libavcodec.hg: ppc/h264_altivec.c comparison

comparison ppc/h264_altivec.c @ 8494:1615d6b75ada libavcodec

Cleanup _t types in libavcodec/ppc

author	lu_zero
date	Sat, 27 Dec 2008 11:21:28 +0000
parents	11307ea31e57
children	992e2f8bfba7

comparison

equal deleted inserted replaced

-:469f3e5bcf13
+:1615d6b75ada
 {((8 - x) * (8 - y)),
 ((x) * (8 - y)),
 ((8 - x) * (y)),
 ((x) * (y))};
 register int i;
-vec_u8_t fperm;
+vec_u8 fperm;
-const vec_s32_t vABCD = vec_ld(0, ABCD);
+const vec_s32 vABCD = vec_ld(0, ABCD);
-const vec_s16_t vA = vec_splat((vec_s16_t)vABCD, 1);
+const vec_s16 vA = vec_splat((vec_s16)vABCD, 1);
-const vec_s16_t vB = vec_splat((vec_s16_t)vABCD, 3);
+const vec_s16 vB = vec_splat((vec_s16)vABCD, 3);
-const vec_s16_t vC = vec_splat((vec_s16_t)vABCD, 5);
+const vec_s16 vC = vec_splat((vec_s16)vABCD, 5);
-const vec_s16_t vD = vec_splat((vec_s16_t)vABCD, 7);
+const vec_s16 vD = vec_splat((vec_s16)vABCD, 7);
 LOAD_ZERO;
-const vec_s16_t v28ss = vec_sub(vec_sl(vec_splat_s16(1),vec_splat_u16(5)),vec_splat_s16(4));
+const vec_s16 v28ss = vec_sub(vec_sl(vec_splat_s16(1),vec_splat_u16(5)),vec_splat_s16(4));
-const vec_u16_t v6us  = vec_splat_u16(6);
+const vec_u16 v6us  = vec_splat_u16(6);
 register int loadSecond     = (((unsigned long)src) % 16) <= 7 ? 0 : 1;
 register int reallyBadAlign = (((unsigned long)src) % 16) == 15 ? 1 : 0;
-vec_u8_t vsrcAuc, vsrcBuc, vsrcperm0, vsrcperm1;
+vec_u8 vsrcAuc, vsrcBuc, vsrcperm0, vsrcperm1;
-vec_u8_t vsrc0uc, vsrc1uc;
+vec_u8 vsrc0uc, vsrc1uc;
-vec_s16_t vsrc0ssH, vsrc1ssH;
+vec_s16 vsrc0ssH, vsrc1ssH;
-vec_u8_t vsrcCuc, vsrc2uc, vsrc3uc;
+vec_u8 vsrcCuc, vsrc2uc, vsrc3uc;
-vec_s16_t vsrc2ssH, vsrc3ssH, psum;
+vec_s16 vsrc2ssH, vsrc3ssH, psum;
-vec_u8_t vdst, ppsum, fsum;
+vec_u8 vdst, ppsum, fsum;
 if (((unsigned long)dst) % 16 == 0) {
-fperm = (vec_u8_t){0x10, 0x11, 0x12, 0x13,
+fperm = (vec_u8){0x10, 0x11, 0x12, 0x13,
 0x14, 0x15, 0x16, 0x17,
 0x08, 0x09, 0x0A, 0x0B,
 0x0C, 0x0D, 0x0E, 0x0F};
 } else {
-fperm = (vec_u8_t){0x00, 0x01, 0x02, 0x03,
+fperm = (vec_u8){0x00, 0x01, 0x02, 0x03,
 0x04, 0x05, 0x06, 0x07,
 0x18, 0x19, 0x1A, 0x1B,
 0x1C, 0x1D, 0x1E, 0x1F};
 }
 if (reallyBadAlign)
 vsrc1uc = vsrcBuc;
 else
 vsrc1uc = vec_perm(vsrcAuc, vsrcBuc, vsrcperm1);
-vsrc0ssH = (vec_s16_t)vec_mergeh(zero_u8v, (vec_u8_t)vsrc0uc);
+vsrc0ssH = (vec_s16)vec_mergeh(zero_u8v, (vec_u8)vsrc0uc);
-vsrc1ssH = (vec_s16_t)vec_mergeh(zero_u8v, (vec_u8_t)vsrc1uc);
+vsrc1ssH = (vec_s16)vec_mergeh(zero_u8v, (vec_u8)vsrc1uc);
 if (!loadSecond) {// -> !reallyBadAlign
 for (i = 0 ; i < h ; i++) {
 vsrcCuc = vec_ld(stride + 0, src);
 vsrc2uc = vec_perm(vsrcCuc, vsrcCuc, vsrcperm0);
 vsrc3uc = vec_perm(vsrcCuc, vsrcCuc, vsrcperm1);
-vsrc2ssH = (vec_s16_t)vec_mergeh(zero_u8v, (vec_u8_t)vsrc2uc);
+vsrc2ssH = (vec_s16)vec_mergeh(zero_u8v, (vec_u8)vsrc2uc);
-vsrc3ssH = (vec_s16_t)vec_mergeh(zero_u8v, (vec_u8_t)vsrc3uc);
+vsrc3ssH = (vec_s16)vec_mergeh(zero_u8v, (vec_u8)vsrc3uc);
 psum = vec_mladd(vA, vsrc0ssH, vec_splat_s16(0));
 psum = vec_mladd(vB, vsrc1ssH, psum);
 psum = vec_mladd(vC, vsrc2ssH, psum);
 psum = vec_mladd(vD, vsrc3ssH, psum);
 psum = vec_add(v28ss, psum);
 psum = vec_sra(psum, v6us);
 vdst = vec_ld(0, dst);
-ppsum = (vec_u8_t)vec_packsu(psum, psum);
+ppsum = (vec_u8)vec_packsu(psum, psum);
 fsum = vec_perm(vdst, ppsum, fperm);
 vec_st(fsum, 0, dst);
 vsrc0ssH = vsrc2ssH;
 dst += stride;
 src += stride;
 }
 } else {
-vec_u8_t vsrcDuc;
+vec_u8 vsrcDuc;
 for (i = 0 ; i < h ; i++) {
 vsrcCuc = vec_ld(stride + 0, src);
 vsrcDuc = vec_ld(stride + 16, src);
 vsrc2uc = vec_perm(vsrcCuc, vsrcDuc, vsrcperm0);
 if (reallyBadAlign)
 vsrc3uc = vsrcDuc;
 else
 vsrc3uc = vec_perm(vsrcCuc, vsrcDuc, vsrcperm1);
-vsrc2ssH = (vec_s16_t)vec_mergeh(zero_u8v, (vec_u8_t)vsrc2uc);
+vsrc2ssH = (vec_s16)vec_mergeh(zero_u8v, (vec_u8)vsrc2uc);
-vsrc3ssH = (vec_s16_t)vec_mergeh(zero_u8v, (vec_u8_t)vsrc3uc);
+vsrc3ssH = (vec_s16)vec_mergeh(zero_u8v, (vec_u8)vsrc3uc);
 psum = vec_mladd(vA, vsrc0ssH, vec_splat_s16(0));
 psum = vec_mladd(vB, vsrc1ssH, psum);
 psum = vec_mladd(vC, vsrc2ssH, psum);
 psum = vec_mladd(vD, vsrc3ssH, psum);
 psum = vec_add(v28ss, psum);
 psum = vec_sr(psum, v6us);
 vdst = vec_ld(0, dst);
-ppsum = (vec_u8_t)vec_pack(psum, psum);
+ppsum = (vec_u8)vec_pack(psum, psum);
 fsum = vec_perm(vdst, ppsum, fperm);
 vec_st(fsum, 0, dst);
 vsrc0ssH = vsrc2ssH;
 static inline void put_pixels16_l2_altivec( uint8_t * dst, const uint8_t * src1,
 const uint8_t * src2, int dst_stride,
 int src_stride1, int h)
 {
 int i;
-vec_u8_t a, b, d, tmp1, tmp2, mask, mask_, edges, align;
+vec_u8 a, b, d, tmp1, tmp2, mask, mask_, edges, align;
 mask_ = vec_lvsl(0, src2);
 for (i = 0; i < h; i++) {
 static inline void avg_pixels16_l2_altivec( uint8_t * dst, const uint8_t * src1,
 const uint8_t * src2, int dst_stride,
 int src_stride1, int h)
 {
 int i;
-vec_u8_t a, b, d, tmp1, tmp2, mask, mask_, edges, align;
+vec_u8 a, b, d, tmp1, tmp2, mask, mask_, edges, align;
 mask_ = vec_lvsl(0, src2);
 for (i = 0; i < h; i++) {
 b3 = vec_mergel( a1, a3 )
 #define VEC_LOAD_U8_ADD_S16_STORE_U8(va)                      \
 vdst_orig = vec_ld(0, dst);                               \
 vdst = vec_perm(vdst_orig, zero_u8v, vdst_mask);          \
-vdst_ss = (vec_s16_t) vec_mergeh(zero_u8v, vdst);         \
+vdst_ss = (vec_s16) vec_mergeh(zero_u8v, vdst);         \
 va = vec_add(va, vdst_ss);                                \
 va_u8 = vec_packsu(va, zero_s16v);                        \
-va_u32 = vec_splat((vec_u32_t)va_u8, 0);                  \
+va_u32 = vec_splat((vec_u32)va_u8, 0);                  \
 vec_ste(va_u32, element, (uint32_t*)dst);
 static void ff_h264_idct_add_altivec(uint8_t *dst, DCTELEM *block, int stride)
 {
-vec_s16_t va0, va1, va2, va3;
+vec_s16 va0, va1, va2, va3;
-vec_s16_t vz0, vz1, vz2, vz3;
+vec_s16 vz0, vz1, vz2, vz3;
-vec_s16_t vtmp0, vtmp1, vtmp2, vtmp3;
+vec_s16 vtmp0, vtmp1, vtmp2, vtmp3;
-vec_u8_t va_u8;
+vec_u8 va_u8;
-vec_u32_t va_u32;
+vec_u32 va_u32;
-vec_s16_t vdst_ss;
+vec_s16 vdst_ss;
-const vec_u16_t v6us = vec_splat_u16(6);
+const vec_u16 v6us = vec_splat_u16(6);
-vec_u8_t vdst, vdst_orig;
+vec_u8 vdst, vdst_orig;
-vec_u8_t vdst_mask = vec_lvsl(0, dst);
+vec_u8 vdst_mask = vec_lvsl(0, dst);
 int element = ((unsigned long)dst & 0xf) >> 2;
 LOAD_ZERO;
 block[0] += 32;  /* add 32 as a DC-level for rounding */
 VEC_LOAD_U8_ADD_S16_STORE_U8(va3);
 }
 #define IDCT8_1D_ALTIVEC(s0, s1, s2, s3, s4, s5, s6, s7,  d0, d1, d2, d3, d4, d5, d6, d7) {\
 /*        a0  = SRC(0) + SRC(4); */ \
-vec_s16_t a0v = vec_add(s0, s4);    \
+vec_s16 a0v = vec_add(s0, s4);    \
 /*        a2  = SRC(0) - SRC(4); */ \
-vec_s16_t a2v = vec_sub(s0, s4);    \
+vec_s16 a2v = vec_sub(s0, s4);    \
 /*        a4  =           (SRC(2)>>1) - SRC(6); */ \
-vec_s16_t a4v = vec_sub(vec_sra(s2, onev), s6);    \
+vec_s16 a4v = vec_sub(vec_sra(s2, onev), s6);    \
 /*        a6  =           (SRC(6)>>1) + SRC(2); */ \
-vec_s16_t a6v = vec_add(vec_sra(s6, onev), s2);    \
+vec_s16 a6v = vec_add(vec_sra(s6, onev), s2);    \
 /*        b0  =         a0 + a6; */ \
-vec_s16_t b0v = vec_add(a0v, a6v);  \
+vec_s16 b0v = vec_add(a0v, a6v);  \
 /*        b2  =         a2 + a4; */ \
-vec_s16_t b2v = vec_add(a2v, a4v);  \
+vec_s16 b2v = vec_add(a2v, a4v);  \
 /*        b4  =         a2 - a4; */ \
-vec_s16_t b4v = vec_sub(a2v, a4v);  \
+vec_s16 b4v = vec_sub(a2v, a4v);  \
 /*        b6  =         a0 - a6; */ \
-vec_s16_t b6v = vec_sub(a0v, a6v);  \
+vec_s16 b6v = vec_sub(a0v, a6v);  \
 /* a1 =  SRC(5) - SRC(3) - SRC(7) - (SRC(7)>>1); */ \
 /*        a1 =             (SRC(5)-SRC(3)) -  (SRC(7)  +  (SRC(7)>>1)); */ \
-vec_s16_t a1v = vec_sub( vec_sub(s5, s3), vec_add(s7, vec_sra(s7, onev)) ); \
+vec_s16 a1v = vec_sub( vec_sub(s5, s3), vec_add(s7, vec_sra(s7, onev)) ); \
 /* a3 =  SRC(7) + SRC(1) - SRC(3) - (SRC(3)>>1); */ \
 /*        a3 =             (SRC(7)+SRC(1)) -  (SRC(3)  +  (SRC(3)>>1)); */ \
-vec_s16_t a3v = vec_sub( vec_add(s7, s1), vec_add(s3, vec_sra(s3, onev)) );\
+vec_s16 a3v = vec_sub( vec_add(s7, s1), vec_add(s3, vec_sra(s3, onev)) );\
 /* a5 =  SRC(7) - SRC(1) + SRC(5) + (SRC(5)>>1); */ \
 /*        a5 =             (SRC(7)-SRC(1)) +   SRC(5) +   (SRC(5)>>1); */ \
-vec_s16_t a5v = vec_add( vec_sub(s7, s1), vec_add(s5, vec_sra(s5, onev)) );\
+vec_s16 a5v = vec_add( vec_sub(s7, s1), vec_add(s5, vec_sra(s5, onev)) );\
 /*        a7 =                SRC(5)+SRC(3) +  SRC(1) +   (SRC(1)>>1); */ \
-vec_s16_t a7v = vec_add( vec_add(s5, s3), vec_add(s1, vec_sra(s1, onev)) );\
+vec_s16 a7v = vec_add( vec_add(s5, s3), vec_add(s1, vec_sra(s1, onev)) );\
 /*        b1 =                  (a7>>2)  +  a1; */ \
-vec_s16_t b1v = vec_add( vec_sra(a7v, twov), a1v); \
+vec_s16 b1v = vec_add( vec_sra(a7v, twov), a1v); \
 /*        b3 =          a3 +        (a5>>2); */ \
-vec_s16_t b3v = vec_add(a3v, vec_sra(a5v, twov)); \
+vec_s16 b3v = vec_add(a3v, vec_sra(a5v, twov)); \
 /*        b5 =                  (a3>>2)  -   a5; */ \
-vec_s16_t b5v = vec_sub( vec_sra(a3v, twov), a5v); \
+vec_s16 b5v = vec_sub( vec_sra(a3v, twov), a5v); \
 /*        b7 =           a7 -        (a1>>2); */ \
-vec_s16_t b7v = vec_sub( a7v, vec_sra(a1v, twov)); \
+vec_s16 b7v = vec_sub( a7v, vec_sra(a1v, twov)); \
 /* DST(0,    b0 + b7); */ \
 d0 = vec_add(b0v, b7v); \
 /* DST(1,    b2 + b5); */ \
 d1 = vec_add(b2v, b5v); \
 /* DST(2,    b4 + b3); */ \
 d7 = vec_sub(b0v, b7v); \
 }
 #define ALTIVEC_STORE_SUM_CLIP(dest, idctv, perm_ldv, perm_stv, sel) { \
 /* unaligned load */                                       \
-vec_u8_t hv = vec_ld( 0, dest );                           \
+vec_u8 hv = vec_ld( 0, dest );                           \
-vec_u8_t lv = vec_ld( 7, dest );                           \
+vec_u8 lv = vec_ld( 7, dest );                           \
-vec_u8_t dstv   = vec_perm( hv, lv, (vec_u8_t)perm_ldv );  \
+vec_u8 dstv   = vec_perm( hv, lv, (vec_u8)perm_ldv );  \
-vec_s16_t idct_sh6 = vec_sra(idctv, sixv);                 \
+vec_s16 idct_sh6 = vec_sra(idctv, sixv);                 \
-vec_u16_t dst16 = (vec_u16_t)vec_mergeh(zero_u8v, dstv);   \
+vec_u16 dst16 = (vec_u16)vec_mergeh(zero_u8v, dstv);   \
-vec_s16_t idstsum = vec_adds(idct_sh6, (vec_s16_t)dst16);  \
+vec_s16 idstsum = vec_adds(idct_sh6, (vec_s16)dst16);  \
-vec_u8_t idstsum8 = vec_packsu(zero_s16v, idstsum);        \
+vec_u8 idstsum8 = vec_packsu(zero_s16v, idstsum);        \
-vec_u8_t edgehv;                                           \
+vec_u8 edgehv;                                           \
 /* unaligned store */                                      \
-vec_u8_t bodyv  = vec_perm( idstsum8, idstsum8, perm_stv );\
+vec_u8 bodyv  = vec_perm( idstsum8, idstsum8, perm_stv );\
-vec_u8_t edgelv = vec_perm( sel, zero_u8v, perm_stv );     \
+vec_u8 edgelv = vec_perm( sel, zero_u8v, perm_stv );     \
 lv    = vec_sel( lv, bodyv, edgelv );                      \
 vec_st( lv, 7, dest );                                     \
 hv    = vec_ld( 0, dest );                                 \
 edgehv = vec_perm( zero_u8v, sel, perm_stv );              \
 hv    = vec_sel( hv, bodyv, edgehv );                      \
 vec_st( hv, 0, dest );                                     \
 }
 void ff_h264_idct8_add_altivec( uint8_t *dst, DCTELEM *dct, int stride ) {
-vec_s16_t s0, s1, s2, s3, s4, s5, s6, s7;
+vec_s16 s0, s1, s2, s3, s4, s5, s6, s7;
-vec_s16_t d0, d1, d2, d3, d4, d5, d6, d7;
+vec_s16 d0, d1, d2, d3, d4, d5, d6, d7;
-vec_s16_t idct0, idct1, idct2, idct3, idct4, idct5, idct6, idct7;
+vec_s16 idct0, idct1, idct2, idct3, idct4, idct5, idct6, idct7;
-vec_u8_t perm_ldv = vec_lvsl(0, dst);
+vec_u8 perm_ldv = vec_lvsl(0, dst);
-vec_u8_t perm_stv = vec_lvsr(8, dst);
+vec_u8 perm_stv = vec_lvsr(8, dst);
-const vec_u16_t onev = vec_splat_u16(1);
+const vec_u16 onev = vec_splat_u16(1);
-const vec_u16_t twov = vec_splat_u16(2);
+const vec_u16 twov = vec_splat_u16(2);
-const vec_u16_t sixv = vec_splat_u16(6);
+const vec_u16 sixv = vec_splat_u16(6);
-const vec_u8_t sel = (vec_u8_t) {0,0,0,0,0,0,0,0,-1,-1,-1,-1,-1,-1,-1,-1};
+const vec_u8 sel = (vec_u8) {0,0,0,0,0,0,0,0,-1,-1,-1,-1,-1,-1,-1,-1};
 LOAD_ZERO;
 dct[0] += 32; // rounding for the >>6 at the end
 s0 = vec_ld(0x00, (int16_t*)dct);
 }
 }
 }
 #define transpose4x16(r0, r1, r2, r3) {      \
-register vec_u8_t r4;                    \
+register vec_u8 r4;                    \
-register vec_u8_t r5;                    \
+register vec_u8 r5;                    \
-register vec_u8_t r6;                    \
+register vec_u8 r6;                    \
-register vec_u8_t r7;                    \
+register vec_u8 r7;                    \
 \
 r4 = vec_mergeh(r0, r2);  /*0, 2 set 0*/ \
 r5 = vec_mergel(r0, r2);  /*0, 2 set 1*/ \
 r6 = vec_mergeh(r1, r3);  /*1, 3 set 0*/ \
 r7 = vec_mergel(r1, r3);  /*1, 3 set 1*/ \
 r2 = vec_mergeh(r5, r7);  /*all set 2*/  \
 r3 = vec_mergel(r5, r7);  /*all set 3*/  \
 }
 static inline void write16x4(uint8_t *dst, int dst_stride,
-register vec_u8_t r0, register vec_u8_t r1,
+register vec_u8 r0, register vec_u8 r1,
-register vec_u8_t r2, register vec_u8_t r3) {
+register vec_u8 r2, register vec_u8 r3) {
 DECLARE_ALIGNED_16(unsigned char, result[64]);
 uint32_t *src_int = (uint32_t *)result, *dst_int = (uint32_t *)dst;
 int int_dst_stride = dst_stride/4;
 vec_st(r0, 0, result);
 /** \brief performs a 6x16 transpose of data in src, and stores it to dst
 \todo FIXME: see if we can't spare some vec_lvsl() by them factorizing
 out of unaligned_load() */
 #define readAndTranspose16x6(src, src_stride, r8, r9, r10, r11, r12, r13) {\
-register vec_u8_t r0  = unaligned_load(0,             src);            \
+register vec_u8 r0  = unaligned_load(0,             src);            \
-register vec_u8_t r1  = unaligned_load(   src_stride, src);            \
+register vec_u8 r1  = unaligned_load(   src_stride, src);            \
-register vec_u8_t r2  = unaligned_load(2* src_stride, src);            \
+register vec_u8 r2  = unaligned_load(2* src_stride, src);            \
-register vec_u8_t r3  = unaligned_load(3* src_stride, src);            \
+register vec_u8 r3  = unaligned_load(3* src_stride, src);            \
-register vec_u8_t r4  = unaligned_load(4* src_stride, src);            \
+register vec_u8 r4  = unaligned_load(4* src_stride, src);            \
-register vec_u8_t r5  = unaligned_load(5* src_stride, src);            \
+register vec_u8 r5  = unaligned_load(5* src_stride, src);            \
-register vec_u8_t r6  = unaligned_load(6* src_stride, src);            \
+register vec_u8 r6  = unaligned_load(6* src_stride, src);            \
-register vec_u8_t r7  = unaligned_load(7* src_stride, src);            \
+register vec_u8 r7  = unaligned_load(7* src_stride, src);            \
-register vec_u8_t r14 = unaligned_load(14*src_stride, src);            \
+register vec_u8 r14 = unaligned_load(14*src_stride, src);            \
-register vec_u8_t r15 = unaligned_load(15*src_stride, src);            \
+register vec_u8 r15 = unaligned_load(15*src_stride, src);            \
 \
 r8  = unaligned_load( 8*src_stride, src);                              \
 r9  = unaligned_load( 9*src_stride, src);                              \
 r10 = unaligned_load(10*src_stride, src);                              \
 r11 = unaligned_load(11*src_stride, src);                              \
 /* Don't need to compute 14 and 15*/                                   \
 \
 }
 // out: o = |x-y| < a
-static inline vec_u8_t diff_lt_altivec ( register vec_u8_t x,
+static inline vec_u8 diff_lt_altivec ( register vec_u8 x,
-register vec_u8_t y,
+register vec_u8 y,
-register vec_u8_t a) {
+register vec_u8 a) {
-register vec_u8_t diff = vec_subs(x, y);
+register vec_u8 diff = vec_subs(x, y);
-register vec_u8_t diffneg = vec_subs(y, x);
+register vec_u8 diffneg = vec_subs(y, x);
-register vec_u8_t o = vec_or(diff, diffneg); /* |x-y| */
+register vec_u8 o = vec_or(diff, diffneg); /* |x-y| */
-o = (vec_u8_t)vec_cmplt(o, a);
+o = (vec_u8)vec_cmplt(o, a);
 return o;
 }
-static inline vec_u8_t h264_deblock_mask ( register vec_u8_t p0,
+static inline vec_u8 h264_deblock_mask ( register vec_u8 p0,
-register vec_u8_t p1,
+register vec_u8 p1,
-register vec_u8_t q0,
+register vec_u8 q0,
-register vec_u8_t q1,
+register vec_u8 q1,
-register vec_u8_t alpha,
+register vec_u8 alpha,
-register vec_u8_t beta) {
+register vec_u8 beta) {
-register vec_u8_t mask;
+register vec_u8 mask;
-register vec_u8_t tempmask;
+register vec_u8 tempmask;
 mask = diff_lt_altivec(p0, q0, alpha);
 tempmask = diff_lt_altivec(p1, p0, beta);
 mask = vec_and(mask, tempmask);
 tempmask = diff_lt_altivec(q1, q0, beta);
 return mask;
 }
 // out: newp1 = clip((p2 + ((p0 + q0 + 1) >> 1)) >> 1, p1-tc0, p1+tc0)
-static inline vec_u8_t h264_deblock_q1(register vec_u8_t p0,
+static inline vec_u8 h264_deblock_q1(register vec_u8 p0,
-register vec_u8_t p1,
+register vec_u8 p1,
-register vec_u8_t p2,
+register vec_u8 p2,
-register vec_u8_t q0,
+register vec_u8 q0,
-register vec_u8_t tc0) {
+register vec_u8 tc0) {
-register vec_u8_t average = vec_avg(p0, q0);
+register vec_u8 average = vec_avg(p0, q0);
-register vec_u8_t temp;
+register vec_u8 temp;
-register vec_u8_t uncliped;
+register vec_u8 uncliped;
-register vec_u8_t ones;
+register vec_u8 ones;
-register vec_u8_t max;
+register vec_u8 max;
-register vec_u8_t min;
+register vec_u8 min;
-register vec_u8_t newp1;
+register vec_u8 newp1;
 temp = vec_xor(average, p2);
 average = vec_avg(average, p2);     /*avg(p2, avg(p0, q0)) */
 ones = vec_splat_u8(1);
 temp = vec_and(temp, ones);         /*(p2^avg(p0, q0)) & 1 */
 return newp1;
 }
 #define h264_deblock_p0_q0(p0, p1, q0, q1, tc0masked) {                                           \
 \
-const vec_u8_t A0v = vec_sl(vec_splat_u8(10), vec_splat_u8(4));                               \
+const vec_u8 A0v = vec_sl(vec_splat_u8(10), vec_splat_u8(4));                               \
 \
-register vec_u8_t pq0bit = vec_xor(p0,q0);                                                    \
+register vec_u8 pq0bit = vec_xor(p0,q0);                                                    \
-register vec_u8_t q1minus;                                                                    \
+register vec_u8 q1minus;                                                                    \
-register vec_u8_t p0minus;                                                                    \
+register vec_u8 p0minus;                                                                    \
-register vec_u8_t stage1;                                                                     \
+register vec_u8 stage1;                                                                     \
-register vec_u8_t stage2;                                                                     \
+register vec_u8 stage2;                                                                     \
-register vec_u8_t vec160;                                                                     \
+register vec_u8 vec160;                                                                     \
-register vec_u8_t delta;                                                                      \
+register vec_u8 delta;                                                                      \
-register vec_u8_t deltaneg;                                                                   \
+register vec_u8 deltaneg;                                                                   \
 \
 q1minus = vec_nor(q1, q1);                 /* 255 - q1 */                                     \
 stage1 = vec_avg(p1, q1minus);             /* (p1 - q1 + 256)>>1 */                           \
 stage2 = vec_sr(stage1, vec_splat_u8(1));  /* (p1 - q1 + 256)>>2 = 64 + (p1 - q1) >> 2 */     \
 p0minus = vec_nor(p0, p0);                 /* 255 - p0 */                                     \
 q0 = vec_adds(q0, deltaneg);                                                                  \
 }
 #define h264_loop_filter_luma_altivec(p2, p1, p0, q0, q1, q2, alpha, beta, tc0) {            \
 DECLARE_ALIGNED_16(unsigned char, temp[16]);                                             \
-register vec_u8_t alphavec;                                                              \
+register vec_u8 alphavec;                                                              \
-register vec_u8_t betavec;                                                               \
+register vec_u8 betavec;                                                               \
-register vec_u8_t mask;                                                                  \
+register vec_u8 mask;                                                                  \
-register vec_u8_t p1mask;                                                                \
+register vec_u8 p1mask;                                                                \
-register vec_u8_t q1mask;                                                                \
+register vec_u8 q1mask;                                                                \
 register vector signed   char tc0vec;                                                    \
-register vec_u8_t finaltc0;                                                              \
+register vec_u8 finaltc0;                                                              \
-register vec_u8_t tc0masked;                                                             \
+register vec_u8 tc0masked;                                                             \
-register vec_u8_t newp1;                                                                 \
+register vec_u8 newp1;                                                                 \
-register vec_u8_t newq1;                                                                 \
+register vec_u8 newq1;                                                                 \
 \
 temp[0] = alpha;                                                                         \
 temp[1] = beta;                                                                          \
 alphavec = vec_ld(0, temp);                                                              \
 betavec = vec_splat(alphavec, 0x1);                                                      \
 *((int *)temp) = *((int *)tc0);                                                          \
 tc0vec = vec_ld(0, (signed char*)temp);                                                  \
 tc0vec = vec_mergeh(tc0vec, tc0vec);                                                     \
 tc0vec = vec_mergeh(tc0vec, tc0vec);                                                     \
 mask = vec_and(mask, vec_cmpgt(tc0vec, vec_splat_s8(-1)));  /* if tc0[i] >= 0 */         \
-finaltc0 = vec_and((vec_u8_t)tc0vec, mask);     /* tc = tc0 */                           \
+finaltc0 = vec_and((vec_u8)tc0vec, mask);     /* tc = tc0 */                           \
 \
 p1mask = diff_lt_altivec(p2, p0, betavec);                                               \
 p1mask = vec_and(p1mask, mask);                             /* if ( |p2 - p0| < beta) */ \
-tc0masked = vec_and(p1mask, (vec_u8_t)tc0vec);                                           \
+tc0masked = vec_and(p1mask, (vec_u8)tc0vec);                                           \
 finaltc0 = vec_sub(finaltc0, p1mask);                       /* tc++ */                   \
 newp1 = h264_deblock_q1(p0, p1, p2, q0, tc0masked);                                      \
 /*end if*/                                                                               \
 \
 q1mask = diff_lt_altivec(q2, q0, betavec);                                               \
 q1mask = vec_and(q1mask, mask);                             /* if ( |q2 - q0| < beta ) */\
-tc0masked = vec_and(q1mask, (vec_u8_t)tc0vec);                                           \
+tc0masked = vec_and(q1mask, (vec_u8)tc0vec);                                           \
 finaltc0 = vec_sub(finaltc0, q1mask);                       /* tc++ */                   \
 newq1 = h264_deblock_q1(p0, q1, q2, q0, tc0masked);                                      \
 /*end if*/                                                                               \
 \
 h264_deblock_p0_q0(p0, p1, q0, q1, finaltc0);                                            \
 }
 static void h264_v_loop_filter_luma_altivec(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) {
 if ((tc0[0] & tc0[1] & tc0[2] & tc0[3]) >= 0) {
-register vec_u8_t p2 = vec_ld(-3*stride, pix);
+register vec_u8 p2 = vec_ld(-3*stride, pix);
-register vec_u8_t p1 = vec_ld(-2*stride, pix);
+register vec_u8 p1 = vec_ld(-2*stride, pix);
-register vec_u8_t p0 = vec_ld(-1*stride, pix);
+register vec_u8 p0 = vec_ld(-1*stride, pix);
-register vec_u8_t q0 = vec_ld(0, pix);
+register vec_u8 q0 = vec_ld(0, pix);
-register vec_u8_t q1 = vec_ld(stride, pix);
+register vec_u8 q1 = vec_ld(stride, pix);
-register vec_u8_t q2 = vec_ld(2*stride, pix);
+register vec_u8 q2 = vec_ld(2*stride, pix);
 h264_loop_filter_luma_altivec(p2, p1, p0, q0, q1, q2, alpha, beta, tc0);
 vec_st(p1, -2*stride, pix);
 vec_st(p0, -1*stride, pix);
 vec_st(q0, 0, pix);
 vec_st(q1, stride, pix);
 }
 }
 static void h264_h_loop_filter_luma_altivec(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) {
-register vec_u8_t line0, line1, line2, line3, line4, line5;
+register vec_u8 line0, line1, line2, line3, line4, line5;
 if ((tc0[0] & tc0[1] & tc0[2] & tc0[3]) < 0)
 return;
 readAndTranspose16x6(pix-3, stride, line0, line1, line2, line3, line4, line5);
 h264_loop_filter_luma_altivec(line0, line1, line2, line3, line4, line5, alpha, beta, tc0);
 transpose4x16(line1, line2, line3, line4);

Mercurial > libavcodec.hg

comparison ppc/h264_altivec.c @ 8494:1615d6b75ada libavcodec