSXWD256_VdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedXmmRegister op = BX_READ_XMM_REG(i->src());
|
|
BxPackedYmmRegister result;
|
|
|
|
result.ymm32u(0) = op.xmm16s(0);
|
|
result.ymm32u(1) = op.xmm16s(1);
|
|
result.ymm32u(2) = op.xmm16s(2);
|
|
result.ymm32u(3) = op.xmm16s(3);
|
|
result.ymm32u(4) = op.xmm16s(4);
|
|
result.ymm32u(5) = op.xmm16s(5);
|
|
result.ymm32u(6) = op.xmm16s(6);
|
|
result.ymm32u(7) = op.xmm16s(7);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVSXWQ256_VdqWqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister result;
|
|
BxPackedMmxRegister op;
|
|
|
|
// use MMX register as 64-bit value with convinient accessors
|
|
MMXUQ(op) = BX_READ_XMM_REG_LO_QWORD(i->src());
|
|
|
|
result.ymm64u(0) = MMXSW0(op);
|
|
result.ymm64u(1) = MMXSW1(op);
|
|
result.ymm64u(2) = MMXSW2(op);
|
|
result.ymm64u(3) = MMXSW3(op);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVSXDQ256_VdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedXmmRegister op = BX_READ_XMM_REG(i->src());
|
|
BxPackedYmmRegister result;
|
|
|
|
result.ymm64u(0) = op.xmm32s(0);
|
|
result.ymm64u(1) = op.xmm32s(1);
|
|
result.ymm64u(2) = op.xmm32s(2);
|
|
result.ymm64u(3) = op.xmm32s(3);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVZXBW256_VdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedXmmRegister op = BX_READ_XMM_REG(i->src());
|
|
BxPackedYmmRegister result;
|
|
|
|
for (int n=0; n<16; n++)
|
|
result.ymm16u(n) = op.xmmubyte(n);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVZXBD256_VdqWqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister result;
|
|
BxPackedMmxRegister op;
|
|
|
|
// use MMX register as 64-bit value with convinient accessors
|
|
MMXUQ(op) = BX_READ_XMM_REG_LO_QWORD(i->src());
|
|
|
|
result.ymm32u(0) = MMXUB0(op);
|
|
result.ymm32u(1) = MMXUB1(op);
|
|
result.ymm32u(2) = MMXUB2(op);
|
|
result.ymm32u(3) = MMXUB3(op);
|
|
result.ymm32u(4) = MMXUB4(op);
|
|
result.ymm32u(5) = MMXUB5(op);
|
|
result.ymm32u(6) = MMXUB6(op);
|
|
result.ymm32u(7) = MMXUB7(op);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVZXBQ256_VdqWdR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister result;
|
|
Bit32u val32 = BX_READ_XMM_REG_LO_DWORD(i->src());
|
|
|
|
result.ymm64u(0) = (Bit8u) (val32 & 0xFF);
|
|
result.ymm64u(1) = (Bit8u) ((val32 >> 8) & 0xFF);
|
|
result.ymm64u(2) = (Bit8u) ((val32 >> 16) & 0xFF);
|
|
result.ymm64u(3) = (Bit8u) (val32 >> 24);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVZXWD256_VdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedXmmRegister op = BX_READ_XMM_REG(i->src());
|
|
BxPackedYmmRegister result;
|
|
|
|
result.ymm32u(0) = op.xmm16u(0);
|
|
result.ymm32u(1) = op.xmm16u(1);
|
|
result.ymm32u(2) = op.xmm16u(2);
|
|
result.ymm32u(3) = op.xmm16u(3);
|
|
result.ymm32u(4) = op.xmm16u(4);
|
|
result.ymm32u(5) = op.xmm16u(5);
|
|
result.ymm32u(6) = op.xmm16u(6);
|
|
result.ymm32u(7) = op.xmm16u(7);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVZXWQ256_VdqWqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister result;
|
|
BxPackedMmxRegister op;
|
|
|
|
// use MMX register as 64-bit value with convinient accessors
|
|
MMXUQ(op) = BX_READ_XMM_REG_LO_QWORD(i->src());
|
|
|
|
result.ymm64u(0) = MMXUW0(op);
|
|
result.ymm64u(1) = MMXUW1(op);
|
|
result.ymm64u(2) = MMXUW2(op);
|
|
result.ymm64u(3) = MMXUW3(op);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPMOVZXDQ256_VdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedXmmRegister op = BX_READ_XMM_REG(i->src());
|
|
BxPackedYmmRegister result;
|
|
|
|
result.ymm64u(0) = op.xmm32u(0);
|
|
result.ymm64u(1) = op.xmm32u(1);
|
|
result.ymm64u(2) = op.xmm32u(2);
|
|
result.ymm64u(3) = op.xmm32u(3);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPALIGNR_VdqHdqWdqIbR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister op1 = BX_READ_YMM_REG(i->src1()), op2 = BX_READ_YMM_REG(i->src2());
|
|
unsigned len = i->getVL();
|
|
|
|
for (unsigned n=0; n<len; n++)
|
|
xmm_palignr(&op2.ymm128(n), &op1.ymm128(n), i->Ib());
|
|
|
|
BX_WRITE_YMM_REGZ_VLEN(i->dst(), op2, i->getVL());
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPERMD_VdqHdqWdqR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister op1 = BX_READ_YMM_REG(i->src1());
|
|
BxPackedYmmRegister op2 = BX_READ_YMM_REG(i->src2()), result;
|
|
|
|
result.ymm32u(0) = op2.ymm32u(op1.ymm32u(0) & 0x7);
|
|
result.ymm32u(1) = op2.ymm32u(op1.ymm32u(1) & 0x7);
|
|
result.ymm32u(2) = op2.ymm32u(op1.ymm32u(2) & 0x7);
|
|
result.ymm32u(3) = op2.ymm32u(op1.ymm32u(3) & 0x7);
|
|
result.ymm32u(4) = op2.ymm32u(op1.ymm32u(4) & 0x7);
|
|
result.ymm32u(5) = op2.ymm32u(op1.ymm32u(5) & 0x7);
|
|
result.ymm32u(6) = op2.ymm32u(op1.ymm32u(6) & 0x7);
|
|
result.ymm32u(7) = op2.ymm32u(op1.ymm32u(7) & 0x7);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::VPERMQ_VdqWdqIbR(bxInstruction_c *i)
|
|
{
|
|
BxPackedYmmRegister op2 = BX_READ_YMM_REG(i->src()), result;
|
|
Bit8u control = i->Ib();
|
|
|
|
result.ymm64u(0) = op2.ymm64u((control) & 0x3);
|
|
result.ymm64u(1) = op2.ymm64u((control >> 2) & 0x3);
|
|
result.ymm64u(2) = op2.ymm64u((control >> 4) & 0x3);
|
|
result.ymm64u(3) = op2.ymm64u((control >> 6) & 0x3);
|
|
|
|
BX_WRITE_YMM_REGZ(i->dst(), result);
|
|
|
|
BX_NEXT_INSTR(i);
|
|
}
|
|
|
|
#endif
|