Skip to content

GitLab

Commit 37131cfd authored by Nathan E. Egge's avatar Nathan E. Egge Committed by Nathan Egge
Browse files

daala_tx: New flattened 16-point Type-IV DST. 

Change-Id: Ic741f269d0bd5e5e295b55f95bfef05050bc31e5
parent 11eac7bf
Hide whitespace changes
Inline Side-by-side
av1/common/daala_tx.c
View file @ 37131cfd
......@@ -2119,6 +2119,471 @@
} \
while (0)
#define OD_FDST_16_FLAT(s0, s8, s4, sc, s2, sa, s6, se, \
s1, s9, s5, sd, s3, sb, s7, sf) \
/* Embedded 16-point orthonormal Type-IV fDST with flattened rotations. */ \
do { \
int t_; \
int u_; \
int s0h; \
int s4h; \
int sbh; \
int sfh; \
t_ = s1 + se; \
/* 32729/32768 ~= (Sin[17*Pi/64] + Cos[17*Pi/64])/Sqrt[2] ~=
0.9987954562051723 */ \
u_ = (se*32729 + 16384) >> 15; \
/* 201/2048 ~= (Sin[17*Pi/64] - Cos[17*Pi/64])*Sqrt[2] ~=
0.09813534865483615 */ \
se = (s1*201 + 1024) >> 11; \
/* 31121/32768 ~= Cos[17*Pi/64]*Sqrt[2] = 0.9497277818777543 */ \
t_ = (t_*31121 + 16384) >> 15; \
se += t_; \
s1 = u_ - OD_RSHIFT1(t_); \
t_ = s6 - s9; \
/* 32413/32768 ~= (Sin[19*Pi/64] + Cos[19*Pi/64])/Sqrt[2] ~=
0.9891765099647809 */ \
u_ = (s9*32413 + 16384) >> 15; \
/* 601/2048 ~= (Sin[19*Pi/64] - Cos[19*Pi/64])*Sqrt[2]
~= 0.29346094891072355 */ \
s9 = (s6*601 + 1024) >> 11; \
/* 27605/32768 ~= Cos[19*Pi/64]*Sqrt[2] = 0.8424460355094193 */ \
t_ = (t_*27605 + 16384) >> 15; \
s9 += t_; \
s6 = u_ + OD_RSHIFT1(t_); \
t_ = s5 + sa; \
/* 15893/16384 ~= (Sin[21*Pi/64] + Cos[21*Pi/64])/Sqrt[2] ~=
0.970031253194544 */ \
u_ = (sa*15893 + 8192) >> 14; \
/* 3981/8192 ~= (Sin[21*Pi/64] - Cos[21*Pi/64])*Sqrt[2] ~=
0.48596035980652796 */ \
sa = (s5*3981 + 4096) >> 13; \
/* 1489/2048 ~= Cos[21*Pi/64]*Sqrt[2] ~= 0.72705107329128 */ \
t_ = (t_*1489 + 1024) >> 11; \
sa += t_; \
s5 = OD_RSHIFT1(t_) - u_; \
t_ = sd - s2; \
/* 30853/32768 ~= (Sin[23*Pi/64] + Cos[23*Pi/64])/Sqrt[2] ~=
0.9415440651830208 */ \
u_ = (sd*30853 + 16384) >> 15; \
/* 11039/16384 ~= (Sin[23*Pi/64] - Cos[23*Pi/64])*Sqrt[2] ~=
0.6737797067844402 */ \
sd = (s2*11039 + 8192) >> 14; \
/* 19813/32768 ~= Cos[23*Pi/64]*Sqrt[2] ~= 0.6046542117908008 */ \
t_ = (t_*19813 + 16384) >> 15; \
sd -= t_; \
s2 = OD_RSHIFT1(t_) - u_; \
t_ = s3 + sc; \
/* 14811/16384 ~= (Sin[25*Pi/64] + Cos[25*Pi/64])/Sqrt[2] ~=
0.9039892931234433 */ \
u_ = (sc*14811 + 8192) >> 14; \
/* 7005/8192 ~= (Sin[25*Pi/64] - Cos[25*Pi/64])*Sqrt[2] ~=
0.8551101868605642 */ \
sc = (s3*7005 + 4096) >> 13; \
/* 3903/8192 ~= Cos[25*Pi/64]*Sqrt[2] ~= 0.47643419969316125 */ \
t_ = (t_*3903 + 4096) >> 13; \
sc += t_; \
s3 = u_ - OD_RSHIFT1(t_); \
t_ = sb - s4; \
/* 14053/16384 ~= (Sin[27*Pi/64] + Cos[27*Pi/64])/Sqrt[2] ~=
0.857728610000272 */ \
u_ = (sb*14053 + 8192) >> 14; \
/* 8423/8192 ~= (Sin[27*Pi/64] - Cos[27*Pi/64])*Sqrt[2] ~=
1.0282054883864435 */ \
sb = (s4*8423 + 4096) >> 13; \
/* 2815/8192 ~= Cos[27*Pi/64]*Sqrt[2] = 0.34362586580705035 */ \
t_ = (t_*2815 + 4096) >> 13; \
sb -= t_; \
s4 = OD_RSHIFT1(t_) - u_; \
t_ = s7 + s8; \
/* 1645/2048 ~= (Sin[29*Pi/64] + Cos[29*Pi/64])/Sqrt[2] ~=
0.8032075314806449 */ \
u_ = (s8*1645 + 1024) >> 11; \
/* 305/256 ~= (Sin[29*Pi/64] - Cos[29*Pi/64])*Sqrt[2] ~=
1.1913986089848667 */ \
s8 = (s7*305 + 128) >> 8; \
/* 425/2048 ~= Cos[29*Pi/64]*Sqrt[2] ~= 0.20750822698821159 */ \
t_ = (t_*425 + 1024) >> 11; \
s8 += t_; \
s7 = u_ - OD_RSHIFT1(t_); \
t_ = s0 - sf; \
/* 24279/32768 ~= (Sin[31*Pi/64] + Cos[31*Pi/64])/Sqrt[2] ~=
0.7409511253549591 */ \
u_ = (sf*24279 + 16384) >> 15; \
/* 44011/32768 ~= (Sin[31*Pi/64] - Cos[31*Pi/64])*Sqrt[2] ~=
1.3431179096940369 */ \
sf = (s0*44011 + 16384) >> 15; \
/* 1137/16384 ~= Cos[31*Pi/64]*Sqrt[2] ~= 0.0693921705079406 */ \
t_ = (t_*1137 + 8192) >> 14; \
s0 = u_ + OD_RSHIFT1(t_); \
sf += t_; \
s3 -= OD_RSHIFT1(sd); \
sd += s3; \
s2 += OD_RSHIFT1(sc); \
sc -= s2; \
s5 -= OD_RSHIFT1(sb); \
sb += s5; \
s4 -= OD_RSHIFT1(sa); \
sa += s4; \
s1 += OD_RSHIFT1(sf); \
sf -= s1; \
s7 -= OD_RSHIFT1(s9); \
s9 += s7; \
s6 -= OD_RSHIFT1(s8); \
s8 += s6; \
s0 += OD_RSHIFT1(se); \
se -= s0; \
sa -= s9; \
s9 += OD_RSHIFT1(sa); \
s5 += s6; \
s6 -= OD_RSHIFT1(s5); \
s1 -= s2; \
s2 += OD_RSHIFT1(s1); \
se += sd; \
sd -= OD_RSHIFT1(se); \
s0 += sc; \
s0h = OD_RSHIFT1(s0); \
sc -= s0h; \
sf -= s3; \
sfh = OD_RSHIFT1(sf); \
s3 += sfh; \
sb += s7; \
sbh = OD_RSHIFT1(sb); \
s7 -= sbh; \
s4 += s8; \
s4h = OD_RSHIFT1(s4); \
s8 -= s4h; \
t_ = OD_PAVG(s1, se); \
/* 9633/8192 ~= Sin[7*Pi/16] + Cos[7*Pi/16] ~= 1.1758756024193586 */ \
u_ = (s1*9633 + 4096) >> 13; \
/* 12873/16384 ~= Sin[7*Pi/16] - Cos[7*Pi/16] ~= 0.7856949583871022 */ \
s1 = (se*12873 + 8192) >> 14; \
/* 12785/32768 ~= 2*Cos[7*Pi/16] ~= 0.3901806440322565 */ \
t_ = (t_*12785 + 16384) >> 15; \
s1 += t_; \
se = u_ - t_; \
t_ = s6 + s9; \
/* 45451/32768 ~= Sin[5*Pi/16] + Cos[5*Pi/16] ~= 1.3870398453221475 */ \
u_ = (s9*45451 + 16384) >> 15; \
/* 9041/32768 ~= Sin[5*Pi/16] - Cos[5*Pi/16] ~= 0.27589937928294306 */ \
s9 = (s6*9041 + 16384) >> 15; \
/* 18205/32768 ~= Cos[5*Pi/16] ~= 0.5555702330196022 */ \
t_ = (t_*18205 + 16384) >> 15; \
s9 += t_; \
s6 = u_ - t_; \
t_ = OD_PAVG(s5, sa); \
/* 22725/16384 ~= Sin[5*Pi/16] + Cos[5*Pi/16] ~= 1.3870398453221475 */ \
u_ = (sa*22725 + 8192) >> 14; \
/* 9041/32768 ~= Sin[5*Pi/16] - Cos[5*Pi/16] ~= 0.27589937928294306 */ \
sa = (s5*9041 + 16384) >> 15; \
/* 18205/16384 ~= 2*Cos[5*Pi/16] ~= 1.1111404660392044 */ \
t_ = (t_*18205 + 8192) >> 14; \
sa += t_; \
s5 = t_ - u_; \
t_ = s2 + sd; \
/* 38531/32768 ~= Sin[7*Pi/16] + Cos[7*Pi/16] ~= 1.1758756024193586 */ \
u_ = (s2*38531 + 16384) >> 15; \
/* 12873/16384 ~= Sin[7*Pi/16] - Cos[7*Pi/16] ~= 0.7856949583871022 */ \
s2 = (sd*12873 + 8192) >> 14; \
/* 6393/32768 ~= Cos[7*Pi/16] ~= 0.19509032201612825 */ \
t_ = (t_*6393 + 16384) >> 15; \
s2 += t_; \
sd = u_ - t_; \
s3 -= s4h; \
s4 += s3; \
s8 -= s0h; \
s0 += s8; \
s7 += sfh; \
sf -= s7; \
sc += sbh; \
sb -= sc; \
s6 += OD_RSHIFT1(se) ;\
se -= s6; \
s9 -= OD_RSHIFT1(s1); \
s1 += s9; \
sd -= OD_RSHIFT1(s5); \
s5 += sd; \
s2 -= OD_RSHIFT1(sa); \
sa += s2; \
t_ = OD_PAVG(s3, sc); \
/* 21407/16384 ~= Sin[3*Pi/8] + Cos[3*Pi/8] ~= 1.3065629648763766 */ \
u_ = (s3*21407 + 8192) >> 14; \
/* 8867/16384 ~= Sin[3*Pi/8] - Cos[3*Pi/8] ~= 0.5411961001461969 */ \
s3 = (sc*8867 + 8192) >> 14; \
/* 3135/4096 ~= 2*Cos[3*Pi/8] ~= 0.7653668647301796 */ \
t_ = (t_*3135 + 2048) >> 12; \
s3 += t_; \
sc = u_ - t_; \
t_ = OD_PAVG(s4, sb); \
/* 21407/16384 ~= Sin[3*Pi/8] + Cos[3*Pi/8] ~= 1.3065629648763766 */ \
u_ = (s4*21407 + 8192) >> 14; \
/* 8867/16384 ~= Sin[3*Pi/8] - Cos[3*Pi/8] ~= 0.5411961001461969 */ \
s4 = (sb*8867 + 8192) >> 14; \
/* 3135/4096 ~= 2*Cos[3*Pi/8] ~= 0.7653668647301796 */ \
t_ = (t_*3134 + 2048) >> 12; \
s4 += t_; \
sb = u_ - t_; \
t_ = OD_PAVG(s5, sa); \
/* 11585/8192 ~= Sin[Pi/4] + Cos[Pi/4] ~= 1.4142135623730951 */ \
u_ = (sa*11585 + 4096) >> 13; \
/* 11585/8192 ~= 2*Cos[Pi/4] ~= 1.4142135623730951 */ \
sa = (t_*11585 + 4096) >> 13; \
s5 = sa - u_; \
t_ = OD_PAVG(s6, -s9); \
/* 11585/8192 ~= Sin[Pi/4] + Cos[Pi/4] ~= 1.4142135623730951 */ \
s6 = (s9*11585 + 4096) >> 13; \
/* 11585/8192 ~= 2*Cos[Pi/4] ~= 1.4142135623730951 */ \
s9 = (t_*11585 + 4096) >> 13; \
s6 += s9; \
t_ = OD_PAVG(s7, -s8); \
/* 11585/8192 ~= Sin[Pi/4] + Cos[Pi/4] ~= 1.4142135623730951 */ \
s7 = (s8*11585 + 4096) >> 13; \
/* 11585/8192 ~= 2*Cos[Pi/4] ~= 1.4142135623730951 */ \
s8 = (t_*11585 + 4096) >> 13; \
s7 += s8; \
} \
while (0)
#define OD_IDST_16_FLAT(s0, s1, s2, s3, s4, s5, s6, s7, \
s8, s9, sa, sb, sc, sd, se, sf) \
/* Embedded 16-point orthonormal Type-IV iDST with flattened rotations. */ \
do { \
int t_; \
int u_; \
int s0h; \
int s1h; \
int s2h; \
int s3h; \
int s4h; \
int s5h; \
int s6h; \
int s7h; \
int sbh; \
int sfh; \
t_ = OD_PAVG(s6, s9); \
/* 11585/8192 ~= Sin[Pi/4] + Cos[Pi/4] ~= 1.4142135623730951 */ \
s9 = (s6*11585 + 4096) >> 13; \
/* 11585/8192 ~= 2*Cos[Pi/4] ~= 1.4142135623730951 */ \
s6 = (t_*11585 + 4096) >> 13; \
s9 -= s6; \
t_ = OD_PAVG(s5, sa); \
/* 11585/8192 ~= Sin[Pi/4] + Cos[Pi/4] ~= 1.4142135623730951 */ \
sa = (s5*11585 + 4096) >> 13; \
/* 11585/8192 ~= 2*Cos[Pi/4] ~= 1.4142135623730951 */ \
s5 = (t_*11585 + 4096) >> 13; \
sa -= s5; \
t_ = OD_PAVG(s7, s8); \
/* 11585/8192 ~= Sin[Pi/4] + Cos[Pi/4] ~= 1.4142135623730951 */ \
s8 = (s7*11585 + 4096) >> 13; \
/* 11585/8192 ~= 2*Cos[Pi/4] ~= 1.4142135623730951 */ \
s7 = (t_*11585 + 4096) >> 13; \
s8 -= s7; \
t_ = OD_PAVG(s3, sc); \
/* 21407/16384 ~= Sin[3*Pi/8] + Cos[3*Pi/8] ~= 1.3065629648763766 */ \
u_ = (s3*21407 + 8192) >> 14; \
/* 8867/16384 ~= Sin[3*Pi/8] - Cos[3*Pi/8] ~= 0.5411961001461969 */ \
s3 = (sc*8867 + 8192) >> 14; \
/* 3135/4096 ~= 2*Cos[3*Pi/8] ~= 0.7653668647301796 */ \
t_ = (t_*3135 + 2048) >> 12; \
s3 += t_; \
sc = u_ - t_; \
t_ = OD_PAVG(sb, -s4); \
/* 21407/16384 ~= Sin[3*Pi/8] + Cos[3*Pi/8] ~= 1.3065629648763766 */ \
u_ = (sb*21407 + 8192) >> 14; \
/* 8867/16384 ~= Sin[3*Pi/8] - Cos[3*Pi/8] ~= 0.5411961001461969 */ \
sb = (s4*8867 + 8192) >> 14; \
/* 3135/4096 ~= 2*Cos[3*Pi/8] ~= 0.7653668647301796 */ \
t_ = (t_*3135 + 2048) >> 12; \
sb -= t_; \
s4 = t_ - u_; \
sa += s2; \
s2 -= OD_RSHIFT1(sa); \
s5 -= sd; \
sd += OD_RSHIFT1(s5); \
s1 -= s9; \
s9 += OD_RSHIFT1(s1); \
se += s6; \
s6 -= OD_RSHIFT1(se); \
sb += sc; \
sbh = OD_RSHIFT1(sb); \
sc -= sbh; \
sf += s7; \
sfh = OD_RSHIFT1(sf); \
s7 -= sfh; \
s0 -= s8; \
s0h = OD_RSHIFT1(s0); \
s8 += s0h; \
s4 += s3; \
s4h = OD_RSHIFT1(s4); \
s3 -= s4h; \
t_ = sd - s2; \
/* 38531/32768 ~= Sin[7*Pi/16] + Cos[7*Pi/16] ~= 1.1758756024193586 */ \
u_ = (sd*38531 + 16384) >> 15; \
/* 12873/16384 ~= Sin[7*Pi/16] - Cos[7*Pi/16] ~= 0.7856949583871022 */ \
sd = (s2*12873 + 8192) >> 14; \
/* 6393/32768 ~= Cos[7*Pi/16] ~= 0.19509032201612825 */ \
t_ = (t_*6393 + 16384) >> 15; \
sd -= t_; \
s2 = t_ - u_; \
t_ = OD_PAVG(s5, -sa); \
/* 22725/16384 ~= Sin[5*Pi/16] + Cos[5*Pi/16] ~= 1.3870398453221475 */ \
u_ = (s5*22725 + 8192) >> 14; \
/* 9041/32768 ~= Sin[5*Pi/16] - Cos[5*Pi/16] ~= 0.27589937928294306 */ \
s5 = (sa*9041 + 16384) >> 15; \
/* 18205/16384 ~= 2*Cos[5*Pi/16] ~= 1.1111404660392044 */ \
t_ = (t_*18205 + 8192) >> 14; \
s5 -= t_; \
sa = t_ - u_; \
t_ = s6 + s9; \
/* 45451/32768 ~= Sin[5*Pi/16] + Cos[5*Pi/16] ~= 1.3870398453221475 */ \
u_ = (s9*45451 + 16384) >> 15; \
/* 9041/32768 ~= Sin[5*Pi/16] - Cos[5*Pi/16] ~= 0.27589937928294306 */ \
s9 = (s6*9041 + 16384) >> 15; \
/* 18205/32768 ~= Cos[5*Pi/16] ~= 0.5555702330196022 */ \
t_ = (t_*18205 + 16384) >> 15; \
s9 += t_; \
s6 = u_ - t_; \
t_ = OD_PAVG(s1, se); \
/* 9633/8192 ~= Sin[7*Pi/16] + Cos[7*Pi/16] ~= 1.1758756024193586 */ \
u_ = (s1*9633 + 4096) >> 13; \
/* 12873/16384 ~= Sin[7*Pi/16] - Cos[7*Pi/16] ~= 0.7856949583871022 */ \
s1 = (se*12873 + 8192) >> 14; \
/* 12785/32768 ~= 2*Cos[7*Pi/16] ~= 0.3901806440322565 */ \
t_ = (t_*12785 + 16384) >> 15; \
s1 += t_; \
se = u_ - t_; \
s8 -= s4h; \
s4 += s8; \
s7 += sbh; \
sb -= s7; \
s3 -= sfh; \
sf += s3; \
sc += s0h; \
s0 -= sc; \
sd += OD_RSHIFT1(se); \
se -= sd; \
s2 += OD_RSHIFT1(s1); \
s1 -= s2; \
s6 -= OD_RSHIFT1(s5); \
s5 += s6; \
s9 -= OD_RSHIFT1(sa); \
sa += s9; \
s0 -= se; \
s0h = OD_RSHIFT1(s0); \
se += s0h; \
s1 -= sf; \
s1h = OD_RSHIFT1(s1); \
sf += s1h; \
s2 += sc; \
s2h = OD_RSHIFT1(s2); \
sc -= s2h; \
s3 += sd; \
s3h = OD_RSHIFT1(s3); \
sd -= s3h; \
s4 -= sa; \
s4h = OD_RSHIFT1(s4); \
sa += s4h; \
s5 -= sb; \
s5h = OD_RSHIFT1(s5); \
sb += s5h; \
s6 += s8; \
s6h = OD_RSHIFT1(s6); \
s8 -= s6h; \
s7 += s9; \
s7h = OD_RSHIFT1(s7); \
s9 -= s7h; \
t_ = se - s1h; \
/* 32729/32768 ~= (Sin[17*Pi/64] + Cos[17*Pi/64])/Sqrt[2] ~=
0.9987954562051723 */ \
u_ = (s1*32729 + 16384) >> 15; \
/* 201/2048 ~= (Sin[17*Pi/64] - Cos[17*Pi/64])*Sqrt[2] ~=
0.09813534865483615 */ \
s1 = (se*201 + 1024) >> 11; \
/* 31121/32768 ~= Cos[17*Pi/64]*Sqrt[2] ~=
0.9497277818777543 */ \
t_ = (t_*31121 + 16384) >> 15; \
s1 += t_; \
se = u_ + t_; \
t_ = s6h + s9; \
/* 32413/32768 ~= (Sin[19*Pi/64] + Cos[19*Pi/64])/Sqrt[2] ~=
0.9891765099647809 */ \
u_ = (s6*32413 + 16384) >> 15; \
/* 601/2048 ~= (Sin[19*Pi/64] - Cos[19*Pi/64])*Sqrt[2] ~=
0.29346094891072355 */ \
s6 = (s9*601 + 1024) >> 11; \
/* 27605/32768 ~= Cos[19*Pi/64]*Sqrt[2] ~= 0.8424460355094193 */ \
t_ = (t_*27605 + 16384) >> 15; \
s6 += t_; \
s9 = u_ - t_; \
t_ = sa - s5h; \
/* 15893/16384 ~= (Sin[21*Pi/64] + Cos[21*Pi/64])/Sqrt[2] ~=
0.970031253194544 */ \
u_ = (s5*15893 + 8192) >> 14; \
/* 3981/8192 ~= (Sin[21*Pi/64] - Cos[21*Pi/64])*Sqrt[2] ~=
0.48596035980652796 */ \
s5 = (sa*3981 + 4096) >> 13; \
/* 1489/2048 ~= Cos[21*Pi/64]*Sqrt[2] ~= 0.72705107329128 */ \
t_ = (t_*1489 + 1024) >> 11; \
s5 += t_; \
sa = u_ + t_; \
t_ = s2h + sd; \
/* 30853/32768 ~= (Sin[23*Pi/64] + Cos[23*Pi/64])/Sqrt[2] ~=
0.9415440651830208 */ \
u_ = (s2*30853 + 16384) >> 15; \
/* 11039/16384 ~= (Sin[23*Pi/64] - Cos[23*Pi/64])*Sqrt[2] ~=
0.6737797067844402 */ \
s2 = (sd*11039 + 8192) >> 14; \
/* 19813/32768 ~= Cos[23*Pi/64]*Sqrt[2] ~= 0.6046542117908008 */ \
t_ = (t_*19813 + 16384) >> 15; \
s2 += t_; \
sd = u_ - t_; \
t_ = sc - s3h; \
/* 14811/16384 ~= (Sin[25*Pi/64] + Cos[25*Pi/64])/Sqrt[2] ~=
0.9039892931234433 */ \
u_ = (s3*14811 + 8192) >> 14; \
/* 7005/8192 ~= (Sin[25*Pi/64] - Cos[25*Pi/64])*Sqrt[2] ~=
0.8551101868605642 */ \
s3 = (sc*7005 + 4096) >> 13; \
/* 3903/8192 ~= Cos[25*Pi/64]*Sqrt[2] ~= 0.47643419969316125 */ \
t_ = (t_*3903 + 4096) >> 13; \
s3 += t_; \
sc = u_ + t_; \
t_ = s4h + sb; \
/* 14053/16384 ~= (Sin[27*Pi/64] + Cos[27*Pi/64])/Sqrt[2] ~=
0.857728610000272 */ \
u_ = (s4*14053 + 8192) >> 14; \
/* 8423/8192 ~= (Sin[27*Pi/64] - Cos[27*Pi/64])*Sqrt[2] ~=
1.0282054883864435 */ \
s4 = (sb*8423 + 4096) >> 13; \
/* 2815/8192 ~= Cos[27*Pi/64]*Sqrt[2] ~= 0.34362586580705035 */ \
t_ = (t_*2815 + 4096) >> 13; \
s4 += t_; \
sb = u_ - t_; \
t_ = s8 - s7h; \
/* 1645/2048 ~= (Sin[29*Pi/64] + Cos[29*Pi/64])/Sqrt[2] ~=
0.8032075314806449 */ \
u_ = (s7*1645 + 1024) >> 11; \
/* 305/256 ~= (Sin[29*Pi/64] - Cos[29*Pi/64])*Sqrt[2] ~=
1.1913986089848667 */ \
s7 = (s8*305 + 128) >> 8; \
/* 425/2048 ~= Cos[29*Pi/64]*Sqrt[2] ~= 0.20750822698821159 */ \
t_ = (t_*425 + 1024) >> 11; \
s7 += t_; \
s8 = u_ + t_; \
t_ = s0h + sf; \
/* 24279/32768 ~= (Sin[31*Pi/64] + Cos[31*Pi/64])/Sqrt[2] ~=
0.7409511253549591 */ \
u_ = (s0*24279 + 16384) >> 15; \
/* 44011/32768 ~= (Sin[31*Pi/64] - Cos[31*Pi/64])*Sqrt[2] ~=
1.3431179096940369 */ \
s0 = (sf*44011 + 16384) >> 15; \
/* 1137/16384 ~= Cos[31*Pi/64]*Sqrt[2] ~= 0.06939217050794069 */ \
t_ = (t_*1137 + 8192) >> 14; \
s0 += t_; \
sf = u_ - t_; \
} \
while (0)
/* TODO: rewrite this to match OD_FDST_16. */
#define OD_FDST_16_ASYM_PR(t0, t0h, t8, t4, t4h, tc, t2, ta, t6, te, \
t1, t9, t5, td, t3, tb, t7, t7h, tf) \
......@@ -5772,39 +6237,40 @@ void od_bin_fdst16(od_coeff y[16], const od_coeff *x, int xstride) {
int sd;
int se;
int sf;
s0 = x[15*xstride];
s8 = x[14*xstride];
s4 = x[13*xstride];
sc = x[12*xstride];
s2 = x[11*xstride];
sa = x[10*xstride];
s6 = x[9*xstride];
se = x[8*xstride];
s1 = x[7*xstride];
s9 = x[6*xstride];
s5 = x[5*xstride];
sd = x[4*xstride];
s3 = x[3*xstride];
sb = x[2*xstride];
s7 = x[1*xstride];
sf = x[0*xstride];
OD_FDST_16_PR(s0, s8, s4, sc, s2, sa, s6, se, s1, s9, s5, sd, s3, sb, s7, sf);
y[0] = (od_coeff)sf;
y[1] = (od_coeff)-se;
y[2] = (od_coeff)sd;
y[3] = (od_coeff)-sc;
y[4] = (od_coeff)sb;
y[5] = (od_coeff)-sa;
y[6] = (od_coeff)s9;
y[7] = (od_coeff)-s8;
y[8] = (od_coeff)s7;
y[9] = (od_coeff)-s6;
y[10] = (od_coeff)s5;
y[11] = (od_coeff)-s4;
y[12] = (od_coeff)s3;
y[13] = (od_coeff)-s2;
y[14] = (od_coeff)s1;
y[15] = (od_coeff)-s0;
s0 = x[0*xstride];
s8 = x[1*xstride];
s4 = x[2*xstride];
sc = x[3*xstride];
s2 = x[4*xstride];
sa = x[5*xstride];
s6 = x[6*xstride];
se = x[7*xstride];
s1 = x[8*xstride];
s9 = x[9*xstride];
s5 = x[10*xstride];
sd = x[11*xstride];
s3 = x[12*xstride];
sb = x[13*xstride];
s7 = x[14*xstride];
sf = x[15*xstride];
OD_FDST_16_FLAT(s0, s8, s4, sc, s2, sa, s6, se,
s1, s9, s5, sd, s3, sb, s7, sf);
y[0] = (od_coeff)s0;
y[1] = (od_coeff)s1;
y[2] = (od_coeff)s2;
y[3] = (od_coeff)s3;
y[4] = (od_coeff)s4;
y[5] = (od_coeff)s5;
y[6] = (od_coeff)s6;
y[7] = (od_coeff)s7;
y[8] = (od_coeff)s8;
y[9] = (od_coeff)s9;
y[10] = (od_coeff)sa;
y[11] = (od_coeff)sb;
y[12] = (od_coeff)sc;
y[13] = (od_coeff)sd;
y[14] = (od_coeff)se;
y[15] = (od_coeff)sf;
}
void od_bin_idst16(od_coeff *x, int xstride, const od_coeff y[16]) {
......@@ -5824,39 +6290,40 @@ void od_bin_idst16(od_coeff *x, int xstride, const od_coeff y[16]) {
int sd;
int se;
int sf;
s0 = -y[15];
s8 = y[14];
s4 = -y[13];
sc = y[12];
s2 = -y[11];
sa = y[10];
s6 = -y[9];
se = y[8];
s1 = -y[7];
s9 = y[6];
s5 = -y[5];
sd = y[4];
s3 = -y[3];
sb = y[2];
s7 = -y[1];
sf = y[0];
OD_IDST_16_PR(s0, s8, s4, sc, s2, sa, s6, se, s1, s9, s5, sd, s3, sb, s7, sf);
x[0*xstride] = (od_coeff)sf;
x[1*xstride] = (od_coeff)se;
x[2*xstride] = (od_coeff)sd;
x[3*xstride] = (od_coeff)sc;
x[4*xstride] = (od_coeff)sb;
x[5*xstride] = (od_coeff)sa;
x[6*xstride] = (od_coeff)s9;
x[7*xstride] = (od_coeff)s8;
x[8*xstride] = (od_coeff)s7;
x[9*xstride] = (od_coeff)s6;
x[10*xstride] = (od_coeff)s5;
x[11*xstride] = (od_coeff)s4;
x[12*xstride] = (od_coeff)s3;
x[13*xstride] = (od_coeff)s2;
x[14*xstride] = (od_coeff)s1;
x[15*xstride] = (od_coeff)s0;
s0 = y[0];
s8 = y[1];
s4 = y[2];
sc = y[3];
s2 = y[4];
sa = y[5];
s6 = y[6];
se = y[7];
s1 = y[8];
s9 = y[9];
s5 = y[10];
sd = y[11];
s3 = y[12];
sb = y[13];
s7 = y[14];
sf = y[15];
OD_IDST_16_FLAT(s0, s8, s4, sc, s2, sa, s6, se,
s1, s9, s5, sd, s3, sb, s7, sf);
x[0*xstride] = (od_coeff)s0;
x[1*xstride] = (od_coeff)s1;
x[2*xstride] = (od_coeff)s2;
x[3*xstride] = (od_coeff)s3;
x[4*xstride] = (od_coeff)s4;
x[5*xstride] = (od_coeff)s5;
x[6*xstride] = (od_coeff)s6;
x[7*xstride] = (od_coeff)s7;
x[8*xstride] = (od_coeff)s8;
x[9*xstride] = (od_coeff)s9;
x[10*xstride] = (od_coeff)sa;
x[11*xstride] = (od_coeff)sb;
x[12*xstride] = (od_coeff)sc;
x[13*xstride] = (od_coeff)sd;
x[14*xstride] = (od_coeff)se;
x[15*xstride] = (od_coeff)sf;
}
void od_bin_fdct32(od_coeff y[32], const od_coeff *x, int xstride) {
......