/*
 * SYNOPSYS CONFIDENTIAL - This is an unpublished, proprietary work of Synopsys,
 * Inc., and is fully protected under copyright and trade secret laws. You may
 * not view, use, disclose, copy, or distribute this file or any information
 * contained herein except pursuant to a valid written license from Synopsys.
 */

//
//	The purpose of this file is to define SSE2 data types to abstacr from the compiler 
//	specific constructs. Currently the target compilers are GCC and the MS VC 2005.
//

#ifndef _SSE2_CMPL_ABSTRACTION_MSC_PCKINT8_H_
#define _SSE2_CMPL_ABSTRACTION_MSC_PCKINT8_H_

//
//	Namespace sse2
//
namespace sse2
{

//
///	class epi64 (packed single precision)
//
class epi64
{
public:

//
///	The type.
//
typedef rxmm128l my_rxmm;

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//
///	Packed integer arithmetic
//

/*!
	r0 := a0 + b0 
	r1 := a1 + b1 
*/
static inline rxmm128d add( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_add_pd( a, b );
}

/*!
	r0 := a0 - b0 
	r1 := a1 - b1 
*/
static inline rxmm128d sub( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_sub_pd( a, b );
}

/*!
	r0 := a0 * b0 
	r1 := a1 * b1 
*/
static inline rxmm128d mul( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_mul_pd( a, b );
}

/*!
	r0 := a0 / b0 
	r1 := a1 / b1 
*/
static inline rxmm128d div( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_div_pd( a, b );
}

/*!
	r0 := max( a0, b0 )
	r1 := max( a1, b1 )
*/
static inline rxmm128d max( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_max_pd( a, b );
}

/*!
	r0 := min( a0, b0 )
	r1 := min( a1, b1 )
*/
static inline rxmm128d min( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_min_pd( a, b );
}

/*!
	r0 := sqrt( a0 )
	r1 := sqrt( a1 )
*/
static inline rxmm128d sqrt( rxmm128d a )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_sqrt_pd( a, b );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//
///	Packed integer logic
//

/*!
	r0 := (~a0) & b0 
	r1 := (~a1) & b1 
*/
static inline rxmm128d andnot( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_andnot_pd( a, b );
}

/*!
	r0 := a0 & b0 
	r1 := a1 & b1 
*/
static inline XMM_TYPE and( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_and_pd( a, b );
}

/*!
	r0 := a0 | b0 
	r1 := a1 | b1 
*/
static inline XMM_TYPE or( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_or_pd( a, b );
}

/*!
	r0 := a0 ^ b0 
	r1 := a1 ^ b1 
*/
static inline XMM_TYPE xor( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_xor_pd( a, b );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//
///	Packed integer comparision
//

/*!
	r0 := (a0 == b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 == b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_eq( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_cmpeq_pd( a, b );
}

/*!
	r0 := (a0 != b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 != b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_neq( rxmm128d a, rxmm128d b )
{
	return _mm_cmpneq_pd( a, b );
}

/*!
	r0 := (a0 < b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 < b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_lt( rxmm128d a, rxmm128d b )
{
	return _mm_cmplt_pd( a, b );
}

/*!
	r0 := (a0 <= b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 <= b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_le( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_cmple_pd( a, b );
}

/*!
	r0 := (a0 > b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 > b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_gt( rxmm128d a, rxmm128d b )
{
	BOOST_STATIC_ASSERT( false );
	return _mm_cmpgt_pd( a, b );
}

/*!
	r0 := (a0 >= b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 >= b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_ge( rxmm128d a, rxmm128d b )
{
	return _mm_cmpge_pd( a, b );
}

/*!
	r0 := (a0 ord b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 ord b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_ord( rxmm128d a, rxmm128d b )
{
	return _mm_cmpord_pd( a, b );
}

/*!
	r0 := (a0 unord b0) ? 0xffffffffffffffff : 0x0
	r1 := (a1 unord b1) ? 0xffffffffffffffff : 0x0
*/
static inline rxmm128d cmp_unord( rxmm128d a, rxmm128d b )
{
	return _mm_cmpunord_pd( a, b );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//
///	Packed integer load
//

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//	Misc

/*!
	r0 := a1
	r1 := b1
*/
static inline rxmm128d unpckh( rxmm128d a, rxmm128d b )
{
	return _mm_unpackhi_pd( a, b );
}

/*!
	r0 := a0
	r1 := b0
*/
static inline rxmm128d unpckl( rxmm128d a, rxmm128d b )
{
	return _mm_unpacklo_pd( a, b );
}

/*!
	r := sign(a1) << 1 | sign(a0)
*/
static inline int movmsk( rxmm128d a, rxmm128d b )
{
	return _mm_movemask_pd( a, b );
}

/*!
	r0 := (i0 == 1) ? b0 : a0 
	r1 := (i1 == 1) ? b1 : a1
	\sa movmsk
*/
static inline int shuffle( rxmm128d a, rxmm128d b, int i )
{
	return _mm_shuffle_pd( a, b, i );
}

/*!
	== shuffle( a, b, 1 )
	r0 := b0
	r1 := a1
*/
static inline rxmm128d move_sd( rxmm128d a, rxmm128d b )
{
	return _mm_move_sd( a0 );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//	Memory load

/*!
	The address \arg p must be 16-byte aligned.
	r0 := p[0]
	r1 := p[1]
*/
static inline rxmm128d load( double * p )
{
	return _mm_load_pd( p );
}

/*!
	The address \arg p must be 16-byte aligned.
	r0 := p[1]
	r1 := p[0]
*/
static inline rxmm128d load_reverse( double * p )
{
	return _mm_loadr_pd( p );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	r0 := p[0]
	r1 := p[1]
*/
static inline rxmm128d load_unaligned( double * p )
{
	return _mm_loadu_pd( p );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	r0 := a0
	r1 := *p
*/
static inline rxmm128d load_hi( rxmm128d a, double * p )
{
	return _mm_loadh_pd( a, p );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	r0 := *p
	r1 := a1
*/
static inline rxmm128d load_lo( rxmm128d a, double * p )
{
	return _mm_loadl_pd( a, p );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	r0 := *p
	r1 := *p
*/
static inline rxmm128d load_both( double * p )
{
	return _mm_load1_pd( p );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	r0 := *p
	r1 := 0.0
*/
static inline rxmm128d load_sd( double * p )
{
	return _mm_load_sd( p );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//	Memory store

/*!
	The address \arg p must be 16-byte aligned.
	p[0] := a0 
	p[1] := a1 
*/
static inline void store( double * p, rxmm128d a )
{
	_mm_load_pd( p, a );
}

/*!
	The address \arg p must be 16-byte aligned.
	p[0] := a1 
	p[1] := a0 
*/
static inline void store_reverse( double * p, rxmm128d a )
{
	_mm_storer_pd( p, a );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	p[0] := a0 
	p[1] := a1 
*/
static inline void store_unaligned(double * p, rxmm128d a )
{
	_mm_storeu_pd( p, a );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	*p := a1
*/
static inline void store_hi( double * p, rxmm128d a )
{
	_mm_storeh_pd( p, a );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	*p := a0
*/
static inline void store_lo( double * p, rxmm128d a )
{
	_mm_storel_pd( p, a );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	p[0] := a0
	p[1] := a0
*/
static inline void store_both( double * p, rxmm128d a )
{
	return _mm_store1_pd( p );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	*p := a0
*/
static inline void store_sd( double * p, rxmm128d a )
{
	return _mm_store_sd( p );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//	Memory set

/*!
	r0 := a0
	r1 := a1
*/
static inline rxmm128d set( double a1, double a0 )
{
	return _mm_set_pd( a1, a0 );
}

/*!
	r0 := 0.0
	r1 := 0.0
*/
static inline rxmm128d set_zero()
{
	return _mm_setzero_pd( a0 );
}

/*!
	r0 := a0
	r1 := a0
*/
static inline rxmm128d set_both( double a0 )
{
	return _mm_set1_pd( a0 );
}

/*!
	The address \arg p does not need to be 16-byte aligned.
	r0 := a0
	r1 := 0.0
*/
static inline rxmm128d set_sd( double a0 )
{
	return _mm_set_sd( a0 );
}

/*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*/
//
///	Packed integer convertion
//

/*!
	r0 := (float) a0
	r1 := (float) a1
	r2 := 0.0
	r3 := 0.0
*/
static inline rxmm128s cvtpd2ps( rxmm128d a )
{
	return _mm_cvtpd_ps( a );
}

/*!
	r0 := (double) a0
	r1 := (double) a1
*/
static inline rxmm128d cvtps2pd( rxmm128s a )
{
	return _mm_cvtps_pd( a );
}

/*!
	r0 := (int) a0
	r1 := (int) a1
	r2 := 0.0
	r3 := 0.0
*/
static inline rxmm128l cvtpd2dq( rxmm128d a )
{
	return _mm_cvtpd_epi32( a );
}

/*!
	r0 := (double) a0
	r1 := (double) a1
*/
static inline rxmm128d cvtdq2pd( rxmm128l a )
{
	return _mm_cvtepi32_pd( a );
}

/*!
	r := (int) a0
*/
static inline int cvtsd2si( rxmm128d a )
{
	return _mm_cvtsd_si32( a );
}

/*!
	r0 := (float) b0
	r1 := a1
	r2 := a2
	r3 := a3
*/
static inline rxmm128s cvtsd2ss( rxmm128l a, rxmm128d b )
{
	return _mm_cvtsd_ss( a, b );
}

/*!
	r0 := (double) b
	r1 := a1
*/
static inline rxmm128d cvtsi2sd( rxmm128d a, int b )
{
	return _mm_cvtsi32_sd( a, b );
}

/*!
	r0 := (double) b0
	r1 := a1
*/
static inline rxmm128d cvtss2sd( rxmm128d a, rxmm128s b )
{
	return _mm_cvtss_sd( a, b );
}

/*!
	using truncate
	r0 := (int) a0
	r1 := (int) a1
	r2 := 0x0
	r3 := 0x0
*/
static inline rxmm128l cvttpd2dq( rxmm128d a )
{
	return _mm_cvttpd_epi32( a );
}

/*!
	using truncate
	r := (int) a0
*/
static inline int cvttsd2si( rxmm128d a )
{
	return _mm_cvttsd_si32( a );
}

/*!
	r0 := (float) a0
	r1 := (float) a1
	r2 := (float) a2
	r3 := (float) a3
*/
static inline rxmm128s cvtdq2ps( rxmm128l a )
{
	return _mm_cvtepi32_ps( a );
}

/*!
	r0 := (int) a0
	r1 := (int) a1
	r2 := (int) a2
	r3 := (int) a3
*/
static inline rxmm128l cvtps2dq( rxmm128s a )
{
	return _mm_cvtps_epi32( a );
}

/*!
	uses trancate
	r0 := (int) a0
	r1 := (int) a1
	r2 := (int) a2
	r3 := (int) a3
*/
static inline rxmm128l cvttps2dq( rxmm128s a )
{
	return _mm_cvttps_epi32( a );
}

//
//	class epi64
//
};

//
//	Namespace sse2
//
}

#endif/*_SSE2_CMPL_ABSTRACTION_MSC_PCKINT8_H_*/