diff options
Diffstat (limited to 'sdk/rwsdk/include/d3d8/rtintel.h')
| -rw-r--r-- | sdk/rwsdk/include/d3d8/rtintel.h | 1206 |
1 files changed, 1206 insertions, 0 deletions
diff --git a/sdk/rwsdk/include/d3d8/rtintel.h b/sdk/rwsdk/include/d3d8/rtintel.h new file mode 100644 index 00000000..c11329df --- /dev/null +++ b/sdk/rwsdk/include/d3d8/rtintel.h @@ -0,0 +1,1206 @@ +/** + * Intel specific support toolkit + */ + +/********************************************************************** + * + * File : rtintel.h + * + * Abstract : Intel specific support/emulation + * + ********************************************************************** + * + * This file is a product of Criterion Software Ltd. + * + * This file is provided as is with no warranties of any kind and is + * provided without any obligation on Criterion Software Ltd. or + * Canon Inc. to assist in its use or modification. + * + * Criterion Software Ltd. will not, under any + * circumstances, be liable for any lost revenue or other damages arising + * from the use of this file. + * + * Copyright (c) 1998 Criterion Software Ltd. + * All Rights Reserved. + * + * RenderWare is a trademark of Canon Inc. + * + ************************************************************************/ + +#ifndef RTINTEL_H +#define RTINTEL_H + +/** + * \defgroup rtintel RtIntel + * \ingroup rttool + * + * Intel CPU Toolkit for RenderWare. + */ + +/**************************************************************************** + Include files + */ + +/* + * Pick up + * typedef struct _rwResEntryTag RwResEntry; + * from baresour.h (internal) / rwcore.h (external) + */ + +/**************************************************************************** + Global Types + */ + +#include "rtintel.rpe" /* automatically generated header file */ + +#if (!defined(RW_FIXED_64)) +typedef struct RwFixed64 RwFixed64; +struct RwFixed64 +{ + RwInt32 msb; + RwUInt32 lsb; +}; + +#define RW_FIXED_64 +#endif /* (!defined(RW_FIXED_64)) */ + +#define doubleFromRwFixed64(x) \ + ( ((double)((x).msb))*((double)(1<<16))*((double)(1<<16)) \ + + ((double)((x).lsb)) ) + +#if (!defined(RPINTELTIMEFUNCTION)) +typedef RwBool(*RtIntelTimeFunction) (void *data); + +#define RPINTELTIMEFUNCTION +#endif /* (!defined(RPINTELTIMEFUNCTION)) */ + +typedef struct RtIntelOverload RtIntelOverload; + +typedef void (*RwTransformFunction) (RwResEntry * repEntry); + +struct RtIntelOverload +{ + rwMatrixMultFn MatrixMultiplyFunction; + rwVectorMultFn VectorMultPointFunction; + rwVectorMultFn VectorMultVectorFunction; + RwTransformFunction TransformFunction; +}; + +#if (defined(__ICL)) +#define DEFINED__ICL " __ICL" +#define UNDEFINED__ICL "" +#else /* (defined(__ICL)) */ +#define DEFINED__ICL "" +#define UNDEFINED__ICL " __ICL" +#endif /* (defined(__ICL)) */ + +#if (defined(_MSC_VER)) + +# pragma comment ( user, "comment:" __FILE__ "(" RW_STRINGIFY_EXPANDED(__LINE__) ") : " "DEFINED " DEFINED__ICL " ; UNDEFINED " UNDEFINED__ICL ) +/* # pragma message (__FILE__ "(" RW_STRINGIFY_EXPANDED(__LINE__) ") : " "DEFINED " DEFINED__ICL " ; UNDEFINED " UNDEFINED__ICL ) */ +/* # pragma comment ( user, "comment:" __FILE__ "(" RW_STRINGIFY_EXPANDED(__LINE__) ") : " "DEFINED " DEFINED__ICL " ; UNDEFINED " UNDEFINED__ICL ) */ +#if (! (defined(__ICL) || defined(XBOX_DRVMODEL_H)) ) +#pragma message (__DATE__ " " __TIME__ " " __FILE__ "(" RW_STRINGIFY_EXPANDED(__LINE__) ") : No MMX intrinsics - defaulting to software emulation") +#pragma message (__DATE__ " " __TIME__ " " __FILE__ "(" RW_STRINGIFY_EXPANDED(__LINE__) ") : No SSE intrinsics - defaulting to software emulation") +#pragma comment ( user, "comment:" __FILE__ "(" RW_STRINGIFY_EXPANDED(__LINE__) ") : No MMX intrinsics - defaulting to software emulation") +#endif /* (! (defined(__ICL) || defined(XBOX_DRVMODEL_H)) ) */ +#endif /* (defined(_MSC_VER)) */ + +/* + * MMX + */ + +#if (defined(__ICL)) + +#if (!defined(MMINTRIN_H)) +#include "mmintrin.h" +#define MMINTRIN_H +#endif /* (!defined(MMINTRIN_H)) */ + +#else /* (defined(__ICL)) */ + +#if (defined(_MSC_VER)) + +#if (!defined(__M64)) +typedef __int64 Rt_m64; + +#define __M64 +#endif /* (!defined(__M64)) */ + +#else /* (defined(_MSC_VER)) -- e.g. __GNUC__ */ + +#if (!defined(__M64)) + +typedef RwInt64 Rt_m64; + +#define __M64 + +#endif /* (!defined(__M64)) */ + +#endif /* (defined(_MSC_VER)) */ + +#endif /* (defined(__ICL)) */ + +/* + * SSE + */ + +/* + * From + * ccomp.pdf + * 12 Intel C/C++ Compiler User's Guide + * for Win32 Systems With Katmai New Instruction Support + * -------------------------------------------------------- + * Functionality Intrinsics Usage + * You need only define one preprocessor symbol and include the header file + * xmmintrin.h in your application to use the following functionality + * intrinsics: + * #define _MM_FUNCTIONALITY + * #include "xmmintrin.h" + * To encourage the compiler to inline the functionality intrinsic functions for + * better performance, consider using the -Qip and -Qipo compiler switches. + */ + +#if (defined(__ICL)) + +/* #define _MM2_FUNCTIONALITY */ + +/* #define _MM_FUNCTIONALITY */ + +/* #define _MM_NO_ABORT */ + +/* #define _MM_NO_ACCURACY */ + +/* #define _MM_NO_ALIGN_CHECK */ + +/* #define _MM_NO_INLINE */ + +/* + * Undefine "or", since this is valid assembler; e.g. in + * SDK10/include/xmm_func.h + * _asm { + * push eax + * fld f + * fstcw saved_cw + * mov eax, saved_cw + * or eax, 3072 + * mov new_cw, eax + * fldcw new_cw + * fistp ret + * fldcw saved_cw + * pop eax + * } + */ + +#if (!defined(XMMINTRIN_H)) +#include "xmmintrin.h" +#define XMMINTRIN_H +#endif /* (!defined(XMMINTRIN_H)) */ + +typedef __m64 Rt_m64; +typedef __m128 Rt_m128; + +#if (450 <= __ICL) +#if (!defined(EMMINTRIN_H)) +#include "emmintrin.h" +#define EMMINTRIN_H +typedef __m128d Rt_m128d; +typedef __m128i Rt_m128i; +#endif /* (!defined(EMMINTRIN_H)) */ +#else /* (450 <= __ICL) */ +typedef __m128 Rt_m128d; +typedef __m128 Rt_m128i; +#endif /* (450 <= __ICL) */ + +/* + * Report SSE options as compiler messages and object file comments + */ + +#ifdef _MM2_FUNCTIONALITY +#define DEFINED__MM2_FUNCTIONALITY " _MM2_FUNCTIONALITY" +#define UNDEFINED__MM2_FUNCTIONALITY "" +#else /* _MM2_FUNCTIONALITY */ +#define DEFINED__MM2_FUNCTIONALITY "" +#define UNDEFINED__MM2_FUNCTIONALITY " _MM2_FUNCTIONALITY" +#endif /* _MM2_FUNCTIONALITY */ + +#ifdef _MM_FUNCTIONALITY +#define DEFINED__MM_FUNCTIONALITY DEFINED__MM2_FUNCTIONALITY ## " _MM_FUNCTIONALITY" +#define UNDEFINED__MM_FUNCTIONALITY UNDEFINED__MM2_FUNCTIONALITY +#else /* _MM_FUNCTIONALITY */ +#define DEFINED__MM_FUNCTIONALITY DEFINED__MM2_FUNCTIONALITY +#define UNDEFINED__MM_FUNCTIONALITY UNDEFINED__MM2_FUNCTIONALITY ## " _MM_FUNCTIONALITY" +#endif /* _MM_FUNCTIONALITY */ + +#ifdef _MM_NO_ABORT +#define DEFINED__MM_NO_ABORT DEFINED__MM_FUNCTIONALITY ## " _MM_NO_ABORT" +#define UNDEFINED__MM_NO_ABORT UNDEFINED__MM_FUNCTIONALITY +#else /* _MM_NO_ABORT */ +#define DEFINED__MM_NO_ABORT DEFINED__MM_FUNCTIONALITY +#define UNDEFINED__MM_NO_ABORT UNDEFINED__MM_FUNCTIONALITY ## " _MM_NO_ABORT" +#endif /* _MM_NO_ABORT */ + +#ifdef _MM_NO_ACCURACY +#define DEFINED__MM_NO_ACCURACY DEFINED__MM_NO_ABORT ## " _MM_NO_ACCURACY" +#define UNDEFINED__MM_NO_ACCURACY UNDEFINED__MM_NO_ABORT +#else /* _MM_NO_ACCURACY */ +#define DEFINED__MM_NO_ACCURACY DEFINED__MM_NO_ABORT +#define UNDEFINED__MM_NO_ACCURACY UNDEFINED__MM_NO_ABORT ## " _MM_NO_ACCURACY" +#endif /* _MM_NO_ACCURACY */ + +#ifdef _MM_NO_ALIGN_CHECK +#define DEFINED__MM_NO_ALIGN_CHECK DEFINED__MM_NO_ACCURACY ## " _MM_NO_ALIGN_CHECK" +#define UNDEFINED__MM_NO_ALIGN_CHECK UNDEFINED__MM_NO_ACCURACY +#else /* _MM_NO_ALIGN_CHECK */ +#define DEFINED__MM_NO_ALIGN_CHECK DEFINED__MM_NO_ACCURACY +#define UNDEFINED__MM_NO_ALIGN_CHECK UNDEFINED__MM_NO_ACCURACY ## " _MM_NO_ALIGN_CHECK" +#endif /* _MM_NO_ALIGN_CHECK */ + +#ifdef _MM_NO_INLINE +#define DEFINED__MM_NO_INLINE DEFINED__MM_NO_ALIGN_CHECK ## " _MM_NO_INLINE" +#define UNDEFINED__MM_NO_INLINE UNDEFINED__MM_NO_ALIGN_CHECK +#else /* _MM_NO_INLINE */ +#define DEFINED__MM_NO_INLINE DEFINED__MM_NO_ALIGN_CHECK +#define UNDEFINED__MM_NO_INLINE UNDEFINED__MM_NO_ALIGN_CHECK ## " _MM_NO_INLINE" +#endif /* _MM_NO_INLINE */ + +#pragma comment ( user, "comment:" __DATE__" " __TIME__ " - " __FILE__ ":" RW_STRINGIFY_EXPANDED(__LINE__) ) +#pragma comment ( user, "comment:" "DEFINED :" DEFINED__MM_NO_INLINE ) +#pragma comment ( user, "comment:" "UNDEFINED:" UNDEFINED__MM_NO_INLINE ) + +#pragma message (__DATE__" " __TIME__ " - " __FILE__ ":" RW_STRINGIFY_EXPANDED(__LINE__) ) +#pragma message ("DEFINED :" DEFINED__MM_NO_INLINE ) +#pragma message ("UNDEFINED:" UNDEFINED__MM_NO_INLINE ) + +#else /* (defined(__ICL)) */ + +#define _MM_HINT_T0 1 +#define _MM_HINT_T1 2 +#define _MM_HINT_T2 3 +#define _MM_HINT_NTA 0 + +#if (defined(__R5900__)) +typedef RwInt128 Rt_m128; +#else /* (defined(__R5900__)) */ +#if (!defined(_PAIR__M64)) +struct Rt_m128 +{ + Rt_m64 lo; + Rt_m64 hi; +}; +#define _PAIR__M64 +typedef struct Rt_m128 Rt_m128; +#endif /* (!defined(_PAIR__M64)) */ +#endif /* (defined(__R5900__)) */ + +typedef Rt_m128 Rt_m128d; +typedef Rt_m128 Rt_m128i; +#endif /* (defined(__ICL)) */ + +typedef struct RtIntelV4d RtIntelV4d; +struct RtIntelV4d +{ + RwReal w; + RwV3d v3d; +}; + +typedef union _RpSSEOverlayM128 RpSSEOverlayM128; + +union _RpSSEOverlayM128 +{ + float _f[4]; + RwInt32 _d[4]; + RwUInt32 ud[4]; + RwInt16 _w[8]; + RwUInt16 uw[8]; + RwInt8 _b[16]; + RwUInt8 ub[16]; + Rt_m64 m64[2]; + Rt_m128 m128; + RtIntelV4d v4d; + RwSplitBits bits[4]; +}; + +typedef RpSSEOverlayM128 RpWNIOverlayM128; + +typedef union _RpWNIOverlayM128d RpWNIOverlayM128d; + +union _RpWNIOverlayM128d +{ + double df[2]; + float _f[4]; + RwInt32 _d[4]; + RwUInt32 ud[4]; + RwInt16 _w[8]; + RwUInt16 uw[8]; + RwInt8 _b[16]; + RwUInt8 ub[16]; + Rt_m64 m64[2]; + Rt_m128d m128d; + RtIntelV4d v4d; + RwSplitBits bits[4]; +}; + +typedef union _RpWNIOverlayM128i RpWNIOverlayM128i; + +union _RpWNIOverlayM128i +{ + double df[2]; + float _f[4]; + RwInt32 _d[4]; + RwUInt32 ud[4]; + RwInt16 _w[8]; + RwUInt16 uw[8]; + RwInt8 _b[16]; + RwUInt8 ub[16]; + Rt_m64 m64[2]; + Rt_m128i m128i; + RtIntelV4d v4d; + RwSplitBits bits[4]; +}; + +#define RWUNALIGNED16BYTE(ptr) (0x0000000FUL & ((RwUInt32)(ptr))) + +/*--- Plugin API Functions ---*/ + +#ifdef __cplusplus +extern "C" +{ +#endif /* __cplusplus */ + +/* + * MMX + */ + +/* General support intrinsics */ +extern void Rt_m_empty(void); +extern Rt_m64 Rt_m_from_int(int i); +extern int Rt_m_to_int(Rt_m64 m); +extern Rt_m64 Rt_m_packsswb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_packssdw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_packuswb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_punpckhbw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_punpckhwd(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_punpckhdq(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_punpcklbw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_punpcklwd(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_punpckldq(Rt_m64 m1, Rt_m64 m2); + +/* Packed arithmetic intrinsics */ +extern Rt_m64 Rt_m_paddb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_paddw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_paddd(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_paddsb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_paddsw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_paddusb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_paddusw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubd(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubsb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubsw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubusb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_psubusw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pmaddwd(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pmulhw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pmullw(Rt_m64 m1, Rt_m64 m2); + +/* Shift intrinsics */ +extern Rt_m64 Rt_m_psllw(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psllwi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_pslld(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_pslldi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_psllq(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psllqi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_psraw(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psrawi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_psrad(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psradi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_psrlw(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psrlwi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_psrld(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psrldi(Rt_m64 m, int count); +extern Rt_m64 Rt_m_psrlq(Rt_m64 m, Rt_m64 count); +extern Rt_m64 Rt_m_psrlqi(Rt_m64 m, int count); + +/* Logical intrinsics */ +extern Rt_m64 Rt_m_pand(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pandn(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_por(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pxor(Rt_m64 m1, Rt_m64 m2); + +/* Comparision intrinsics */ +extern Rt_m64 Rt_m_pcmpeqb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pcmpeqw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pcmpeqd(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pcmpgtb(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pcmpgtw(Rt_m64 m1, Rt_m64 m2); +extern Rt_m64 Rt_m_pcmpgtd(Rt_m64 m1, Rt_m64 m2); + +/* + * SSE + */ + +/* + * Arithmetic Operations + */ + +extern Rt_m128 Rt_mm_add_ss(Rt_m128 a, Rt_m128 b) /* ADDSS */ ; +extern Rt_m128 Rt_mm_add_ps(Rt_m128 a, Rt_m128 b) /* ADDPS */ ; +extern Rt_m128 Rt_mm_sub_ss(Rt_m128 a, Rt_m128 b) /* SUBSS */ ; +extern Rt_m128 Rt_mm_sub_ps(Rt_m128 a, Rt_m128 b) /* SUBPS */ ; +extern Rt_m128 Rt_mm_mul_ss(Rt_m128 a, Rt_m128 b) /* MULSS */ ; +extern Rt_m128 Rt_mm_mul_ps(Rt_m128 a, Rt_m128 b) /* MULPS */ ; +extern Rt_m128 Rt_mm_div_ss(Rt_m128 a, Rt_m128 b) /* DIVSS */ ; +extern Rt_m128 Rt_mm_div_ps(Rt_m128 a, Rt_m128 b) /* DIVPS */ ; +extern Rt_m128 Rt_mm_sqrt_ss(Rt_m128 a) /* SQRTSS */ ; +extern Rt_m128 Rt_mm_sqrt_ps(Rt_m128 a) /* SQRTPS */ ; +extern Rt_m128 Rt_mm_rcp_ss(Rt_m128 a) /* RCPSS */ ; +extern Rt_m128 Rt_mm_rcp_ps(Rt_m128 a) /* RCPPS */ ; +extern Rt_m128 Rt_mm_rsqrt_ss(Rt_m128 a) /* RSQRTSS */ ; +extern Rt_m128 Rt_mm_rsqrt_ps(Rt_m128 a) /* RSQRTPS */ ; +extern Rt_m128 Rt_mm_min_ss(Rt_m128 a, Rt_m128 b) /* MINSS */ ; +extern Rt_m128 Rt_mm_min_ps(Rt_m128 a, Rt_m128 b) /* MINPS */ ; +extern Rt_m128 Rt_mm_max_ss(Rt_m128 a, Rt_m128 b) /* MAXSS */ ; +extern Rt_m128 Rt_mm_max_ps(Rt_m128 a, Rt_m128 b) /* MAXPS */ ; + +/* + * Logical Operations + */ + +extern Rt_m128 Rt_mm_and_ps(Rt_m128 a, Rt_m128 b) /* ANDPS */ ; +extern Rt_m128 Rt_mm_andnot_ps(Rt_m128 a, + Rt_m128 b) /* ANDNPS */ ; +extern Rt_m128 Rt_mm_or_ps(Rt_m128 a, Rt_m128 b) /* ORPS */ ; +extern Rt_m128 Rt_mm_xor_ps(Rt_m128 a, Rt_m128 b) /* XORPS */ ; + +/* + * Comparisons + */ + +extern Rt_m128 Rt_mm_cmpeq_ss(Rt_m128 a, + Rt_m128 b) /* CMPEQSS */ ; +extern Rt_m128 Rt_mm_cmpeq_ps(Rt_m128 a, + Rt_m128 b) /* CMPEQPS */ ; +extern Rt_m128 Rt_mm_cmplt_ss(Rt_m128 a, + Rt_m128 b) /* CMPLTSS */ ; +extern Rt_m128 Rt_mm_cmplt_ps(Rt_m128 a, + Rt_m128 b) /* CMPLTPS */ ; +extern Rt_m128 Rt_mm_cmple_ss(Rt_m128 a, + Rt_m128 b) /* CMPLESS */ ; +extern Rt_m128 Rt_mm_cmple_ps(Rt_m128 a, + Rt_m128 b) /* CMPLEPS */ ; +extern Rt_m128 Rt_mm_cmpgt_ss(Rt_m128 a, Rt_m128 b) /* CMPLTSS r */ + ; +extern Rt_m128 Rt_mm_cmpgt_ps(Rt_m128 a, Rt_m128 b) /* CMPLTPS r */ + ; +extern Rt_m128 Rt_mm_cmpge_ss(Rt_m128 a, Rt_m128 b) /* CMPLESS r */ + ; +extern Rt_m128 Rt_mm_cmpge_ps(Rt_m128 a, Rt_m128 b) /* CMPLEPS r */ + ; +extern Rt_m128 Rt_mm_cmpneq_ss(Rt_m128 a, + Rt_m128 b) /* CMPNEQSS */ ; +extern Rt_m128 Rt_mm_cmpneq_ps(Rt_m128 a, + Rt_m128 b) /* CMPNEQPS */ ; +extern Rt_m128 Rt_mm_cmpnlt_ss(Rt_m128 a, + Rt_m128 b) /* CMPNLTSS */ ; +extern Rt_m128 Rt_mm_cmpnlt_ps(Rt_m128 a, + Rt_m128 b) /* CMPNLTPS */ ; +extern Rt_m128 Rt_mm_cmpnle_ss(Rt_m128 a, + Rt_m128 b) /* CMPNLESS */ ; +extern Rt_m128 Rt_mm_cmpnle_ps(Rt_m128 a, + Rt_m128 b) /* CMPNLEPS */ ; +extern Rt_m128 Rt_mm_cmpngt_ss(Rt_m128 a, Rt_m128 b) + /* CMPNLTSS r */ ; +extern Rt_m128 Rt_mm_cmpngt_ps(Rt_m128 a, Rt_m128 b) + /* CMPNLTPS r */ ; +extern Rt_m128 Rt_mm_cmpnge_ss(Rt_m128 a, Rt_m128 b) + /* CMPNLESS r */ ; +extern Rt_m128 Rt_mm_cmpnge_ps(Rt_m128 a, Rt_m128 b) + /* CMPNLEPS r */ ; +extern Rt_m128 Rt_mm_cmpord_ss(Rt_m128 a, + Rt_m128 b) /* CMPORDSS */ ; +extern Rt_m128 Rt_mm_cmpord_ps(Rt_m128 a, + Rt_m128 b) /* CMPORDPS */ ; +extern Rt_m128 Rt_mm_cmpunord_ss(Rt_m128 a, + Rt_m128 b) /* CMPUNORDSS */ ; +extern Rt_m128 Rt_mm_cmpunord_ps(Rt_m128 a, + Rt_m128 b) /* CMPUNORDPS */ ; +extern int Rt_mm_comieq_ss(Rt_m128 a, + Rt_m128 b) /* COMISS */ ; +extern int Rt_mm_comilt_ss(Rt_m128 a, + Rt_m128 b) /* COMISS */ ; +extern int Rt_mm_comile_ss(Rt_m128 a, + Rt_m128 b) /* COMISS */ ; +extern int Rt_mm_comigt_ss(Rt_m128 a, + Rt_m128 b) /* COMISS */ ; +extern int Rt_mm_comige_ss(Rt_m128 a, + Rt_m128 b) /* COMISS */ ; +extern int Rt_mm_comineq_ss(Rt_m128 a, + Rt_m128 b) /* COMISS */ ; +extern int Rt_mm_ucomieq_ss(Rt_m128 a, + Rt_m128 b) /* UCOMISS */ ; +extern int Rt_mm_ucomilt_ss(Rt_m128 a, + Rt_m128 b) /* UCOMISS */ ; +extern int Rt_mm_ucomile_ss(Rt_m128 a, + Rt_m128 b) /* UCOMISS */ ; +extern int Rt_mm_ucomigt_ss(Rt_m128 a, + Rt_m128 b) /* UCOMISS */ ; +extern int Rt_mm_ucomige_ss(Rt_m128 a, + Rt_m128 b) /* UCOMISS */ ; +extern int Rt_mm_ucomineq_ss(Rt_m128 a, + Rt_m128 b) /* UCOMISS */ ; + +/* + * Conversion Operations + */ + +extern int Rt_mm_cvt_ss2si(Rt_m128 a) /* CVTSS2SI */ ; +extern Rt_m64 Rt_mm_cvt_ps2pi(Rt_m128 a) /* CVTPS2PI */ ; +extern int Rt_mm_cvtt_ss2si(Rt_m128 a) /* CVTTSS2SI */ ; +extern Rt_m64 Rt_mm_cvtt_ps2pi(Rt_m128 a) /* CVTTPS2PI */ ; +extern Rt_m128 Rt_mm_cvt_si2ss(Rt_m128 a, + int b) /* CVTSI2SS */ ; +extern Rt_m128 Rt_mm_cvt_pi2ps(Rt_m128 a, + Rt_m64 b) /* CVTPI2PS */ ; + +/* + * Miscellaneous + */ + +extern Rt_m128 Rt_mm_shuffle_ps(Rt_m128 a, Rt_m128 b, + int i) /* SHUFPS */ ; +extern Rt_m128 Rt_mm_unpackhi_ps(Rt_m128 a, + Rt_m128 b) /* UNPCKHPS */ ; +extern Rt_m128 Rt_mm_unpacklo_ps(Rt_m128 a, + Rt_m128 b) /* UNPCKLPS */ ; +extern Rt_m128 Rt_mm_loadh_pi(Rt_m128 a, Rt_m64 * p) /* MOVHPS reg, mem */ + ; +extern void Rt_mm_storeh_pi(Rt_m64 * p, Rt_m128 a) /* MOVHPS mem, reg */ + ; +extern Rt_m128 Rt_mm_movehl_ps(Rt_m128 a, + Rt_m128 b) /* MOVHLPS */ ; +extern Rt_m128 Rt_mm_movelh_ps(Rt_m128 a, + Rt_m128 b) /* MOVLHPS */ ; +extern Rt_m128 Rt_mm_loadl_pi(Rt_m128 a, Rt_m64 * p) + /* MOVLPS reg, mem */ ; +extern void Rt_mm_storel_pi(Rt_m64 * p, Rt_m128 a) /* MOVLPS mem, reg */ + ; +extern int Rt_mm_movemask_ps(Rt_m128 a) /* MOVMSKPS */ ; +extern unsigned int Rt_mm_getcsr(void) /* STMXCSR */ ; +extern void Rt_mm_setcsr(unsigned int i) /* LDMXCSR */ ; + +/* + * Load Operations + */ + +extern Rt_m128 Rt_mm_load_ss(float *p) /* MOVSS */ ; +extern Rt_m128 Rt_mm_load_ps1(float *p) /* MOVSS + shuffling */ + ; +extern Rt_m128 Rt_mm_load_ps(float *p) /* MOVAPS */ ; +extern Rt_m128 Rt_mm_loadu_ps(float *p) /* MOVUPS */ ; +extern Rt_m128 Rt_mm_loadr_ps(float *p) + /* MOVAPS + shuffling */ ; + +/* + * Set Operations + */ + +extern Rt_m128 Rt_mm_set_ss(float w) /* (composite) */ ; +extern Rt_m128 Rt_mm_set_ps1(float w) /* (composite) */ ; +extern Rt_m128 Rt_mm_set_ps(float z, float y, float x, + float w) /* (composite) */ ; +extern Rt_m128 Rt_mm_setr_ps(float z, float y, float x, + float w) /* (composite) */ ; +extern Rt_m128 Rt_mm_setzero_ps(void) /* (composite) */ ; + +/* + * Store Operations + */ + +extern void Rt_mm_store_ss(float *p, + Rt_m128 a) /* MOVSS */ ; +extern void Rt_mm_store_ps1(float *p, Rt_m128 a) + /* MOVSS + shuffling */ ; +extern void Rt_mm_store_ps(float *p, + Rt_m128 a) /* MOVAPS */ ; +extern void Rt_mm_storeu_ps(float *p, + Rt_m128 a) /* MOVUPS */ ; +extern void Rt_mm_storer_ps(float *p, Rt_m128 a) + /* MOVAPS + shuffling */ ; +extern Rt_m128 Rt_mm_move_ss(Rt_m128 a, + Rt_m128 b) /* MOVSS */ ; + +/* + * Integer Intrinsics + */ + +extern int Rt_m_pextrw(Rt_m64 a, int n) /* PEXTRW */ ; +extern Rt_m64 Rt_m_pinsrw(Rt_m64 a, int d, + int n) /* PINSRW */ ; +extern Rt_m64 Rt_m_pmaxsw(Rt_m64 a, Rt_m64 b) /* PMAXSW */ ; +extern Rt_m64 Rt_m_pmaxub(Rt_m64 a, Rt_m64 b) /* PMAXUB */ ; +extern Rt_m64 Rt_m_pminsw(Rt_m64 a, Rt_m64 b) /* PMINSW */ ; +extern Rt_m64 Rt_m_pminub(Rt_m64 a, Rt_m64 b) /* PMINUB */ ; +extern int Rt_m_pmovmskb(Rt_m64 a) /* PMOVMSKB */ ; +extern Rt_m64 Rt_m_pmulhuw(Rt_m64 a, Rt_m64 b) /* PMULHUW */ ; +extern Rt_m64 Rt_m_pshufw(Rt_m64 a, int n) /* PSHUFW */ ; +extern void Rt_m_lwmaskmovq(Rt_m64 d, Rt_m64 n, + char *p) /* MASKMOVQ */ ; + +/* + * Cacheability Support + */ + +extern void Rt_mm_prefetch(char *p, int i) /* PREFETCH */ ; +extern void Rt_mm_stream_pi(Rt_m64 * p, + Rt_m64 a) /* MOVNTQ */ ; +extern void Rt_mm_stream_ps(float *p, + Rt_m128 a) /* MOVNTPS */ ; +extern void Rt_mm_sfence(void) /* SFENCE */ ; + +/* + * WNI + */ + +/* Arithmetic Operations */ + +extern Rt_m128d Rt_mm_add_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_add_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_div_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_div_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_max_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_max_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_min_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_min_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_mul_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_mul_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_sqrt_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_sqrt_pd(Rt_m128d a); +extern Rt_m128d Rt_mm_sub_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_sub_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_andnot_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_and_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_or_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_xor_pd(Rt_m128d a, Rt_m128d b); + +/* Comparisons */ + +extern Rt_m128d Rt_mm_cmpeq_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmplt_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmple_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpgt_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpge_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpord_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpunord_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpneq_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpnlt_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpnle_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpngt_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpnge_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpeq_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmplt_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmple_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpgt_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpge_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpord_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpunord_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpneq_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpnlt_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpnle_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpngt_sd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_cmpnge_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_comieq_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_comilt_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_comile_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_comigt_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_comige_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_comineq_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_ucomieq_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_ucomilt_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_ucomile_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_ucomigt_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_ucomige_sd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_ucomineq_sd(Rt_m128d a, Rt_m128d b); + +/* Conversion Operations */ + +extern Rt_m128 Rt_mm_cvtpd_ps(Rt_m128d a); +extern Rt_m128d Rt_mm_cvtps_pd(Rt_m128 a); +extern Rt_m128d Rt_mm_cvtepi32_pd(Rt_m128i a); +extern Rt_m128i Rt_mm_cvtpd_epi32(Rt_m128d a); +extern int Rt_mm_cvtsd_si32(Rt_m128d a); +extern Rt_m128 Rt_mm_cvtsd_ss(Rt_m128 a, Rt_m128d b); +extern Rt_m128d Rt_mm_cvtsi32_sd(Rt_m128d a, int b); +extern Rt_m128d Rt_mm_cvtss_sd(Rt_m128d a, Rt_m128 b); +extern Rt_m128i Rt_mm_cvttpd_epi32(Rt_m128d a); +extern int Rt_mm_cvttsd_si32(Rt_m128d a); +extern Rt_m128 Rt_mm_cvtepi32_ps(Rt_m128i a); +extern Rt_m128i Rt_mm_cvtps_epi32(Rt_m128 a); +extern Rt_m128i Rt_mm_cvttps_epi32(Rt_m128 a); +extern Rt_m64 Rt_mm_cvtpd_pi32(Rt_m128d a); +extern Rt_m64 Rt_mm_cvttpd_pi32(Rt_m128d a); +extern Rt_m128d Rt_mm_cvtpi32_pd(Rt_m64 a); + +/* Miscellaneous Operations */ + +extern Rt_m128d Rt_mm_unpackhi_pd(Rt_m128d a, Rt_m128d b); +extern Rt_m128d Rt_mm_unpacklo_pd(Rt_m128d a, Rt_m128d b); +extern int Rt_mm_movemask_pd(Rt_m128d a); +extern Rt_m128d Rt_mm_shuffle_pd(Rt_m128d a, Rt_m128d b, int i); +extern Rt_m128d Rt_mm_load_pd(const double *p); +extern Rt_m128d Rt_mm_load1_pd(const double *p); +extern Rt_m128d Rt_mm_loadr_pd(const double *p); +extern Rt_m128d Rt_mm_loadu_pd(const double *p); +extern Rt_m128d Rt_mm_load_sd(const double *p); +extern Rt_m128d Rt_mm_loadh_pd(Rt_m128d a, const double *p); +extern Rt_m128d Rt_mm_loadl_pd(Rt_m128d a, const double *p); +extern Rt_m128d Rt_mm_set_sd(double w); +extern Rt_m128d Rt_mm_set1_pd(double w); +extern Rt_m128d Rt_mm_set_pd(double w, double x); +extern Rt_m128d Rt_mm_setr_pd(double w, double x); +extern Rt_m128d Rt_mm_setzero_pd(void); +extern Rt_m128d Rt_mm_move_sd(Rt_m128d a, Rt_m128d b); +extern void Rt_mm_stream_pd(double *p, Rt_m128d a); +extern void Rt_mm_store_sd(double *p, Rt_m128d a); +extern void Rt_mm_store1_pd(double *p, Rt_m128d a); +extern void Rt_mm_store_pd(double *p, Rt_m128d a); +extern void Rt_mm_storeu_pd(double *p, Rt_m128d a); +extern void Rt_mm_storer_pd(double *p, Rt_m128d a); +extern void Rt_mm_storeh_pd(double *p, Rt_m128d a); +extern void Rt_mm_storel_pd(double *p, Rt_m128d a); +extern Rt_m128i Rt_mm_add_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_add_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_add_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m64 Rt_mm_add_si64(Rt_m64 a, Rt_m64 b); +extern Rt_m128i Rt_mm_add_epi64(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_adds_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_adds_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_adds_epu8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_adds_epu16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_avg_epu8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_avg_epu16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_madd_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_max_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_max_epu8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_min_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_min_epu8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_mulhi_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_mulhi_epu16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_mullo_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m64 Rt_mm_mul_su32(Rt_m64 a, Rt_m64 b); +extern Rt_m128i Rt_mm_mul_epu32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_sad_epu8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_sub_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_sub_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_sub_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m64 Rt_mm_sub_si64(Rt_m64 a, Rt_m64 b); +extern Rt_m128i Rt_mm_sub_epi64(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_subs_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_subs_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_subs_epu8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_subs_epu16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_and_si128(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_andnot_si128(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_or_si128(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_xor_si128(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_slli_si128(Rt_m128i a, int imm); +extern Rt_m128i Rt_mm_slli_epi16(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_sll_epi16(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_slli_epi32(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_sll_epi32(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_slli_epi64(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_sll_epi64(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_srai_epi16(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_sra_epi16(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_srai_epi32(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_sra_epi32(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_srli_si128(Rt_m128i a, int imm); +extern Rt_m128i Rt_mm_srli_epi16(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_srl_epi16(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_srli_epi32(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_srl_epi32(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_srli_epi64(Rt_m128i a, int count); +extern Rt_m128i Rt_mm_srl_epi64(Rt_m128i a, Rt_m128i count); +extern Rt_m128i Rt_mm_cmpeq_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmpeq_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmpeq_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmpgt_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmpgt_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmpgt_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmplt_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmplt_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cmplt_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_cvtsi32_si128(int a); +extern int Rt_mm_cvtsi128_si32(Rt_m128i a); + +/* Miscellaneous Operations */ + +extern Rt_m64 Rt_mm_movepi64_pi64(Rt_m128i a); +extern Rt_m128i Rt_mm_movpi64_epi64(Rt_m64 a); +extern Rt_m128i Rt_mm_move_epi64(Rt_m128i a); +extern Rt_m128i Rt_mm_packs_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_packs_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_packus_epi16(Rt_m128i a, Rt_m128i b); +extern int Rt_mm_extract_epi16(Rt_m128i a, int imm); +extern Rt_m128i Rt_mm_insert_epi16(Rt_m128i a, int b, int imm); +extern int Rt_mm_movemask_epi8(Rt_m128i a); +extern Rt_m128i Rt_mm_shuffle_epi32(Rt_m128i a, int imm); +extern Rt_m128i Rt_mm_shufflehi_epi16(Rt_m128i a, int imm); +extern Rt_m128i Rt_mm_shufflelo_epi16(Rt_m128i a, int imm); +extern Rt_m128i Rt_mm_unpackhi_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpackhi_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpackhi_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpackhi_epi64(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpacklo_epi8(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpacklo_epi16(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpacklo_epi32(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_unpacklo_epi64(Rt_m128i a, Rt_m128i b); +extern Rt_m128i Rt_mm_loadl_epi64(Rt_m128i const *p); +extern Rt_m128i Rt_mm_load_si128(const Rt_m128i * p); +extern Rt_m128i Rt_mm_loadu_si128(const Rt_m128i * p); +extern Rt_m128i Rt_mm_set_epi64(Rt_m64 q1, Rt_m64 q0); +extern Rt_m128i Rt_mm_set_epi32(int i3, int i2, int i1, int i0); +extern Rt_m128i + Rt_mm_set_epi16(short w7, short w6, + short w5, short w4, short w3, short w2, + short w1, short w0); +extern Rt_m128i Rt_mm_set_epi8(char b15, char b14, + char b13, char b12, + char b11, char b10, + char b9, char b8, + char b7, char b6, + char b5, char b4, + char b3, char b2, + char b1, char b0); +extern Rt_m128i Rt_mm_set1_epi64(Rt_m64 q); +extern Rt_m128i Rt_mm_set1_epi32(int i); +extern Rt_m128i Rt_mm_set1_epi16(short w); +extern Rt_m128i Rt_mm_set1_epi8(char b); +extern Rt_m128i Rt_mm_setr_epi64(Rt_m64 q0, Rt_m64 q1); +extern Rt_m128i Rt_mm_setr_epi32(int i0, int i1, int i2, + int i3); +extern Rt_m128i Rt_mm_setr_epi16(short w0, short w1, + short w2, short w3, + short w4, short w5, + short w6, short w7); +extern Rt_m128i Rt_mm_setr_epi8(char b0, char b1, + char b2, char b3, + char b4, char b5, + char b6, char b7, + char b8, char b9, + char b10, char b11, + char b12, char b13, + char b14, char b15); +extern Rt_m128i Rt_mm_setzero_si128(void); + +/* Store Operations */ + +extern void Rt_mm_store_si128(Rt_m128i * p, Rt_m128i a); +extern void Rt_mm_storeu_si128(Rt_m128i * p, Rt_m128i a); +extern void Rt_mm_maskmoveu_si128(Rt_m128i s, Rt_m128i n, + char *p); +extern void Rt_mm_storel_epi64(Rt_m128i * p, Rt_m128i a); +extern void Rt_mm_stream_si128(Rt_m128i * p, Rt_m128i a); +extern void Rt_mm_stream_si32(int *p, int a); +extern void Rt_mm_clflush(void const *p); +extern void Rt_mm_lfence(void); +extern void Rt_mm_mfence(void); + +/* + * API + */ +extern RwUInt32 RtIntelRDTSC(void); +extern RwUInt32 RtIntelToggleEFLAGS(int mask); +extern RwUInt32 RtIntelCPUID(RwUInt32 level, + void *pb, void *pc, void *pd); +extern RwUInt32 RtIntelHaveCPUID(void); +extern RwUInt32 RtIntelHaveRDTSC(void); +extern RwUInt32 RtIntelHaveMMX(void); +extern RwUInt32 RtIntelHaveSSE(void); +extern RwUInt32 RtIntelHaveWNI(void); +extern RwUInt32 RtIntelCpuType(void); + +extern RwBool RtIntelStartTiming(void * data); +extern RwBool RtIntelStopTiming(void *data); +extern RwBool RtIntelTime(RwFixed64 * result, + RtIntelTimeFunction func, + void *data); + +extern RwBool RtIntelPluginAttach(void); + +extern RtIntelOverload *_rtIntelOverloadGetHandle(void); + +extern RxNodeDefinition *RxNodeDefinitionGetSSETransformCSL(void); + + +#ifdef __cplusplus +} +#endif /* __cplusplus */ + +/* + * LEGACY-SUPPORT -- e.g. + * rwsdk/driver/d3d/baintd3d.c + * is locked at time of writing + */ + +#define RpIntelRDTSC() RtIntelRDTSC() +#define RpIntelToggleEFLAGS(_mask) RtIntelToggleEFLAGS(_mask) +#define RpIntelCPUID(_level, _pb, _pc, _pd) \ + RtIntelCPUID(_level, _pb, _pc, _pd) +#define RpIntelHaveCPUID() RtIntelHaveCPUID() +#define RpIntelHaveRDTSC() RtIntelHaveRDTSC() +#define RpIntelHaveMMX() RtIntelHaveMMX() +#define RpIntelHaveSSE() RtIntelHaveSSE() +#define RpIntelHaveWNI() RtIntelHaveWNI() +#define RpIntelCpuType() RtIntelCpuType() +#define RpIntelStartTiming(_data) RtIntelStartTiming(_data) +#define RpIntelStopTiming(_data) RtIntelStopTiming(_data) +#define RpIntelTime(_result, _func, _data) \ + RtIntelTime(_result, _func, _data) +#define RpIntelPluginAttach() RtIntelPluginAttach() +#define RpNodeDefinitionGetSSETransformCSL() \ + RxNodeDefinitionGetSSETransformCSL() + +typedef RtIntelOverload RwIntelOverload; +typedef RtIntelOverload RpIntelOverload; + +#define _rwIntelOverloadGetHandle() _rtIntelOverloadGetHandle() +#define _rpIntelOverloadGetHandle() _rtIntelOverloadGetHandle() +#define RwIntelHaveSSE() RtIntelHaveSSE() +#define RpIntelHaveSSE() RtIntelHaveSSE() + +#if (defined(RWEMULATEINTELSIMD) || !defined(__ICL)) + +/* + * MMX + */ + +/* General support intrinsics */ +#define _m_empty() Rt_m_empty() +#define _m_from_int(i) Rt_m_from_int(i) +#define _m_to_int(m) Rt_m_to_int(m) +#define _m_packsswb(m1, m2) Rt_m_packsswb(m1, m2) +#define _m_packssdw(m1, m2) Rt_m_packssdw(m1, m2) +#define _m_packuswb(m1, m2) Rt_m_packuswb(m1, m2) +#define _m_punpckhbw(m1, m2) Rt_m_punpckhbw(m1, m2) +#define _m_punpckhwd(m1, m2) Rt_m_punpckhwd(m1, m2) +#define _m_punpckhdq(m1, m2) Rt_m_punpckhdq(m1, m2) +#define _m_punpcklbw(m1, m2) Rt_m_punpcklbw(m1, m2) +#define _m_punpcklwd(m1, m2) Rt_m_punpcklwd(m1, m2) +#define _m_punpckldq(m1, m2) Rt_m_punpckldq(m1, m2) + +/* Packed arithmetic intrinsics */ +#define _m_paddb(m1, m2) Rt_m_paddb(m1, m2) +#define _m_paddw(m1, m2) Rt_m_paddw(m1, m2) +#define _m_paddd(m1, m2) Rt_m_paddd(m1, m2) +#define _m_paddsb(m1, m2) Rt_m_paddsb(m1, m2) +#define _m_paddsw(m1, m2) Rt_m_paddsw(m1, m2) +#define _m_paddusb(m1, m2) Rt_m_paddusb(m1, m2) +#define _m_paddusw(m1, m2) Rt_m_paddusw(m1, m2) +#define _m_psubb(m1, m2) Rt_m_psubb(m1, m2) +#define _m_psubw(m1, m2) Rt_m_psubw(m1, m2) +#define _m_psubd(m1, m2) Rt_m_psubd(m1, m2) +#define _m_psubsb(m1, m2) Rt_m_psubsb(m1, m2) +#define _m_psubsw(m1, m2) Rt_m_psubsw(m1, m2) +#define _m_psubusb(m1, m2) Rt_m_psubusb(m1, m2) +#define _m_psubusw(m1, m2) Rt_m_psubusw(m1, m2) +#define _m_pmaddwd(m1, m2) Rt_m_pmaddwd(m1, m2) +#define _m_pmulhw(m1, m2) Rt_m_pmulhw(m1, m2) +#define _m_pmullw(m1, m2) Rt_m_pmullw(m1, m2) + +/* Shift intrinsics */ +#define _m_psllw(m, count) Rt_m_psllw(m, count) +#define _m_psllwi(m, count) Rt_m_psllwi(m, count) +#define _m_pslld(m, count) Rt_m_pslld(m, count) +#define _m_pslldi(m, count) Rt_m_pslldi(m, count) +#define _m_psllq(m, count) Rt_m_psllq(m, count) +#define _m_psllqi(m, count) Rt_m_psllqi(m, count) +#define _m_psraw(m, count) Rt_m_psraw(m, count) +#define _m_psrawi(m, count) Rt_m_psrawi(m, count) +#define _m_psrad(m, count) Rt_m_psrad(m, count) +#define _m_psradi(m, count) Rt_m_psradi(m, count) +#define _m_psrlw(m, count) Rt_m_psrlw(m, count) +#define _m_psrlwi(m, count) Rt_m_psrlwi(m, count) +#define _m_psrld(m, count) Rt_m_psrld(m, count) +#define _m_psrldi(m, count) Rt_m_psrldi(m, count) +#define _m_psrlq(m, count) Rt_m_psrlq(m, count) +#define _m_psrlqi(m, count) Rt_m_psrlqi(m, count) + +/* Logical intrinsics */ +#define _m_pand(m1, m2) Rt_m_pand(m1, m2) +#define _m_pandn(m1, m2) Rt_m_pandn(m1, m2) +#define _m_por(m1, m2) Rt_m_por(m1, m2) +#define _m_pxor(m1, m2) Rt_m_pxor(m1, m2) + +/* Comparison intrinsics */ +#define _m_pcmpeqb(m1, m2) Rt_m_pcmpeqb(m1, m2) +#define _m_pcmpeqw(m1, m2) Rt_m_pcmpeqw(m1, m2) +#define _m_pcmpeqd(m1, m2) Rt_m_pcmpeqd(m1, m2) +#define _m_pcmpgtb(m1, m2) Rt_m_pcmpgtb(m1, m2) +#define _m_pcmpgtw(m1, m2) Rt_m_pcmpgtw(m1, m2) +#define _m_pcmpgtd(m1, m2) Rt_m_pcmpgtd(m1, m2) + +/* + * SSE + */ + +/* + * Arithmetic Operations + */ + +#define _mm_add_ss(a, b) Rt_mm_add_ss(a, b) +#define _mm_add_ps(a, b) Rt_mm_add_ps(a, b) +#define _mm_sub_ss(a, b) Rt_mm_sub_ss(a, b) +#define _mm_sub_ps(a, b) Rt_mm_sub_ps(a, b) +#define _mm_mul_ss(a, b) Rt_mm_mul_ss(a, b) +#define _mm_mul_ps(a, b) Rt_mm_mul_ps(a, b) +#define _mm_div_ss(a, b) Rt_mm_div_ss(a, b) +#define _mm_div_ps(a, b) Rt_mm_div_ps(a, b) +#define _mm_sqrt_ss(a) Rt_mm_sqrt_ss(a) +#define _mm_sqrt_ps(a) Rt_mm_sqrt_ps(a) +#define _mm_rcp_ss(a) Rt_mm_rcp_ss(a) +#define _mm_rcp_ps(a) Rt_mm_rcp_ps(a) +#define _mm_rsqrt_ss(a) Rt_mm_rsqrt_ss(a) +#define _mm_rsqrt_ps(a) Rt_mm_rsqrt_ps(a) +#define _mm_min_ss(a, b) Rt_mm_min_ss(a, b) +#define _mm_min_ps(a, b) Rt_mm_min_ps(a, b) +#define _mm_max_ss(a, b) Rt_mm_max_ss(a, b) +#define _mm_max_ps(a, b) Rt_mm_max_ps(a, b) + +/* + * Logical Operations + */ + +#define _mm_and_ps(a, b) Rt_mm_and_ps(a, b) +#define _mm_andnot_ps(a, b) Rt_mm_andnot_ps(a, b) +#define _mm_or_ps(a, b) Rt_mm_or_ps(a, b) +#define _mm_xor_ps(a, b) Rt_mm_xor_ps(a, b) + +/* + * Comparisons + */ + +#define _mm_cmpeq_ss(a, b) Rt_mm_cmpeq_ss(a, b) +#define _mm_cmpeq_ps(a, b) Rt_mm_cmpeq_ps(a, b) +#define _mm_cmplt_ss(a, b) Rt_mm_cmplt_ss(a, b) +#define _mm_cmplt_ps(a, b) Rt_mm_cmplt_ps(a, b) +#define _mm_cmple_ss(a, b) Rt_mm_cmple_ss(a, b) +#define _mm_cmple_ps(a, b) Rt_mm_cmple_ps(a, b) +#define _mm_cmpgt_ss(a, b) Rt_mm_cmpgt_ss(a, b) +#define _mm_cmpgt_ps(a, b) Rt_mm_cmpgt_ps(a, b) +#define _mm_cmpge_ss(a, b) Rt_mm_cmpge_ss(a, b) +#define _mm_cmpge_ps(a, b) Rt_mm_cmpge_ps(a, b) +#define _mm_cmpneq_ss(a, b) Rt_mm_cmpneq_ss(a, b) +#define _mm_cmpneq_ps(a, b) Rt_mm_cmpneq_ps(a, b) +#define _mm_cmpnlt_ss(a, b) Rt_mm_cmpnlt_ss(a, b) +#define _mm_cmpnlt_ps(a, b) Rt_mm_cmpnlt_ps(a, b) +#define _mm_cmpnle_ss(a, b) Rt_mm_cmpnle_ss(a, b) +#define _mm_cmpnle_ps(a, b) Rt_mm_cmpnle_ps(a, b) +#define _mm_cmpngt_ss(a, b) Rt_mm_cmpngt_ss(a, b) +#define _mm_cmpngt_ps(a, b) Rt_mm_cmpngt_ps(a, b) +#define _mm_cmpnge_ss(a, b) Rt_mm_cmpnge_ss(a, b) +#define _mm_cmpnge_ps(a, b) Rt_mm_cmpnge_ps(a, b) +#define _mm_cmpord_ss(a, b) Rt_mm_cmpord_ss(a, b) +#define _mm_cmpord_ps(a, b) Rt_mm_cmpord_ps(a, b) +#define _mm_cmpunord_ss(a, b) Rt_mm_cmpunord_ss(a, b) +#define _mm_cmpunord_ps(a, b) Rt_mm_cmpunord_ps(a, b) +#define _mm_comieq_ss(a, b) Rt_mm_comieq_ss(a, b) +#define _mm_comilt_ss(a, b) Rt_mm_comilt_ss(a, b) +#define _mm_comile_ss(a, b) Rt_mm_comile_ss(a, b) +#define _mm_comigt_ss(a, b) Rt_mm_comigt_ss(a, b) +#define _mm_comige_ss(a, b) Rt_mm_comige_ss(a, b) +#define _mm_comineq_ss(a, b) Rt_mm_comineq_ss(a, b) +#define _mm_ucomieq_ss(a, b) Rt_mm_ucomieq_ss(a, b) +#define _mm_ucomilt_ss(a, b) Rt_mm_ucomilt_ss(a, b) +#define _mm_ucomile_ss(a, b) Rt_mm_ucomile_ss(a, b) +#define _mm_ucomigt_ss(a, b) Rt_mm_ucomigt_ss(a, b) +#define _mm_ucomige_ss(a, b) Rt_mm_ucomige_ss(a, b) +#define _mm_ucomineq_ss(a, b) Rt_mm_ucomineq_ss(a, b) + +/* + * Conversion Operations + */ + +#define _mm_cvt_ss2si(a) Rt_mm_cvt_ss2si(a) +#define _mm_cvt_ps2pi(a) Rt_mm_cvt_ps2pi(a) +#define _mm_cvtt_ss2si(a) Rt_mm_cvtt_ss2si(a) +#define _mm_cvtt_ps2pi(a) Rt_mm_cvtt_ps2pi(a) +#define _mm_cvt_si2ss(a, b) Rt_mm_cvt_si2ss(a, b) +#define _mm_cvt_pi2ps(a, b) Rt_mm_cvt_pi2ps(a, b) + +/* + * Miscellaneous + */ + +#define _mm_shuffle_ps(a, b, i) Rt_mm_shuffle_ps(a, b, i) +#define _mm_unpackhi_ps(a, b) Rt_mm_unpackhi_ps(a, b) +#define _mm_unpacklo_ps(a, b) Rt_mm_unpacklo_ps(a, b) +#define _mm_loadh_pi(a, p) Rt_mm_loadh_pi(a, p) +#define _mm_storeh_pi(p, a) Rt_mm_storeh_pi(p, a) +#define _mm_movehl_ps(a, b) Rt_mm_movehl_ps(a, b) +#define _mm_movelh_ps(a, b) Rt_mm_movelh_ps(a, b) +#define _mm_loadl_pi(a, p) Rt_mm_loadl_pi(a, p) +#define _mm_storel_pi(p, a) Rt_mm_storel_pi(p, a) +#define _mm_movemask_ps(a) Rt_mm_movemask_ps(a) +#define _mm_getcsr() Rt_mm_getcsr() +#define _mm_setcsr(i) Rt_mm_setcsr(i) + +/* + *Load Operations + */ + +#define _mm_load_ss(p) Rt_mm_load_ss(p) +#define _mm_load_ps1(p) Rt_mm_load_ps1(p) +#define _mm_load_ps(p) Rt_mm_load_ps(p) +#define _mm_loadu_ps(p) Rt_mm_loadu_ps(p) +#define _mm_loadr_ps(p) Rt_mm_loadr_ps(p) + +/* + * Set Operations + */ + +#define _mm_set_ss(w) Rt_mm_set_ss(w) +#define _mm_set_ps1(w) Rt_mm_set_ps1(w) +#define _mm_set_ps(z, y, x, w) Rt_mm_set_ps(z, y, x, w) +#define _mm_setr_ps(z, y, x, w) Rt_mm_setr_ps(z, y, x, w) +#define _mm_setzero_ps() Rt_mm_setzero_ps() + +/* + * Store Operations + */ + +#define _mm_store_ss(p, a) Rt_mm_store_ss(p, a) +#define _mm_store_ps1(p, a) Rt_mm_store_ps1(p, a) +#define _mm_store_ps(p, a) Rt_mm_store_ps(p, a) +#define _mm_storeu_ps(p, a) Rt_mm_storeu_ps(p, a) +#define _mm_storer_ps(p, a) Rt_mm_storer_ps(p, a) +#define _mm_move_ss(a, b) Rt_mm_move_ss(a, b) + +/* + * Integer Intrinsics + */ + +#define _m_pextrw(a, n) Rt_m_pextrw(a, n) +#define _m_pinsrw(a, d, n) Rt_m_pinsrw(a, d, n) +#define _m_pmaxsw(a, b) Rt_m_pmaxsw(a, b) +#define _m_pmaxub(a, b) Rt_m_pmaxub(a, b) +#define _m_pminsw(a, b) Rt_m_pminsw(a, b) +#define _m_pminub(a, b) Rt_m_pminub(a, b) +#define _m_pmovmskb(a) Rt_m_pmovmskb(a) +#define _m_pmulhuw(a, b) Rt_m_pmulhuw(a, b) +#define _m_pshufw(a, n) Rt_m_pshufw(a, n) +#define _m_lwmaskmovq(d, n, p) Rt_m_lwmaskmovq(d, n, p) + +/* + * Cacheability Support + */ + +#define _mm_prefetch(p, i) Rt_mm_prefetch(p, i) +#define _mm_stream_pi(p, a) Rt_mm_stream_pi(p, a) +#define _mm_stream_ps(p, a) Rt_mm_stream_ps(p, a) +#define _mm_sfence() Rt_mm_sfence() + +#endif /* (defined(RWEMULATEINTELSIMD) || !defined(__ICL)) */ + +#endif /* RTINTEL_H */ |