cosmetics: Replace // by /* */ comments.

sync with upstream libmpeg2 0.4.1

Originally committed as revision 11915 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Diego Biurrun 2008-02-13 01:18:12 +00:00
parent 6c6007d6a3
commit 5edac5dc94

View File

@ -85,102 +85,102 @@ static inline void idct_row (int16_t * row, int offset,
static inline void mmxext_row_head (int16_t * row, int offset, const int16_t * table)
{
movq_m2r (*(row+offset), mm2); // mm2 = x6 x4 x2 x0
movq_m2r (*(row+offset), mm2); /* mm2 = x6 x4 x2 x0 */
movq_m2r (*(row+offset+4), mm5); // mm5 = x7 x5 x3 x1
movq_r2r (mm2, mm0); // mm0 = x6 x4 x2 x0
movq_m2r (*(row+offset+4), mm5); /* mm5 = x7 x5 x3 x1 */
movq_r2r (mm2, mm0); /* mm0 = x6 x4 x2 x0 */
movq_m2r (*table, mm3); // mm3 = -C2 -C4 C2 C4
movq_r2r (mm5, mm6); // mm6 = x7 x5 x3 x1
movq_m2r (*table, mm3); /* mm3 = -C2 -C4 C2 C4 */
movq_r2r (mm5, mm6); /* mm6 = x7 x5 x3 x1 */
movq_m2r (*(table+4), mm4); // mm4 = C6 C4 C6 C4
pmaddwd_r2r (mm0, mm3); // mm3 = -C4*x4-C2*x6 C4*x0+C2*x2
movq_m2r (*(table+4), mm4); /* mm4 = C6 C4 C6 C4 */
pmaddwd_r2r (mm0, mm3); /* mm3 = -C4*x4-C2*x6 C4*x0+C2*x2 */
pshufw_r2r (mm2, mm2, 0x4e); // mm2 = x2 x0 x6 x4
pshufw_r2r (mm2, mm2, 0x4e); /* mm2 = x2 x0 x6 x4 */
}
static inline void mmxext_row (const int16_t * table, const int32_t * rounder)
{
movq_m2r (*(table+8), mm1); // mm1 = -C5 -C1 C3 C1
pmaddwd_r2r (mm2, mm4); // mm4 = C4*x0+C6*x2 C4*x4+C6*x6
movq_m2r (*(table+8), mm1); /* mm1 = -C5 -C1 C3 C1 */
pmaddwd_r2r (mm2, mm4); /* mm4 = C4*x0+C6*x2 C4*x4+C6*x6 */
pmaddwd_m2r (*(table+16), mm0); // mm0 = C4*x4-C6*x6 C4*x0-C6*x2
pshufw_r2r (mm6, mm6, 0x4e); // mm6 = x3 x1 x7 x5
pmaddwd_m2r (*(table+16), mm0); /* mm0 = C4*x4-C6*x6 C4*x0-C6*x2 */
pshufw_r2r (mm6, mm6, 0x4e); /* mm6 = x3 x1 x7 x5 */
movq_m2r (*(table+12), mm7); // mm7 = -C7 C3 C7 C5
pmaddwd_r2r (mm5, mm1); // mm1 = -C1*x5-C5*x7 C1*x1+C3*x3
movq_m2r (*(table+12), mm7); /* mm7 = -C7 C3 C7 C5 */
pmaddwd_r2r (mm5, mm1); /* mm1 = -C1*x5-C5*x7 C1*x1+C3*x3 */
paddd_m2r (*rounder, mm3); // mm3 += rounder
pmaddwd_r2r (mm6, mm7); // mm7 = C3*x1-C7*x3 C5*x5+C7*x7
paddd_m2r (*rounder, mm3); /* mm3 += rounder */
pmaddwd_r2r (mm6, mm7); /* mm7 = C3*x1-C7*x3 C5*x5+C7*x7 */
pmaddwd_m2r (*(table+20), mm2); // mm2 = C4*x0-C2*x2 -C4*x4+C2*x6
paddd_r2r (mm4, mm3); // mm3 = a1 a0 + rounder
pmaddwd_m2r (*(table+20), mm2); /* mm2 = C4*x0-C2*x2 -C4*x4+C2*x6 */
paddd_r2r (mm4, mm3); /* mm3 = a1 a0 + rounder */
pmaddwd_m2r (*(table+24), mm5); // mm5 = C3*x5-C1*x7 C5*x1-C1*x3
movq_r2r (mm3, mm4); // mm4 = a1 a0 + rounder
pmaddwd_m2r (*(table+24), mm5); /* mm5 = C3*x5-C1*x7 C5*x1-C1*x3 */
movq_r2r (mm3, mm4); /* mm4 = a1 a0 + rounder */
pmaddwd_m2r (*(table+28), mm6); // mm6 = C7*x1-C5*x3 C7*x5+C3*x7
paddd_r2r (mm7, mm1); // mm1 = b1 b0
pmaddwd_m2r (*(table+28), mm6); /* mm6 = C7*x1-C5*x3 C7*x5+C3*x7 */
paddd_r2r (mm7, mm1); /* mm1 = b1 b0 */
paddd_m2r (*rounder, mm0); // mm0 += rounder
psubd_r2r (mm1, mm3); // mm3 = a1-b1 a0-b0 + rounder
paddd_m2r (*rounder, mm0); /* mm0 += rounder */
psubd_r2r (mm1, mm3); /* mm3 = a1-b1 a0-b0 + rounder */
psrad_i2r (ROW_SHIFT, mm3); // mm3 = y6 y7
paddd_r2r (mm4, mm1); // mm1 = a1+b1 a0+b0 + rounder
psrad_i2r (ROW_SHIFT, mm3); /* mm3 = y6 y7 */
paddd_r2r (mm4, mm1); /* mm1 = a1+b1 a0+b0 + rounder */
paddd_r2r (mm2, mm0); // mm0 = a3 a2 + rounder
psrad_i2r (ROW_SHIFT, mm1); // mm1 = y1 y0
paddd_r2r (mm2, mm0); /* mm0 = a3 a2 + rounder */
psrad_i2r (ROW_SHIFT, mm1); /* mm1 = y1 y0 */
paddd_r2r (mm6, mm5); // mm5 = b3 b2
movq_r2r (mm0, mm4); // mm4 = a3 a2 + rounder
paddd_r2r (mm6, mm5); /* mm5 = b3 b2 */
movq_r2r (mm0, mm4); /* mm4 = a3 a2 + rounder */
paddd_r2r (mm5, mm0); // mm0 = a3+b3 a2+b2 + rounder
psubd_r2r (mm5, mm4); // mm4 = a3-b3 a2-b2 + rounder
paddd_r2r (mm5, mm0); /* mm0 = a3+b3 a2+b2 + rounder */
psubd_r2r (mm5, mm4); /* mm4 = a3-b3 a2-b2 + rounder */
}
static inline void mmxext_row_tail (int16_t * row, int store)
{
psrad_i2r (ROW_SHIFT, mm0); // mm0 = y3 y2
psrad_i2r (ROW_SHIFT, mm0); /* mm0 = y3 y2 */
psrad_i2r (ROW_SHIFT, mm4); // mm4 = y4 y5
psrad_i2r (ROW_SHIFT, mm4); /* mm4 = y4 y5 */
packssdw_r2r (mm0, mm1); // mm1 = y3 y2 y1 y0
packssdw_r2r (mm0, mm1); /* mm1 = y3 y2 y1 y0 */
packssdw_r2r (mm3, mm4); // mm4 = y6 y7 y4 y5
packssdw_r2r (mm3, mm4); /* mm4 = y6 y7 y4 y5 */
movq_r2m (mm1, *(row+store)); // save y3 y2 y1 y0
pshufw_r2r (mm4, mm4, 0xb1); // mm4 = y7 y6 y5 y4
movq_r2m (mm1, *(row+store)); /* save y3 y2 y1 y0 */
pshufw_r2r (mm4, mm4, 0xb1); /* mm4 = y7 y6 y5 y4 */
/* slot */
movq_r2m (mm4, *(row+store+4)); // save y7 y6 y5 y4
movq_r2m (mm4, *(row+store+4)); /* save y7 y6 y5 y4 */
}
static inline void mmxext_row_mid (int16_t * row, int store,
int offset, const int16_t * table)
{
movq_m2r (*(row+offset), mm2); // mm2 = x6 x4 x2 x0
psrad_i2r (ROW_SHIFT, mm0); // mm0 = y3 y2
movq_m2r (*(row+offset), mm2); /* mm2 = x6 x4 x2 x0 */
psrad_i2r (ROW_SHIFT, mm0); /* mm0 = y3 y2 */
movq_m2r (*(row+offset+4), mm5); // mm5 = x7 x5 x3 x1
psrad_i2r (ROW_SHIFT, mm4); // mm4 = y4 y5
movq_m2r (*(row+offset+4), mm5); /* mm5 = x7 x5 x3 x1 */
psrad_i2r (ROW_SHIFT, mm4); /* mm4 = y4 y5 */
packssdw_r2r (mm0, mm1); // mm1 = y3 y2 y1 y0
movq_r2r (mm5, mm6); // mm6 = x7 x5 x3 x1
packssdw_r2r (mm0, mm1); /* mm1 = y3 y2 y1 y0 */
movq_r2r (mm5, mm6); /* mm6 = x7 x5 x3 x1 */
packssdw_r2r (mm3, mm4); // mm4 = y6 y7 y4 y5
movq_r2r (mm2, mm0); // mm0 = x6 x4 x2 x0
packssdw_r2r (mm3, mm4); /* mm4 = y6 y7 y4 y5 */
movq_r2r (mm2, mm0); /* mm0 = x6 x4 x2 x0 */
movq_r2m (mm1, *(row+store)); // save y3 y2 y1 y0
pshufw_r2r (mm4, mm4, 0xb1); // mm4 = y7 y6 y5 y4
movq_r2m (mm1, *(row+store)); /* save y3 y2 y1 y0 */
pshufw_r2r (mm4, mm4, 0xb1); /* mm4 = y7 y6 y5 y4 */
movq_m2r (*table, mm3); // mm3 = -C2 -C4 C2 C4
movq_r2m (mm4, *(row+store+4)); // save y7 y6 y5 y4
movq_m2r (*table, mm3); /* mm3 = -C2 -C4 C2 C4 */
movq_r2m (mm4, *(row+store+4)); /* save y7 y6 y5 y4 */
pmaddwd_r2r (mm0, mm3); // mm3 = -C4*x4-C2*x6 C4*x0+C2*x2
pmaddwd_r2r (mm0, mm3); /* mm3 = -C4*x4-C2*x6 C4*x0+C2*x2 */
movq_m2r (*(table+4), mm4); // mm4 = C6 C4 C6 C4
pshufw_r2r (mm2, mm2, 0x4e); // mm2 = x2 x0 x6 x4
movq_m2r (*(table+4), mm4); /* mm4 = C6 C4 C6 C4 */
pshufw_r2r (mm2, mm2, 0x4e); /* mm2 = x2 x0 x6 x4 */
}
@ -197,123 +197,123 @@ static inline void mmxext_row_mid (int16_t * row, int store,
static inline void mmx_row_head (int16_t * row, int offset, const int16_t * table)
{
movq_m2r (*(row+offset), mm2); // mm2 = x6 x4 x2 x0
movq_m2r (*(row+offset), mm2); /* mm2 = x6 x4 x2 x0 */
movq_m2r (*(row+offset+4), mm5); // mm5 = x7 x5 x3 x1
movq_r2r (mm2, mm0); // mm0 = x6 x4 x2 x0
movq_m2r (*(row+offset+4), mm5); /* mm5 = x7 x5 x3 x1 */
movq_r2r (mm2, mm0); /* mm0 = x6 x4 x2 x0 */
movq_m2r (*table, mm3); // mm3 = C6 C4 C2 C4
movq_r2r (mm5, mm6); // mm6 = x7 x5 x3 x1
movq_m2r (*table, mm3); /* mm3 = C6 C4 C2 C4 */
movq_r2r (mm5, mm6); /* mm6 = x7 x5 x3 x1 */
punpckldq_r2r (mm0, mm0); // mm0 = x2 x0 x2 x0
punpckldq_r2r (mm0, mm0); /* mm0 = x2 x0 x2 x0 */
movq_m2r (*(table+4), mm4); // mm4 = -C2 -C4 C6 C4
pmaddwd_r2r (mm0, mm3); // mm3 = C4*x0+C6*x2 C4*x0+C2*x2
movq_m2r (*(table+4), mm4); /* mm4 = -C2 -C4 C6 C4 */
pmaddwd_r2r (mm0, mm3); /* mm3 = C4*x0+C6*x2 C4*x0+C2*x2 */
movq_m2r (*(table+8), mm1); // mm1 = -C7 C3 C3 C1
punpckhdq_r2r (mm2, mm2); // mm2 = x6 x4 x6 x4
movq_m2r (*(table+8), mm1); /* mm1 = -C7 C3 C3 C1 */
punpckhdq_r2r (mm2, mm2); /* mm2 = x6 x4 x6 x4 */
}
static inline void mmx_row (const int16_t * table, const int32_t * rounder)
{
pmaddwd_r2r (mm2, mm4); // mm4 = -C4*x4-C2*x6 C4*x4+C6*x6
punpckldq_r2r (mm5, mm5); // mm5 = x3 x1 x3 x1
pmaddwd_r2r (mm2, mm4); /* mm4 = -C4*x4-C2*x6 C4*x4+C6*x6 */
punpckldq_r2r (mm5, mm5); /* mm5 = x3 x1 x3 x1 */
pmaddwd_m2r (*(table+16), mm0); // mm0 = C4*x0-C2*x2 C4*x0-C6*x2
punpckhdq_r2r (mm6, mm6); // mm6 = x7 x5 x7 x5
pmaddwd_m2r (*(table+16), mm0); /* mm0 = C4*x0-C2*x2 C4*x0-C6*x2 */
punpckhdq_r2r (mm6, mm6); /* mm6 = x7 x5 x7 x5 */
movq_m2r (*(table+12), mm7); // mm7 = -C5 -C1 C7 C5
pmaddwd_r2r (mm5, mm1); // mm1 = C3*x1-C7*x3 C1*x1+C3*x3
movq_m2r (*(table+12), mm7); /* mm7 = -C5 -C1 C7 C5 */
pmaddwd_r2r (mm5, mm1); /* mm1 = C3*x1-C7*x3 C1*x1+C3*x3 */
paddd_m2r (*rounder, mm3); // mm3 += rounder
pmaddwd_r2r (mm6, mm7); // mm7 = -C1*x5-C5*x7 C5*x5+C7*x7
paddd_m2r (*rounder, mm3); /* mm3 += rounder */
pmaddwd_r2r (mm6, mm7); /* mm7 = -C1*x5-C5*x7 C5*x5+C7*x7 */
pmaddwd_m2r (*(table+20), mm2); // mm2 = C4*x4-C6*x6 -C4*x4+C2*x6
paddd_r2r (mm4, mm3); // mm3 = a1 a0 + rounder
pmaddwd_m2r (*(table+20), mm2); /* mm2 = C4*x4-C6*x6 -C4*x4+C2*x6 */
paddd_r2r (mm4, mm3); /* mm3 = a1 a0 + rounder */
pmaddwd_m2r (*(table+24), mm5); // mm5 = C7*x1-C5*x3 C5*x1-C1*x3
movq_r2r (mm3, mm4); // mm4 = a1 a0 + rounder
pmaddwd_m2r (*(table+24), mm5); /* mm5 = C7*x1-C5*x3 C5*x1-C1*x3 */
movq_r2r (mm3, mm4); /* mm4 = a1 a0 + rounder */
pmaddwd_m2r (*(table+28), mm6); // mm6 = C3*x5-C1*x7 C7*x5+C3*x7
paddd_r2r (mm7, mm1); // mm1 = b1 b0
pmaddwd_m2r (*(table+28), mm6); /* mm6 = C3*x5-C1*x7 C7*x5+C3*x7 */
paddd_r2r (mm7, mm1); /* mm1 = b1 b0 */
paddd_m2r (*rounder, mm0); // mm0 += rounder
psubd_r2r (mm1, mm3); // mm3 = a1-b1 a0-b0 + rounder
paddd_m2r (*rounder, mm0); /* mm0 += rounder */
psubd_r2r (mm1, mm3); /* mm3 = a1-b1 a0-b0 + rounder */
psrad_i2r (ROW_SHIFT, mm3); // mm3 = y6 y7
paddd_r2r (mm4, mm1); // mm1 = a1+b1 a0+b0 + rounder
psrad_i2r (ROW_SHIFT, mm3); /* mm3 = y6 y7 */
paddd_r2r (mm4, mm1); /* mm1 = a1+b1 a0+b0 + rounder */
paddd_r2r (mm2, mm0); // mm0 = a3 a2 + rounder
psrad_i2r (ROW_SHIFT, mm1); // mm1 = y1 y0
paddd_r2r (mm2, mm0); /* mm0 = a3 a2 + rounder */
psrad_i2r (ROW_SHIFT, mm1); /* mm1 = y1 y0 */
paddd_r2r (mm6, mm5); // mm5 = b3 b2
movq_r2r (mm0, mm7); // mm7 = a3 a2 + rounder
paddd_r2r (mm6, mm5); /* mm5 = b3 b2 */
movq_r2r (mm0, mm7); /* mm7 = a3 a2 + rounder */
paddd_r2r (mm5, mm0); // mm0 = a3+b3 a2+b2 + rounder
psubd_r2r (mm5, mm7); // mm7 = a3-b3 a2-b2 + rounder
paddd_r2r (mm5, mm0); /* mm0 = a3+b3 a2+b2 + rounder */
psubd_r2r (mm5, mm7); /* mm7 = a3-b3 a2-b2 + rounder */
}
static inline void mmx_row_tail (int16_t * row, int store)
{
psrad_i2r (ROW_SHIFT, mm0); // mm0 = y3 y2
psrad_i2r (ROW_SHIFT, mm0); /* mm0 = y3 y2 */
psrad_i2r (ROW_SHIFT, mm7); // mm7 = y4 y5
psrad_i2r (ROW_SHIFT, mm7); /* mm7 = y4 y5 */
packssdw_r2r (mm0, mm1); // mm1 = y3 y2 y1 y0
packssdw_r2r (mm0, mm1); /* mm1 = y3 y2 y1 y0 */
packssdw_r2r (mm3, mm7); // mm7 = y6 y7 y4 y5
packssdw_r2r (mm3, mm7); /* mm7 = y6 y7 y4 y5 */
movq_r2m (mm1, *(row+store)); // save y3 y2 y1 y0
movq_r2r (mm7, mm4); // mm4 = y6 y7 y4 y5
movq_r2m (mm1, *(row+store)); /* save y3 y2 y1 y0 */
movq_r2r (mm7, mm4); /* mm4 = y6 y7 y4 y5 */
pslld_i2r (16, mm7); // mm7 = y7 0 y5 0
pslld_i2r (16, mm7); /* mm7 = y7 0 y5 0 */
psrld_i2r (16, mm4); // mm4 = 0 y6 0 y4
psrld_i2r (16, mm4); /* mm4 = 0 y6 0 y4 */
por_r2r (mm4, mm7); // mm7 = y7 y6 y5 y4
por_r2r (mm4, mm7); /* mm7 = y7 y6 y5 y4 */
/* slot */
movq_r2m (mm7, *(row+store+4)); // save y7 y6 y5 y4
movq_r2m (mm7, *(row+store+4)); /* save y7 y6 y5 y4 */
}
static inline void mmx_row_mid (int16_t * row, int store,
int offset, const int16_t * table)
{
movq_m2r (*(row+offset), mm2); // mm2 = x6 x4 x2 x0
psrad_i2r (ROW_SHIFT, mm0); // mm0 = y3 y2
movq_m2r (*(row+offset), mm2); /* mm2 = x6 x4 x2 x0 */
psrad_i2r (ROW_SHIFT, mm0); /* mm0 = y3 y2 */
movq_m2r (*(row+offset+4), mm5); // mm5 = x7 x5 x3 x1
psrad_i2r (ROW_SHIFT, mm7); // mm7 = y4 y5
movq_m2r (*(row+offset+4), mm5); /* mm5 = x7 x5 x3 x1 */
psrad_i2r (ROW_SHIFT, mm7); /* mm7 = y4 y5 */
packssdw_r2r (mm0, mm1); // mm1 = y3 y2 y1 y0
movq_r2r (mm5, mm6); // mm6 = x7 x5 x3 x1
packssdw_r2r (mm0, mm1); /* mm1 = y3 y2 y1 y0 */
movq_r2r (mm5, mm6); /* mm6 = x7 x5 x3 x1 */
packssdw_r2r (mm3, mm7); // mm7 = y6 y7 y4 y5
movq_r2r (mm2, mm0); // mm0 = x6 x4 x2 x0
packssdw_r2r (mm3, mm7); /* mm7 = y6 y7 y4 y5 */
movq_r2r (mm2, mm0); /* mm0 = x6 x4 x2 x0 */
movq_r2m (mm1, *(row+store)); // save y3 y2 y1 y0
movq_r2r (mm7, mm1); // mm1 = y6 y7 y4 y5
movq_r2m (mm1, *(row+store)); /* save y3 y2 y1 y0 */
movq_r2r (mm7, mm1); /* mm1 = y6 y7 y4 y5 */
punpckldq_r2r (mm0, mm0); // mm0 = x2 x0 x2 x0
psrld_i2r (16, mm7); // mm7 = 0 y6 0 y4
punpckldq_r2r (mm0, mm0); /* mm0 = x2 x0 x2 x0 */
psrld_i2r (16, mm7); /* mm7 = 0 y6 0 y4 */
movq_m2r (*table, mm3); // mm3 = C6 C4 C2 C4
pslld_i2r (16, mm1); // mm1 = y7 0 y5 0
movq_m2r (*table, mm3); /* mm3 = C6 C4 C2 C4 */
pslld_i2r (16, mm1); /* mm1 = y7 0 y5 0 */
movq_m2r (*(table+4), mm4); // mm4 = -C2 -C4 C6 C4
por_r2r (mm1, mm7); // mm7 = y7 y6 y5 y4
movq_m2r (*(table+4), mm4); /* mm4 = -C2 -C4 C6 C4 */
por_r2r (mm1, mm7); /* mm7 = y7 y6 y5 y4 */
movq_m2r (*(table+8), mm1); // mm1 = -C7 C3 C3 C1
punpckhdq_r2r (mm2, mm2); // mm2 = x6 x4 x6 x4
movq_m2r (*(table+8), mm1); /* mm1 = -C7 C3 C3 C1 */
punpckhdq_r2r (mm2, mm2); /* mm2 = x6 x4 x6 x4 */
movq_r2m (mm7, *(row+store+4)); // save y7 y6 y5 y4
pmaddwd_r2r (mm0, mm3); // mm3 = C4*x0+C6*x2 C4*x0+C2*x2
movq_r2m (mm7, *(row+store+4)); /* save y7 y6 y5 y4 */
pmaddwd_r2r (mm0, mm3); /* mm3 = C4*x0+C6*x2 C4*x0+C2*x2 */
}
#if 0
// C column IDCT - it is just here to document the MMXEXT and MMX versions
/* C column IDCT - it is just here to document the MMXEXT and MMX versions */
static inline void idct_col (int16_t * col, int offset)
{
/* multiplication - as implemented on mmx */
@ -384,7 +384,7 @@ static inline void idct_col (int16_t * col, int offset)
#endif
// MMX column IDCT
/* MMX column IDCT */
static inline void idct_col (int16_t * col, int offset)
{
#define T1 13036
@ -400,132 +400,132 @@ static inline void idct_col (int16_t * col, int offset)
/* column code adapted from Peter Gubanov */
/* http://www.elecard.com/peter/idct.shtml */
movq_m2r (*t1_vector, mm0); // mm0 = T1
movq_m2r (*t1_vector, mm0); /* mm0 = T1 */
movq_m2r (*(col+offset+1*8), mm1); // mm1 = x1
movq_r2r (mm0, mm2); // mm2 = T1
movq_m2r (*(col+offset+1*8), mm1); /* mm1 = x1 */
movq_r2r (mm0, mm2); /* mm2 = T1 */
movq_m2r (*(col+offset+7*8), mm4); // mm4 = x7
pmulhw_r2r (mm1, mm0); // mm0 = T1*x1
movq_m2r (*(col+offset+7*8), mm4); /* mm4 = x7 */
pmulhw_r2r (mm1, mm0); /* mm0 = T1*x1 */
movq_m2r (*t3_vector, mm5); // mm5 = T3
pmulhw_r2r (mm4, mm2); // mm2 = T1*x7
movq_m2r (*t3_vector, mm5); /* mm5 = T3 */
pmulhw_r2r (mm4, mm2); /* mm2 = T1*x7 */
movq_m2r (*(col+offset+5*8), mm6); // mm6 = x5
movq_r2r (mm5, mm7); // mm7 = T3-1
movq_m2r (*(col+offset+5*8), mm6); /* mm6 = x5 */
movq_r2r (mm5, mm7); /* mm7 = T3-1 */
movq_m2r (*(col+offset+3*8), mm3); // mm3 = x3
psubsw_r2r (mm4, mm0); // mm0 = v17
movq_m2r (*(col+offset+3*8), mm3); /* mm3 = x3 */
psubsw_r2r (mm4, mm0); /* mm0 = v17 */
movq_m2r (*t2_vector, mm4); // mm4 = T2
pmulhw_r2r (mm3, mm5); // mm5 = (T3-1)*x3
movq_m2r (*t2_vector, mm4); /* mm4 = T2 */
pmulhw_r2r (mm3, mm5); /* mm5 = (T3-1)*x3 */
paddsw_r2r (mm2, mm1); // mm1 = u17
pmulhw_r2r (mm6, mm7); // mm7 = (T3-1)*x5
paddsw_r2r (mm2, mm1); /* mm1 = u17 */
pmulhw_r2r (mm6, mm7); /* mm7 = (T3-1)*x5 */
/* slot */
movq_r2r (mm4, mm2); // mm2 = T2
paddsw_r2r (mm3, mm5); // mm5 = T3*x3
movq_r2r (mm4, mm2); /* mm2 = T2 */
paddsw_r2r (mm3, mm5); /* mm5 = T3*x3 */
pmulhw_m2r (*(col+offset+2*8), mm4);// mm4 = T2*x2
paddsw_r2r (mm6, mm7); // mm7 = T3*x5
pmulhw_m2r (*(col+offset+2*8), mm4);/* mm4 = T2*x2 */
paddsw_r2r (mm6, mm7); /* mm7 = T3*x5 */
psubsw_r2r (mm6, mm5); // mm5 = v35
paddsw_r2r (mm3, mm7); // mm7 = u35
psubsw_r2r (mm6, mm5); /* mm5 = v35 */
paddsw_r2r (mm3, mm7); /* mm7 = u35 */
movq_m2r (*(col+offset+6*8), mm3); // mm3 = x6
movq_r2r (mm0, mm6); // mm6 = v17
movq_m2r (*(col+offset+6*8), mm3); /* mm3 = x6 */
movq_r2r (mm0, mm6); /* mm6 = v17 */
pmulhw_r2r (mm3, mm2); // mm2 = T2*x6
psubsw_r2r (mm5, mm0); // mm0 = b3
pmulhw_r2r (mm3, mm2); /* mm2 = T2*x6 */
psubsw_r2r (mm5, mm0); /* mm0 = b3 */
psubsw_r2r (mm3, mm4); // mm4 = v26
paddsw_r2r (mm6, mm5); // mm5 = v12
psubsw_r2r (mm3, mm4); /* mm4 = v26 */
paddsw_r2r (mm6, mm5); /* mm5 = v12 */
movq_r2m (mm0, *(col+offset+3*8)); // save b3 in scratch0
movq_r2r (mm1, mm6); // mm6 = u17
movq_r2m (mm0, *(col+offset+3*8)); /* save b3 in scratch0 */
movq_r2r (mm1, mm6); /* mm6 = u17 */
paddsw_m2r (*(col+offset+2*8), mm2);// mm2 = u26
paddsw_r2r (mm7, mm6); // mm6 = b0
paddsw_m2r (*(col+offset+2*8), mm2);/* mm2 = u26 */
paddsw_r2r (mm7, mm6); /* mm6 = b0 */
psubsw_r2r (mm7, mm1); // mm1 = u12
movq_r2r (mm1, mm7); // mm7 = u12
psubsw_r2r (mm7, mm1); /* mm1 = u12 */
movq_r2r (mm1, mm7); /* mm7 = u12 */
movq_m2r (*(col+offset+0*8), mm3); // mm3 = x0
paddsw_r2r (mm5, mm1); // mm1 = u12+v12
movq_m2r (*(col+offset+0*8), mm3); /* mm3 = x0 */
paddsw_r2r (mm5, mm1); /* mm1 = u12+v12 */
movq_m2r (*c4_vector, mm0); // mm0 = C4/2
psubsw_r2r (mm5, mm7); // mm7 = u12-v12
movq_m2r (*c4_vector, mm0); /* mm0 = C4/2 */
psubsw_r2r (mm5, mm7); /* mm7 = u12-v12 */
movq_r2m (mm6, *(col+offset+5*8)); // save b0 in scratch1
pmulhw_r2r (mm0, mm1); // mm1 = b1/2
movq_r2m (mm6, *(col+offset+5*8)); /* save b0 in scratch1 */
pmulhw_r2r (mm0, mm1); /* mm1 = b1/2 */
movq_r2r (mm4, mm6); // mm6 = v26
pmulhw_r2r (mm0, mm7); // mm7 = b2/2
movq_r2r (mm4, mm6); /* mm6 = v26 */
pmulhw_r2r (mm0, mm7); /* mm7 = b2/2 */
movq_m2r (*(col+offset+4*8), mm5); // mm5 = x4
movq_r2r (mm3, mm0); // mm0 = x0
movq_m2r (*(col+offset+4*8), mm5); /* mm5 = x4 */
movq_r2r (mm3, mm0); /* mm0 = x0 */
psubsw_r2r (mm5, mm3); // mm3 = v04
paddsw_r2r (mm5, mm0); // mm0 = u04
psubsw_r2r (mm5, mm3); /* mm3 = v04 */
paddsw_r2r (mm5, mm0); /* mm0 = u04 */
paddsw_r2r (mm3, mm4); // mm4 = a1
movq_r2r (mm0, mm5); // mm5 = u04
paddsw_r2r (mm3, mm4); /* mm4 = a1 */
movq_r2r (mm0, mm5); /* mm5 = u04 */
psubsw_r2r (mm6, mm3); // mm3 = a2
paddsw_r2r (mm2, mm5); // mm5 = a0
psubsw_r2r (mm6, mm3); /* mm3 = a2 */
paddsw_r2r (mm2, mm5); /* mm5 = a0 */
paddsw_r2r (mm1, mm1); // mm1 = b1
psubsw_r2r (mm2, mm0); // mm0 = a3
paddsw_r2r (mm1, mm1); /* mm1 = b1 */
psubsw_r2r (mm2, mm0); /* mm0 = a3 */
paddsw_r2r (mm7, mm7); // mm7 = b2
movq_r2r (mm3, mm2); // mm2 = a2
paddsw_r2r (mm7, mm7); /* mm7 = b2 */
movq_r2r (mm3, mm2); /* mm2 = a2 */
movq_r2r (mm4, mm6); // mm6 = a1
paddsw_r2r (mm7, mm3); // mm3 = a2+b2
movq_r2r (mm4, mm6); /* mm6 = a1 */
paddsw_r2r (mm7, mm3); /* mm3 = a2+b2 */
psraw_i2r (COL_SHIFT, mm3); // mm3 = y2
paddsw_r2r (mm1, mm4); // mm4 = a1+b1
psraw_i2r (COL_SHIFT, mm3); /* mm3 = y2 */
paddsw_r2r (mm1, mm4); /* mm4 = a1+b1 */
psraw_i2r (COL_SHIFT, mm4); // mm4 = y1
psubsw_r2r (mm1, mm6); // mm6 = a1-b1
psraw_i2r (COL_SHIFT, mm4); /* mm4 = y1 */
psubsw_r2r (mm1, mm6); /* mm6 = a1-b1 */
movq_m2r (*(col+offset+5*8), mm1); // mm1 = b0
psubsw_r2r (mm7, mm2); // mm2 = a2-b2
movq_m2r (*(col+offset+5*8), mm1); /* mm1 = b0 */
psubsw_r2r (mm7, mm2); /* mm2 = a2-b2 */
psraw_i2r (COL_SHIFT, mm6); // mm6 = y6
movq_r2r (mm5, mm7); // mm7 = a0
psraw_i2r (COL_SHIFT, mm6); /* mm6 = y6 */
movq_r2r (mm5, mm7); /* mm7 = a0 */
movq_r2m (mm4, *(col+offset+1*8)); // save y1
psraw_i2r (COL_SHIFT, mm2); // mm2 = y5
movq_r2m (mm4, *(col+offset+1*8)); /* save y1 */
psraw_i2r (COL_SHIFT, mm2); /* mm2 = y5 */
movq_r2m (mm3, *(col+offset+2*8)); // save y2
paddsw_r2r (mm1, mm5); // mm5 = a0+b0
movq_r2m (mm3, *(col+offset+2*8)); /* save y2 */
paddsw_r2r (mm1, mm5); /* mm5 = a0+b0 */
movq_m2r (*(col+offset+3*8), mm4); // mm4 = b3
psubsw_r2r (mm1, mm7); // mm7 = a0-b0
movq_m2r (*(col+offset+3*8), mm4); /* mm4 = b3 */
psubsw_r2r (mm1, mm7); /* mm7 = a0-b0 */
psraw_i2r (COL_SHIFT, mm5); // mm5 = y0
movq_r2r (mm0, mm3); // mm3 = a3
psraw_i2r (COL_SHIFT, mm5); /* mm5 = y0 */
movq_r2r (mm0, mm3); /* mm3 = a3 */
movq_r2m (mm2, *(col+offset+5*8)); // save y5
psubsw_r2r (mm4, mm3); // mm3 = a3-b3
movq_r2m (mm2, *(col+offset+5*8)); /* save y5 */
psubsw_r2r (mm4, mm3); /* mm3 = a3-b3 */
psraw_i2r (COL_SHIFT, mm7); // mm7 = y7
paddsw_r2r (mm0, mm4); // mm4 = a3+b3
psraw_i2r (COL_SHIFT, mm7); /* mm7 = y7 */
paddsw_r2r (mm0, mm4); /* mm4 = a3+b3 */
movq_r2m (mm5, *(col+offset+0*8)); // save y0
psraw_i2r (COL_SHIFT, mm3); // mm3 = y4
movq_r2m (mm5, *(col+offset+0*8)); /* save y0 */
psraw_i2r (COL_SHIFT, mm3); /* mm3 = y4 */
movq_r2m (mm6, *(col+offset+6*8)); // save y6
psraw_i2r (COL_SHIFT, mm4); // mm4 = y3
movq_r2m (mm6, *(col+offset+6*8)); /* save y6 */
psraw_i2r (COL_SHIFT, mm4); /* mm4 = y3 */
movq_r2m (mm7, *(col+offset+7*8)); // save y7
movq_r2m (mm7, *(col+offset+7*8)); /* save y7 */
movq_r2m (mm3, *(col+offset+4*8)); // save y4
movq_r2m (mm3, *(col+offset+4*8)); /* save y4 */
movq_r2m (mm4, *(col+offset+3*8)); // save y3
movq_r2m (mm4, *(col+offset+3*8)); /* save y3 */
#undef T1
#undef T2