source/libvpx/vp9/common/vp9_idct.c - Issue 756673003: libvpx: Pull from upstream

Unified Diff: source/libvpx/vp9/common/vp9_idct.c

Issue 756673003: libvpx: Pull from upstream (Closed) Base URL: svn://svn.chromium.org/chrome/trunk/deps/third_party/libvpx/

Patch Set: Created 6 years, 1 month ago

Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.

Jump to:

View side-by-side diff with in-line comments

Download patch

Index: source/libvpx/vp9/common/vp9_idct.c

===================================================================

--- source/libvpx/vp9/common/vp9_idct.c (revision 293081)

+++ source/libvpx/vp9/common/vp9_idct.c (working copy)

@@ -34,7 +34,7 @@

// bd of x uses trans_low with 8+x bits, need to remove 24-x bits

#define WRAPLOW(x, bd) ((((int32_t)(x)) << (24 - bd)) >> (24 - bd))

#else

-#define WRAPLOW(x, bd) (x)

+#define WRAPLOW(x, bd) ((int32_t)(x))

#endif // CONFIG_EMULATE_HARDWARE

#if CONFIG_VP9_HIGHBITDEPTH

@@ -276,10 +276,10 @@

static void iadst4(const tran_low_t *input, tran_low_t *output) {

tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;

- tran_high_t x0 = input[0];

- tran_high_t x1 = input[1];

- tran_high_t x2 = input[2];

- tran_high_t x3 = input[3];

+ tran_low_t x0 = input[0];

+ tran_low_t x1 = input[1];

+ tran_low_t x2 = input[2];

+ tran_low_t x3 = input[3];

if (!(x0 | x1 | x2 | x3)) {

output[0] = output[1] = output[2] = output[3] = 0;

@@ -295,24 +295,19 @@

s6 = sinpi_4_9 * x3;

s7 = x0 - x2 + x3;

- x0 = s0 + s3 + s5;

- x1 = s1 - s4 - s6;

- x2 = sinpi_3_9 * s7;

- x3 = s2;

+ s0 = s0 + s3 + s5;

+ s1 = s1 - s4 - s6;

+ s3 = s2;

+ s2 = sinpi_3_9 * s7;

- s0 = x0 + x3;

- s1 = x1 + x3;

- s2 = x2;

- s3 = x0 + x1 - x3;

// 1-D transform scaling factor is sqrt(2).

// The overall dynamic range is 14b (input) + 14b (multiplication scaling)

// + 1b (addition) = 29b.

// Hence the output bit depth is 15b.

- output[0] = WRAPLOW(dct_const_round_shift(s0), 8);

- output[1] = WRAPLOW(dct_const_round_shift(s1), 8);

+ output[0] = WRAPLOW(dct_const_round_shift(s0 + s3), 8);

+ output[1] = WRAPLOW(dct_const_round_shift(s1 + s3), 8);

output[2] = WRAPLOW(dct_const_round_shift(s2), 8);

- output[3] = WRAPLOW(dct_const_round_shift(s3), 8);

+ output[3] = WRAPLOW(dct_const_round_shift(s0 + s1 - s3), 8);

}

void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride,

@@ -367,14 +362,14 @@

}

// stage 1

- s0 = cospi_2_64 * x0 + cospi_30_64 * x1;

- s1 = cospi_30_64 * x0 - cospi_2_64 * x1;

- s2 = cospi_10_64 * x2 + cospi_22_64 * x3;

- s3 = cospi_22_64 * x2 - cospi_10_64 * x3;

- s4 = cospi_18_64 * x4 + cospi_14_64 * x5;

- s5 = cospi_14_64 * x4 - cospi_18_64 * x5;

- s6 = cospi_26_64 * x6 + cospi_6_64 * x7;

- s7 = cospi_6_64 * x6 - cospi_26_64 * x7;

+ s0 = (int)(cospi_2_64 * x0 + cospi_30_64 * x1);

+ s1 = (int)(cospi_30_64 * x0 - cospi_2_64 * x1);

+ s2 = (int)(cospi_10_64 * x2 + cospi_22_64 * x3);

+ s3 = (int)(cospi_22_64 * x2 - cospi_10_64 * x3);

+ s4 = (int)(cospi_18_64 * x4 + cospi_14_64 * x5);

+ s5 = (int)(cospi_14_64 * x4 - cospi_18_64 * x5);

+ s6 = (int)(cospi_26_64 * x6 + cospi_6_64 * x7);

+ s7 = (int)(cospi_6_64 * x6 - cospi_26_64 * x7);

x0 = WRAPLOW(dct_const_round_shift(s0 + s4), 8);

x1 = WRAPLOW(dct_const_round_shift(s1 + s5), 8);

@@ -386,14 +381,14 @@

x7 = WRAPLOW(dct_const_round_shift(s3 - s7), 8);

// stage 2

- s0 = x0;

- s1 = x1;

- s2 = x2;

- s3 = x3;

- s4 = cospi_8_64 * x4 + cospi_24_64 * x5;

- s5 = cospi_24_64 * x4 - cospi_8_64 * x5;

- s6 = -cospi_24_64 * x6 + cospi_8_64 * x7;

- s7 = cospi_8_64 * x6 + cospi_24_64 * x7;

+ s0 = (int)x0;

+ s1 = (int)x1;

+ s2 = (int)x2;

+ s3 = (int)x3;

+ s4 = (int)(cospi_8_64 * x4 + cospi_24_64 * x5);

+ s5 = (int)(cospi_24_64 * x4 - cospi_8_64 * x5);

+ s6 = (int)(-cospi_24_64 * x6 + cospi_8_64 * x7);

+ s7 = (int)(cospi_8_64 * x6 + cospi_24_64 * x7);

x0 = WRAPLOW(s0 + s2, 8);

x1 = WRAPLOW(s1 + s3, 8);

@@ -405,10 +400,10 @@

x7 = WRAPLOW(dct_const_round_shift(s5 - s7), 8);

// stage 3

- s2 = cospi_16_64 * (x2 + x3);

- s3 = cospi_16_64 * (x2 - x3);

- s6 = cospi_16_64 * (x6 + x7);

- s7 = cospi_16_64 * (x6 - x7);

+ s2 = (int)(cospi_16_64 * (x2 + x3));

+ s3 = (int)(cospi_16_64 * (x2 - x3));

+ s6 = (int)(cospi_16_64 * (x6 + x7));

+ s7 = (int)(cospi_16_64 * (x6 - x7));

x2 = WRAPLOW(dct_const_round_shift(s2), 8);

x3 = WRAPLOW(dct_const_round_shift(s3), 8);

@@ -1702,10 +1697,10 @@

static void highbd_iadst4(const tran_low_t *input, tran_low_t *output, int bd) {

tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;

- tran_high_t x0 = input[0];

- tran_high_t x1 = input[1];

- tran_high_t x2 = input[2];

- tran_high_t x3 = input[3];

+ tran_low_t x0 = input[0];

+ tran_low_t x1 = input[1];

+ tran_low_t x2 = input[2];

+ tran_low_t x3 = input[3];

(void) bd;

if (!(x0 | x1 | x2 | x3)) {

@@ -1720,26 +1715,21 @@

s4 = sinpi_1_9 * x2;

s5 = sinpi_2_9 * x3;

s6 = sinpi_4_9 * x3;

- s7 = x0 - x2 + x3;

+ s7 = (tran_high_t)(x0 - x2 + x3);

- x0 = s0 + s3 + s5;

- x1 = s1 - s4 - s6;

- x2 = sinpi_3_9 * s7;

- x3 = s2;

+ s0 = s0 + s3 + s5;

+ s1 = s1 - s4 - s6;

+ s3 = s2;

+ s2 = sinpi_3_9 * s7;

- s0 = x0 + x3;

- s1 = x1 + x3;

- s2 = x2;

- s3 = x0 + x1 - x3;

// 1-D transform scaling factor is sqrt(2).

// The overall dynamic range is 14b (input) + 14b (multiplication scaling)

// + 1b (addition) = 29b.

// Hence the output bit depth is 15b.

- output[0] = WRAPLOW(dct_const_round_shift(s0), bd);

- output[1] = WRAPLOW(dct_const_round_shift(s1), bd);

+ output[0] = WRAPLOW(dct_const_round_shift(s0 + s3), bd);

+ output[1] = WRAPLOW(dct_const_round_shift(s1 + s3), bd);

output[2] = WRAPLOW(dct_const_round_shift(s2), bd);

- output[3] = WRAPLOW(dct_const_round_shift(s3), bd);

+ output[3] = WRAPLOW(dct_const_round_shift(s0 + s1 - s3), bd);

}

void vp9_highbd_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,

@@ -1779,14 +1769,14 @@

static void highbd_iadst8(const tran_low_t *input, tran_low_t *output, int bd) {

tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;

- tran_high_t x0 = input[7];

- tran_high_t x1 = input[0];

- tran_high_t x2 = input[5];

- tran_high_t x3 = input[2];

- tran_high_t x4 = input[3];

- tran_high_t x5 = input[4];

- tran_high_t x6 = input[1];

- tran_high_t x7 = input[6];

+ tran_low_t x0 = input[7];

+ tran_low_t x1 = input[0];

+ tran_low_t x2 = input[5];

+ tran_low_t x3 = input[2];

+ tran_low_t x4 = input[3];

+ tran_low_t x5 = input[4];

+ tran_low_t x6 = input[1];

+ tran_low_t x7 = input[6];

(void) bd;

if (!(x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7)) {

@@ -2113,22 +2103,22 @@

tran_high_t s0, s1, s2, s3, s4, s5, s6, s7, s8;

tran_high_t s9, s10, s11, s12, s13, s14, s15;

- tran_high_t x0 = input[15];

- tran_high_t x1 = input[0];

- tran_high_t x2 = input[13];

- tran_high_t x3 = input[2];

- tran_high_t x4 = input[11];

- tran_high_t x5 = input[4];

- tran_high_t x6 = input[9];

- tran_high_t x7 = input[6];

- tran_high_t x8 = input[7];

- tran_high_t x9 = input[8];

- tran_high_t x10 = input[5];

- tran_high_t x11 = input[10];

- tran_high_t x12 = input[3];

- tran_high_t x13 = input[12];

- tran_high_t x14 = input[1];

- tran_high_t x15 = input[14];

+ tran_low_t x0 = input[15];

+ tran_low_t x1 = input[0];

+ tran_low_t x2 = input[13];

+ tran_low_t x3 = input[2];

+ tran_low_t x4 = input[11];

+ tran_low_t x5 = input[4];

+ tran_low_t x6 = input[9];

+ tran_low_t x7 = input[6];

+ tran_low_t x8 = input[7];

+ tran_low_t x9 = input[8];

+ tran_low_t x10 = input[5];

+ tran_low_t x11 = input[10];

+ tran_low_t x12 = input[3];

+ tran_low_t x13 = input[12];

+ tran_low_t x14 = input[1];

+ tran_low_t x15 = input[14];

(void) bd;

if (!(x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7 | x8

« no previous file with comments | « source/libvpx/vp9/common/vp9_idct.h ('k') | source/libvpx/vp9/common/vp9_loopfilter.h » ('j') | no next file with comments »