summaryrefslogtreecommitdiff
path: root/src/libffmpeg/libavcodec/vp3dsp.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/libffmpeg/libavcodec/vp3dsp.c')
-rw-r--r--src/libffmpeg/libavcodec/vp3dsp.c240
1 files changed, 76 insertions, 164 deletions
diff --git a/src/libffmpeg/libavcodec/vp3dsp.c b/src/libffmpeg/libavcodec/vp3dsp.c
index 0cbe8d551..f5a1fb6ff 100644
--- a/src/libffmpeg/libavcodec/vp3dsp.c
+++ b/src/libffmpeg/libavcodec/vp3dsp.c
@@ -35,14 +35,15 @@
#define xC6S2 25080
#define xC7S1 12785
+#define M(a,b) (((a) * (b))>>16)
+
static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
{
int16_t *ip = input;
uint8_t *cm = cropTbl + MAX_NEG_CROP;
- int A_, B_, C_, D_, _Ad, _Bd, _Cd, _Dd, E_, F_, G_, H_;
- int _Ed, _Gd, _Add, _Bdd, _Fd, _Hd;
- int t1, t2;
+ int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
+ int Ed, Gd, Add, Bdd, Fd, Hd;
int i;
@@ -50,86 +51,44 @@ static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int typ
for (i = 0; i < 8; i++) {
/* Check for non-zero values */
if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
- t1 = (int32_t)(xC1S7 * ip[1]);
- t2 = (int32_t)(xC7S1 * ip[7]);
- t1 >>= 16;
- t2 >>= 16;
- A_ = t1 + t2;
-
- t1 = (int32_t)(xC7S1 * ip[1]);
- t2 = (int32_t)(xC1S7 * ip[7]);
- t1 >>= 16;
- t2 >>= 16;
- B_ = t1 - t2;
-
- t1 = (int32_t)(xC3S5 * ip[3]);
- t2 = (int32_t)(xC5S3 * ip[5]);
- t1 >>= 16;
- t2 >>= 16;
- C_ = t1 + t2;
-
- t1 = (int32_t)(xC3S5 * ip[5]);
- t2 = (int32_t)(xC5S3 * ip[3]);
- t1 >>= 16;
- t2 >>= 16;
- D_ = t1 - t2;
-
-
- t1 = (int32_t)(xC4S4 * (A_ - C_));
- t1 >>= 16;
- _Ad = t1;
+ A = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
+ B = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
+ C = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
+ D = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
- t1 = (int32_t)(xC4S4 * (B_ - D_));
- t1 >>= 16;
- _Bd = t1;
+ Ad = M(xC4S4, (A - C));
+ Bd = M(xC4S4, (B - D));
+ Cd = A + C;
+ Dd = B + D;
- _Cd = A_ + C_;
- _Dd = B_ + D_;
+ E = M(xC4S4, (ip[0] + ip[4]));
+ F = M(xC4S4, (ip[0] - ip[4]));
- t1 = (int32_t)(xC4S4 * (ip[0] + ip[4]));
- t1 >>= 16;
- E_ = t1;
+ G = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
+ H = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
- t1 = (int32_t)(xC4S4 * (ip[0] - ip[4]));
- t1 >>= 16;
- F_ = t1;
+ Ed = E - G;
+ Gd = E + G;
- t1 = (int32_t)(xC2S6 * ip[2]);
- t2 = (int32_t)(xC6S2 * ip[6]);
- t1 >>= 16;
- t2 >>= 16;
- G_ = t1 + t2;
+ Add = F + Ad;
+ Bdd = Bd - H;
- t1 = (int32_t)(xC6S2 * ip[2]);
- t2 = (int32_t)(xC2S6 * ip[6]);
- t1 >>= 16;
- t2 >>= 16;
- H_ = t1 - t2;
-
-
- _Ed = E_ - G_;
- _Gd = E_ + G_;
-
- _Add = F_ + _Ad;
- _Bdd = _Bd - H_;
-
- _Fd = F_ - _Ad;
- _Hd = _Bd + H_;
+ Fd = F - Ad;
+ Hd = Bd + H;
/* Final sequence of operations over-write original inputs. */
- ip[0] = _Gd + _Cd ;
- ip[7] = _Gd - _Cd ;
-
- ip[1] = _Add + _Hd;
- ip[2] = _Add - _Hd;
+ ip[0] = Gd + Cd ;
+ ip[7] = Gd - Cd ;
- ip[3] = _Ed + _Dd ;
- ip[4] = _Ed - _Dd ;
+ ip[1] = Add + Hd;
+ ip[2] = Add - Hd;
- ip[5] = _Fd + _Bdd;
- ip[6] = _Fd - _Bdd;
+ ip[3] = Ed + Dd ;
+ ip[4] = Ed - Dd ;
+ ip[5] = Fd + Bdd;
+ ip[6] = Fd - Bdd;
}
ip += 8; /* next row */
@@ -142,121 +101,74 @@ static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int typ
if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
- t1 = (int32_t)(xC1S7 * ip[1*8]);
- t2 = (int32_t)(xC7S1 * ip[7*8]);
- t1 >>= 16;
- t2 >>= 16;
- A_ = t1 + t2;
-
- t1 = (int32_t)(xC7S1 * ip[1*8]);
- t2 = (int32_t)(xC1S7 * ip[7*8]);
- t1 >>= 16;
- t2 >>= 16;
- B_ = t1 - t2;
+ A = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
+ B = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
+ C = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
+ D = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
- t1 = (int32_t)(xC3S5 * ip[3*8]);
- t2 = (int32_t)(xC5S3 * ip[5*8]);
- t1 >>= 16;
- t2 >>= 16;
- C_ = t1 + t2;
+ Ad = M(xC4S4, (A - C));
+ Bd = M(xC4S4, (B - D));
- t1 = (int32_t)(xC3S5 * ip[5*8]);
- t2 = (int32_t)(xC5S3 * ip[3*8]);
- t1 >>= 16;
- t2 >>= 16;
- D_ = t1 - t2;
+ Cd = A + C;
+ Dd = B + D;
+ E = M(xC4S4, (ip[0*8] + ip[4*8])) + 8;
+ F = M(xC4S4, (ip[0*8] - ip[4*8])) + 8;
- t1 = (int32_t)(xC4S4 * (A_ - C_));
- t1 >>= 16;
- _Ad = t1;
-
- t1 = (int32_t)(xC4S4 * (B_ - D_));
- t1 >>= 16;
- _Bd = t1;
-
-
- _Cd = A_ + C_;
- _Dd = B_ + D_;
-
- t1 = (int32_t)(xC4S4 * (ip[0*8] + ip[4*8]));
- t1 >>= 16;
- E_ = t1;
-
- t1 = (int32_t)(xC4S4 * (ip[0*8] - ip[4*8]));
- t1 >>= 16;
- F_ = t1;
-
- t1 = (int32_t)(xC2S6 * ip[2*8]);
- t2 = (int32_t)(xC6S2 * ip[6*8]);
- t1 >>= 16;
- t2 >>= 16;
- G_ = t1 + t2;
-
- t1 = (int32_t)(xC6S2 * ip[2*8]);
- t2 = (int32_t)(xC2S6 * ip[6*8]);
- t1 >>= 16;
- t2 >>= 16;
- H_ = t1 - t2;
-
+ if(type==1){ //HACK
+ E += 16*128;
+ F += 16*128;
+ }
- _Ed = E_ - G_;
- _Gd = E_ + G_;
+ G = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
+ H = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
- _Add = F_ + _Ad;
- _Bdd = _Bd - H_;
+ Ed = E - G;
+ Gd = E + G;
- _Fd = F_ - _Ad;
- _Hd = _Bd + H_;
+ Add = F + Ad;
+ Bdd = Bd - H;
- if(type==1){ //HACK
- _Gd += 16*128;
- _Add+= 16*128;
- _Ed += 16*128;
- _Fd += 16*128;
- }
- _Gd += IdctAdjustBeforeShift;
- _Add += IdctAdjustBeforeShift;
- _Ed += IdctAdjustBeforeShift;
- _Fd += IdctAdjustBeforeShift;
+ Fd = F - Ad;
+ Hd = Bd + H;
/* Final sequence of operations over-write original inputs. */
if(type==0){
- ip[0*8] = (_Gd + _Cd ) >> 4;
- ip[7*8] = (_Gd - _Cd ) >> 4;
+ ip[0*8] = (Gd + Cd ) >> 4;
+ ip[7*8] = (Gd - Cd ) >> 4;
- ip[1*8] = (_Add + _Hd ) >> 4;
- ip[2*8] = (_Add - _Hd ) >> 4;
+ ip[1*8] = (Add + Hd ) >> 4;
+ ip[2*8] = (Add - Hd ) >> 4;
- ip[3*8] = (_Ed + _Dd ) >> 4;
- ip[4*8] = (_Ed - _Dd ) >> 4;
+ ip[3*8] = (Ed + Dd ) >> 4;
+ ip[4*8] = (Ed - Dd ) >> 4;
- ip[5*8] = (_Fd + _Bdd ) >> 4;
- ip[6*8] = (_Fd - _Bdd ) >> 4;
+ ip[5*8] = (Fd + Bdd ) >> 4;
+ ip[6*8] = (Fd - Bdd ) >> 4;
}else if(type==1){
- dst[0*stride] = cm[(_Gd + _Cd ) >> 4];
- dst[7*stride] = cm[(_Gd - _Cd ) >> 4];
+ dst[0*stride] = cm[(Gd + Cd ) >> 4];
+ dst[7*stride] = cm[(Gd - Cd ) >> 4];
- dst[1*stride] = cm[(_Add + _Hd ) >> 4];
- dst[2*stride] = cm[(_Add - _Hd ) >> 4];
+ dst[1*stride] = cm[(Add + Hd ) >> 4];
+ dst[2*stride] = cm[(Add - Hd ) >> 4];
- dst[3*stride] = cm[(_Ed + _Dd ) >> 4];
- dst[4*stride] = cm[(_Ed - _Dd ) >> 4];
+ dst[3*stride] = cm[(Ed + Dd ) >> 4];
+ dst[4*stride] = cm[(Ed - Dd ) >> 4];
- dst[5*stride] = cm[(_Fd + _Bdd ) >> 4];
- dst[6*stride] = cm[(_Fd - _Bdd ) >> 4];
+ dst[5*stride] = cm[(Fd + Bdd ) >> 4];
+ dst[6*stride] = cm[(Fd - Bdd ) >> 4];
}else{
- dst[0*stride] = cm[dst[0*stride] + ((_Gd + _Cd ) >> 4)];
- dst[7*stride] = cm[dst[7*stride] + ((_Gd - _Cd ) >> 4)];
+ dst[0*stride] = cm[dst[0*stride] + ((Gd + Cd ) >> 4)];
+ dst[7*stride] = cm[dst[7*stride] + ((Gd - Cd ) >> 4)];
- dst[1*stride] = cm[dst[1*stride] + ((_Add + _Hd ) >> 4)];
- dst[2*stride] = cm[dst[2*stride] + ((_Add - _Hd ) >> 4)];
+ dst[1*stride] = cm[dst[1*stride] + ((Add + Hd ) >> 4)];
+ dst[2*stride] = cm[dst[2*stride] + ((Add - Hd ) >> 4)];
- dst[3*stride] = cm[dst[3*stride] + ((_Ed + _Dd ) >> 4)];
- dst[4*stride] = cm[dst[4*stride] + ((_Ed - _Dd ) >> 4)];
+ dst[3*stride] = cm[dst[3*stride] + ((Ed + Dd ) >> 4)];
+ dst[4*stride] = cm[dst[4*stride] + ((Ed - Dd ) >> 4)];
- dst[5*stride] = cm[dst[5*stride] + ((_Fd + _Bdd ) >> 4)];
- dst[6*stride] = cm[dst[6*stride] + ((_Fd - _Bdd ) >> 4)];
+ dst[5*stride] = cm[dst[5*stride] + ((Fd + Bdd ) >> 4)];
+ dst[6*stride] = cm[dst[6*stride] + ((Fd - Bdd ) >> 4)];
}
} else {
generated by cgit v1.2.3 (git 2.25.1) at 2025-07-02 21:58:29 +0000