void x264_sub4x4_dct_altivec( int16_t dct[16], uint8_t *pix1, uint8_t *pix2 )
{
PREP_DIFF_8BYTEALIGNED;
vec_s16_t dct0v, dct1v, dct2v, dct3v;
vec_s16_t tmp0v, tmp1v, tmp2v, tmp3v;
vec_u8_t permHighv;
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 4, dct0v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 4, dct1v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 4, dct2v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 4, dct3v );
VEC_DCT( dct0v, dct1v, dct2v, dct3v, tmp0v, tmp1v, tmp2v, tmp3v );
VEC_TRANSPOSE_4( tmp0v, tmp1v, tmp2v, tmp3v,
dct0v, dct1v, dct2v, dct3v );
permHighv = (vec_u8_t) CV(0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17);
VEC_DCT( dct0v, dct1v, dct2v, dct3v, tmp0v, tmp1v, tmp2v, tmp3v );
vec_st(vec_perm(tmp0v, tmp1v, permHighv), 0, dct);
vec_st(vec_perm(tmp2v, tmp3v, permHighv), 16, dct);
}
void x264_sub8x8_dct8_altivec( int16_t dct[64], uint8_t *pix1, uint8_t *pix2 )
{
vec_u16_t onev = vec_splat_u16(1);
vec_u16_t twov = vec_add( onev, onev );
PREP_DIFF_8BYTEALIGNED;
vec_s16_t dct0v, dct1v, dct2v, dct3v,
dct4v, dct5v, dct6v, dct7v;
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct0v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct1v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct2v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct3v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct4v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct5v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct6v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct7v );
DCT8_1D_ALTIVEC( dct0v, dct1v, dct2v, dct3v,
dct4v, dct5v, dct6v, dct7v );
vec_s16_t dct_tr0v, dct_tr1v, dct_tr2v, dct_tr3v,
dct_tr4v, dct_tr5v, dct_tr6v, dct_tr7v;
VEC_TRANSPOSE_8(dct0v, dct1v, dct2v, dct3v,
dct4v, dct5v, dct6v, dct7v,
dct_tr0v, dct_tr1v, dct_tr2v, dct_tr3v,
dct_tr4v, dct_tr5v, dct_tr6v, dct_tr7v );
DCT8_1D_ALTIVEC( dct_tr0v, dct_tr1v, dct_tr2v, dct_tr3v,
dct_tr4v, dct_tr5v, dct_tr6v, dct_tr7v );
vec_st( dct_tr0v, 0, dct );
vec_st( dct_tr1v, 16, dct );
vec_st( dct_tr2v, 32, dct );
vec_st( dct_tr3v, 48, dct );
vec_st( dct_tr4v, 64, dct );
vec_st( dct_tr5v, 80, dct );
vec_st( dct_tr6v, 96, dct );
vec_st( dct_tr7v, 112, dct );
}
void x264_sub8x8_dct_altivec( int16_t dct[4][4][4],
uint8_t *pix1, uint8_t *pix2 )
{
PREP_DIFF_8BYTEALIGNED;
vec_s16_t dct0v, dct1v, dct2v, dct3v, dct4v, dct5v, dct6v, dct7v;
vec_s16_t tmp0v, tmp1v, tmp2v, tmp3v, tmp4v, tmp5v, tmp6v, tmp7v;
vec_u8_t permHighv, permLowv;
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct0v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct1v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct2v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct3v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct4v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct5v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct6v );
VEC_DIFF_H_8BYTE_ALIGNED( pix1, FENC_STRIDE, pix2, FDEC_STRIDE, 8, dct7v );
VEC_DCT( dct0v, dct1v, dct2v, dct3v, tmp0v, tmp1v, tmp2v, tmp3v );
VEC_DCT( dct4v, dct5v, dct6v, dct7v, tmp4v, tmp5v, tmp6v, tmp7v );
VEC_TRANSPOSE_8( tmp0v, tmp1v, tmp2v, tmp3v,
tmp4v, tmp5v, tmp6v, tmp7v,
dct0v, dct1v, dct2v, dct3v,
dct4v, dct5v, dct6v, dct7v );
permHighv = (vec_u8_t) CV(0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17);
permLowv = (vec_u8_t) CV(0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F);
VEC_DCT( dct0v, dct1v, dct2v, dct3v, tmp0v, tmp1v, tmp2v, tmp3v );
VEC_DCT( dct4v, dct5v, dct6v, dct7v, tmp4v, tmp5v, tmp6v, tmp7v );
vec_st(vec_perm(tmp0v, tmp1v, permHighv), 0, (int16_t*)dct);
vec_st(vec_perm(tmp2v, tmp3v, permHighv), 16, (int16_t*)dct);
vec_st(vec_perm(tmp4v, tmp5v, permHighv), 32, (int16_t*)dct);
vec_st(vec_perm(tmp6v, tmp7v, permHighv), 48, (int16_t*)dct);
vec_st(vec_perm(tmp0v, tmp1v, permLowv), 64, (int16_t*)dct);
vec_st(vec_perm(tmp2v, tmp3v, permLowv), 80, (int16_t*)dct);
vec_st(vec_perm(tmp4v, tmp5v, permLowv), 96, (int16_t*)dct);
vec_st(vec_perm(tmp6v, tmp7v, permLowv), 112, (int16_t*)dct);
}
