Remove two more LPF macros Change-Id: I60278e399f4f65aa63526e459947e88084f0e889
diff --git a/aom_dsp/loopfilter.c b/aom_dsp/loopfilter.c index 252908b..a4ae230 100644 --- a/aom_dsp/loopfilter.c +++ b/aom_dsp/loopfilter.c
@@ -20,14 +20,6 @@ return (int8_t)clamp(t, -128, 127); } -#if CONFIG_DEBLOCK_13TAP -#define PARALLEL_DEBLOCKING_13_TAP 1 -#define PARALLEL_DEBLOCKING_5_TAP_CHROMA 1 -#else -#define PARALLEL_DEBLOCKING_13_TAP 0 -#define PARALLEL_DEBLOCKING_5_TAP_CHROMA 0 -#endif - static INLINE int16_t signed_char_clamp_high(int t, int bd) { switch (bd) { case 10: return (int16_t)clamp(t, -128 * 4, 128 * 4 - 1); @@ -61,7 +53,6 @@ return ~mask; } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA static INLINE int8_t filter_mask3_chroma(uint8_t limit, uint8_t blimit, uint8_t p2, uint8_t p1, uint8_t p0, uint8_t q0, uint8_t q1, uint8_t q2) { @@ -84,7 +75,6 @@ mask |= (abs(q2 - q0) > thresh) * -1; return ~mask; } -#endif static INLINE int8_t flat_mask4(uint8_t thresh, uint8_t p3, uint8_t p2, uint8_t p1, uint8_t p0, uint8_t q0, uint8_t q1, @@ -99,18 +89,6 @@ return ~mask; } -#if !PARALLEL_DEBLOCKING_13_TAP -static INLINE int8_t flat_mask5(uint8_t thresh, uint8_t p4, uint8_t p3, - uint8_t p2, uint8_t p1, uint8_t p0, uint8_t q0, - uint8_t q1, uint8_t q2, uint8_t q3, - uint8_t q4) { - int8_t mask = ~flat_mask4(thresh, p3, p2, p1, p0, q0, q1, q2, q3); - mask |= (abs(p4 - p0) > thresh) * -1; - mask |= (abs(q4 - q0) > thresh) * -1; - return ~mask; -} -#endif - // is there high edge variance internal edge: 11111111 yes, 00000000 no static INLINE int8_t hev_mask(uint8_t thresh, uint8_t p1, uint8_t p0, uint8_t q0, uint8_t q1) { @@ -201,7 +179,6 @@ aom_lpf_vertical_4_c(s + 4 * pitch, pitch, blimit1, limit1, thresh1); } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA static INLINE void filter6(int8_t mask, uint8_t thresh, int8_t flat, uint8_t *op2, uint8_t *op1, uint8_t *op0, uint8_t *oq0, uint8_t *oq1, uint8_t *oq2) { @@ -218,7 +195,6 @@ filter4(mask, thresh, op1, op0, oq0, oq1); } } -#endif static INLINE void filter8(int8_t mask, uint8_t thresh, int8_t flat, uint8_t *op3, uint8_t *op2, uint8_t *op1, @@ -240,7 +216,6 @@ } } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA void aom_lpf_horizontal_6_c(uint8_t *s, int p, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh) { int i; @@ -260,7 +235,6 @@ ++s; } } -#endif void aom_lpf_horizontal_8_c(uint8_t *s, int p, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh) { @@ -290,7 +264,6 @@ aom_lpf_horizontal_8_c(s + 4, p, blimit1, limit1, thresh1); } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA void aom_lpf_vertical_6_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh) { int i; @@ -306,7 +279,6 @@ s += pitch; } } -#endif void aom_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh) { @@ -333,7 +305,6 @@ aom_lpf_vertical_8_c(s + 4 * pitch, pitch, blimit1, limit1, thresh1); } -#if PARALLEL_DEBLOCKING_13_TAP static INLINE void filter14(int8_t mask, uint8_t thresh, int8_t flat, int8_t flat2, uint8_t *op6, uint8_t *op5, uint8_t *op4, uint8_t *op3, uint8_t *op2, @@ -381,63 +352,6 @@ filter8(mask, thresh, flat, op3, op2, op1, op0, oq0, oq1, oq2, oq3); } } -#endif - -#if !PARALLEL_DEBLOCKING_13_TAP -static INLINE void filter16(int8_t mask, uint8_t thresh, int8_t flat, - int8_t flat2, uint8_t *op7, uint8_t *op6, - uint8_t *op5, uint8_t *op4, uint8_t *op3, - uint8_t *op2, uint8_t *op1, uint8_t *op0, - uint8_t *oq0, uint8_t *oq1, uint8_t *oq2, - uint8_t *oq3, uint8_t *oq4, uint8_t *oq5, - uint8_t *oq6, uint8_t *oq7) { - if (flat2 && flat && mask) { - const uint8_t p7 = *op7, p6 = *op6, p5 = *op5, p4 = *op4, p3 = *op3, - p2 = *op2, p1 = *op1, p0 = *op0; - - const uint8_t q0 = *oq0, q1 = *oq1, q2 = *oq2, q3 = *oq3, q4 = *oq4, - q5 = *oq5, q6 = *oq6, q7 = *oq7; - - // 15-tap filter [1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1] - *op6 = ROUND_POWER_OF_TWO( - p7 * 7 + p6 * 2 + p5 + p4 + p3 + p2 + p1 + p0 + q0, 4); - *op5 = ROUND_POWER_OF_TWO( - p7 * 6 + p6 + p5 * 2 + p4 + p3 + p2 + p1 + p0 + q0 + q1, 4); - *op4 = ROUND_POWER_OF_TWO( - p7 * 5 + p6 + p5 + p4 * 2 + p3 + p2 + p1 + p0 + q0 + q1 + q2, 4); - *op3 = ROUND_POWER_OF_TWO( - p7 * 4 + p6 + p5 + p4 + p3 * 2 + p2 + p1 + p0 + q0 + q1 + q2 + q3, 4); - *op2 = ROUND_POWER_OF_TWO( - p7 * 3 + p6 + p5 + p4 + p3 + p2 * 2 + p1 + p0 + q0 + q1 + q2 + q3 + q4, - 4); - *op1 = ROUND_POWER_OF_TWO(p7 * 2 + p6 + p5 + p4 + p3 + p2 + p1 * 2 + p0 + - q0 + q1 + q2 + q3 + q4 + q5, - 4); - *op0 = ROUND_POWER_OF_TWO(p7 + p6 + p5 + p4 + p3 + p2 + p1 + p0 * 2 + q0 + - q1 + q2 + q3 + q4 + q5 + q6, - 4); - *oq0 = ROUND_POWER_OF_TWO(p6 + p5 + p4 + p3 + p2 + p1 + p0 + q0 * 2 + q1 + - q2 + q3 + q4 + q5 + q6 + q7, - 4); - *oq1 = ROUND_POWER_OF_TWO(p5 + p4 + p3 + p2 + p1 + p0 + q0 + q1 * 2 + q2 + - q3 + q4 + q5 + q6 + q7 * 2, - 4); - *oq2 = ROUND_POWER_OF_TWO( - p4 + p3 + p2 + p1 + p0 + q0 + q1 + q2 * 2 + q3 + q4 + q5 + q6 + q7 * 3, - 4); - *oq3 = ROUND_POWER_OF_TWO( - p3 + p2 + p1 + p0 + q0 + q1 + q2 + q3 * 2 + q4 + q5 + q6 + q7 * 4, 4); - *oq4 = ROUND_POWER_OF_TWO( - p2 + p1 + p0 + q0 + q1 + q2 + q3 + q4 * 2 + q5 + q6 + q7 * 5, 4); - *oq5 = ROUND_POWER_OF_TWO( - p1 + p0 + q0 + q1 + q2 + q3 + q4 + q5 * 2 + q6 + q7 * 6, 4); - *oq6 = ROUND_POWER_OF_TWO( - p0 + q0 + q1 + q2 + q3 + q4 + q5 + q6 * 2 + q7 * 7, 4); - } else { - filter8(mask, thresh, flat, op3, op2, op1, op0, oq0, oq1, oq2, oq3); - } -} -#endif static void mb_lpf_horizontal_edge_w(uint8_t *s, int p, const uint8_t *blimit, const uint8_t *limit, @@ -457,7 +371,6 @@ filter_mask(*limit, *blimit, p3, p2, p1, p0, q0, q1, q2, q3); const int8_t flat = flat_mask4(1, p3, p2, p1, p0, q0, q1, q2, q3); -#if PARALLEL_DEBLOCKING_13_TAP (void)p7; (void)q7; const int8_t flat2 = flat_mask4(1, p6, p5, p4, p0, q0, q4, q5, q6); @@ -465,15 +378,6 @@ filter14(mask, *thresh, flat, flat2, s - 7 * p, s - 6 * p, s - 5 * p, s - 4 * p, s - 3 * p, s - 2 * p, s - 1 * p, s, s + 1 * p, s + 2 * p, s + 3 * p, s + 4 * p, s + 5 * p, s + 6 * p); -#else - const int8_t flat2 = flat_mask5(1, p7, p6, p5, p4, p0, q0, q4, q5, q6, q7); - - filter16(mask, *thresh, flat, flat2, s - 8 * p, s - 7 * p, s - 6 * p, - s - 5 * p, s - 4 * p, s - 3 * p, s - 2 * p, s - 1 * p, s, - s + 1 * p, s + 2 * p, s + 3 * p, s + 4 * p, s + 5 * p, s + 6 * p, - s + 7 * p); -#endif - ++s; } } @@ -502,21 +406,12 @@ filter_mask(*limit, *blimit, p3, p2, p1, p0, q0, q1, q2, q3); const int8_t flat = flat_mask4(1, p3, p2, p1, p0, q0, q1, q2, q3); -#if PARALLEL_DEBLOCKING_13_TAP (void)p7; (void)q7; const int8_t flat2 = flat_mask4(1, p6, p5, p4, p0, q0, q4, q5, q6); filter14(mask, *thresh, flat, flat2, s - 7, s - 6, s - 5, s - 4, s - 3, s - 2, s - 1, s, s + 1, s + 2, s + 3, s + 4, s + 5, s + 6); -#else - const int8_t flat2 = flat_mask5(1, p7, p6, p5, p4, p0, q0, q4, q5, q6, q7); - - filter16(mask, *thresh, flat, flat2, s - 8, s - 7, s - 6, s - 5, s - 4, - s - 3, s - 2, s - 1, s, s + 1, s + 2, s + 3, s + 4, s + 5, s + 6, - s + 7); -#endif - s += p; } } @@ -562,7 +457,6 @@ return ~mask; } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA static INLINE int8_t highbd_filter_mask3_chroma(uint8_t limit, uint8_t blimit, uint16_t p2, uint16_t p1, uint16_t p0, uint16_t q0, @@ -591,7 +485,6 @@ mask |= (abs(q2 - q0) > thresh16) * -1; return ~mask; } -#endif static INLINE int8_t highbd_flat_mask4(uint8_t thresh, uint16_t p3, uint16_t p2, uint16_t p1, uint16_t p0, uint16_t q0, @@ -608,19 +501,6 @@ return ~mask; } -#if !PARALLEL_DEBLOCKING_13_TAP -static INLINE int8_t highbd_flat_mask5(uint8_t thresh, uint16_t p4, uint16_t p3, - uint16_t p2, uint16_t p1, uint16_t p0, - uint16_t q0, uint16_t q1, uint16_t q2, - uint16_t q3, uint16_t q4, int bd) { - int8_t mask = ~highbd_flat_mask4(thresh, p3, p2, p1, p0, q0, q1, q2, q3, bd); - int16_t thresh16 = (uint16_t)thresh << (bd - 8); - mask |= (abs(p4 - p0) > thresh16) * -1; - mask |= (abs(q4 - q0) > thresh16) * -1; - return ~mask; -} -#endif - // Is there high edge variance internal edge: // 11111111_11111111 yes, 00000000_00000000 no ? static INLINE int16_t highbd_hev_mask(uint8_t thresh, uint16_t p1, uint16_t p0, @@ -722,7 +602,6 @@ bd); } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA static INLINE void highbd_filter6(int8_t mask, uint8_t thresh, int8_t flat, uint16_t *op2, uint16_t *op1, uint16_t *op0, uint16_t *oq0, uint16_t *oq1, uint16_t *oq2, @@ -740,7 +619,6 @@ highbd_filter4(mask, thresh, op1, op0, oq0, oq1, bd); } } -#endif static INLINE void highbd_filter8(int8_t mask, uint8_t thresh, int8_t flat, uint16_t *op3, uint16_t *op2, uint16_t *op1, @@ -784,7 +662,6 @@ } } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA void aom_highbd_lpf_horizontal_6_c(uint16_t *s, int p, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd) { @@ -805,7 +682,6 @@ ++s; } } -#endif void aom_highbd_lpf_horizontal_8_dual_c( uint16_t *s, int p, const uint8_t *blimit0, const uint8_t *limit0, @@ -815,7 +691,6 @@ aom_highbd_lpf_horizontal_8_c(s + 4, p, blimit1, limit1, thresh1, bd); } -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA void aom_highbd_lpf_vertical_6_c(uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd) { @@ -833,7 +708,6 @@ s += pitch; } } -#endif void aom_highbd_lpf_vertical_8_c(uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, @@ -863,7 +737,6 @@ bd); } -#if PARALLEL_DEBLOCKING_13_TAP static INLINE void highbd_filter14(int8_t mask, uint8_t thresh, int8_t flat, int8_t flat2, uint16_t *op6, uint16_t *op5, uint16_t *op4, uint16_t *op3, uint16_t *op2, @@ -923,75 +796,6 @@ bd); } } -#endif - -#if !PARALLEL_DEBLOCKING_13_TAP -static INLINE void highbd_filter16(int8_t mask, uint8_t thresh, int8_t flat, - int8_t flat2, uint16_t *op7, uint16_t *op6, - uint16_t *op5, uint16_t *op4, uint16_t *op3, - uint16_t *op2, uint16_t *op1, uint16_t *op0, - uint16_t *oq0, uint16_t *oq1, uint16_t *oq2, - uint16_t *oq3, uint16_t *oq4, uint16_t *oq5, - uint16_t *oq6, uint16_t *oq7, int bd) { - if (flat2 && flat && mask) { - const uint16_t p7 = *op7; - const uint16_t p6 = *op6; - const uint16_t p5 = *op5; - const uint16_t p4 = *op4; - const uint16_t p3 = *op3; - const uint16_t p2 = *op2; - const uint16_t p1 = *op1; - const uint16_t p0 = *op0; - const uint16_t q0 = *oq0; - const uint16_t q1 = *oq1; - const uint16_t q2 = *oq2; - const uint16_t q3 = *oq3; - const uint16_t q4 = *oq4; - const uint16_t q5 = *oq5; - const uint16_t q6 = *oq6; - const uint16_t q7 = *oq7; - - // 15-tap filter [1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1] - *op6 = ROUND_POWER_OF_TWO( - p7 * 7 + p6 * 2 + p5 + p4 + p3 + p2 + p1 + p0 + q0, 4); - *op5 = ROUND_POWER_OF_TWO( - p7 * 6 + p6 + p5 * 2 + p4 + p3 + p2 + p1 + p0 + q0 + q1, 4); - *op4 = ROUND_POWER_OF_TWO( - p7 * 5 + p6 + p5 + p4 * 2 + p3 + p2 + p1 + p0 + q0 + q1 + q2, 4); - *op3 = ROUND_POWER_OF_TWO( - p7 * 4 + p6 + p5 + p4 + p3 * 2 + p2 + p1 + p0 + q0 + q1 + q2 + q3, 4); - *op2 = ROUND_POWER_OF_TWO( - p7 * 3 + p6 + p5 + p4 + p3 + p2 * 2 + p1 + p0 + q0 + q1 + q2 + q3 + q4, - 4); - *op1 = ROUND_POWER_OF_TWO(p7 * 2 + p6 + p5 + p4 + p3 + p2 + p1 * 2 + p0 + - q0 + q1 + q2 + q3 + q4 + q5, - 4); - *op0 = ROUND_POWER_OF_TWO(p7 + p6 + p5 + p4 + p3 + p2 + p1 + p0 * 2 + q0 + - q1 + q2 + q3 + q4 + q5 + q6, - 4); - *oq0 = ROUND_POWER_OF_TWO(p6 + p5 + p4 + p3 + p2 + p1 + p0 + q0 * 2 + q1 + - q2 + q3 + q4 + q5 + q6 + q7, - 4); - *oq1 = ROUND_POWER_OF_TWO(p5 + p4 + p3 + p2 + p1 + p0 + q0 + q1 * 2 + q2 + - q3 + q4 + q5 + q6 + q7 * 2, - 4); - *oq2 = ROUND_POWER_OF_TWO( - p4 + p3 + p2 + p1 + p0 + q0 + q1 + q2 * 2 + q3 + q4 + q5 + q6 + q7 * 3, - 4); - *oq3 = ROUND_POWER_OF_TWO( - p3 + p2 + p1 + p0 + q0 + q1 + q2 + q3 * 2 + q4 + q5 + q6 + q7 * 4, 4); - *oq4 = ROUND_POWER_OF_TWO( - p2 + p1 + p0 + q0 + q1 + q2 + q3 + q4 * 2 + q5 + q6 + q7 * 5, 4); - *oq5 = ROUND_POWER_OF_TWO( - p1 + p0 + q0 + q1 + q2 + q3 + q4 + q5 * 2 + q6 + q7 * 6, 4); - *oq6 = ROUND_POWER_OF_TWO( - p0 + q0 + q1 + q2 + q3 + q4 + q5 + q6 * 2 + q7 * 7, 4); - } else { - highbd_filter8(mask, thresh, flat, op3, op2, op1, op0, oq0, oq1, oq2, oq3, - bd); - } -} -#endif static void highbd_mb_lpf_horizontal_edge_w(uint16_t *s, int p, const uint8_t *blimit, @@ -1017,7 +821,6 @@ const int8_t flat = highbd_flat_mask4(1, p3, p2, p1, p0, q0, q1, q2, q3, bd); -#if PARALLEL_DEBLOCKING_13_TAP const int8_t flat2 = highbd_flat_mask4(1, s[-7 * p], s[-6 * p], s[-5 * p], p0, q0, s[4 * p], s[5 * p], s[6 * p], bd); @@ -1025,16 +828,6 @@ highbd_filter14(mask, *thresh, flat, flat2, s - 7 * p, s - 6 * p, s - 5 * p, s - 4 * p, s - 3 * p, s - 2 * p, s - 1 * p, s, s + 1 * p, s + 2 * p, s + 3 * p, s + 4 * p, s + 5 * p, s + 6 * p, bd); -#else - const int8_t flat2 = - highbd_flat_mask5(1, s[-8 * p], s[-7 * p], s[-6 * p], s[-5 * p], p0, q0, - s[4 * p], s[5 * p], s[6 * p], s[7 * p], bd); - - highbd_filter16(mask, *thresh, flat, flat2, s - 8 * p, s - 7 * p, s - 6 * p, - s - 5 * p, s - 4 * p, s - 3 * p, s - 2 * p, s - 1 * p, s, - s + 1 * p, s + 2 * p, s + 3 * p, s + 4 * p, s + 5 * p, - s + 6 * p, s + 7 * p, bd); -#endif ++s; } } @@ -1072,21 +865,12 @@ highbd_filter_mask(*limit, *blimit, p3, p2, p1, p0, q0, q1, q2, q3, bd); const int8_t flat = highbd_flat_mask4(1, p3, p2, p1, p0, q0, q1, q2, q3, bd); -#if PARALLEL_DEBLOCKING_13_TAP const int8_t flat2 = highbd_flat_mask4(1, s[-7], s[-6], s[-5], p0, q0, s[4], s[5], s[6], bd); highbd_filter14(mask, *thresh, flat, flat2, s - 7, s - 6, s - 5, s - 4, s - 3, s - 2, s - 1, s, s + 1, s + 2, s + 3, s + 4, s + 5, s + 6, bd); -#else - const int8_t flat2 = highbd_flat_mask5(1, s[-8], s[-7], s[-6], s[-5], p0, - q0, s[4], s[5], s[6], s[7], bd); - - highbd_filter16(mask, *thresh, flat, flat2, s - 8, s - 7, s - 6, s - 5, - s - 4, s - 3, s - 2, s - 1, s, s + 1, s + 2, s + 3, s + 4, - s + 5, s + 6, s + 7, bd); -#endif s += p; } }
diff --git a/aom_dsp/x86/highbd_loopfilter_sse2.c b/aom_dsp/x86/highbd_loopfilter_sse2.c index 35f2c35..ecbee3f 100644 --- a/aom_dsp/x86/highbd_loopfilter_sse2.c +++ b/aom_dsp/x86/highbd_loopfilter_sse2.c
@@ -130,7 +130,6 @@ flat_mask_internal(th, p, q, bd, 1, 4, flat); } -#if CONFIG_DEBLOCK_13TAP // Note: // access p[6-4], p[0], and q[6-4], q[0] static INLINE void highbd_flat_mask4_13(const __m128i *th, const __m128i *p, @@ -138,7 +137,6 @@ int bd) { flat_mask_internal(th, p, q, bd, 4, 7, flat); } -#endif // Note: // access p[7-4], p[0], and q[7-4], q[0] @@ -224,13 +222,8 @@ __m128i blimit, limit, thresh; get_limit(blt, lt, thr, bd, &blimit, &limit, &thresh); -#if CONFIG_DEBLOCK_13TAP __m128i p[7], q[7]; load_highbd_pixel(s, 7, pitch, p, q); -#else - __m128i p[8], q[8]; - load_highbd_pixel(s, 8, pitch, p, q); -#endif __m128i mask; highbd_filter_mask(p, q, &limit, &blimit, &mask); @@ -238,12 +231,7 @@ __m128i flat, flat2; const __m128i one = _mm_set1_epi16(1); highbd_flat_mask4(&one, p, q, &flat, bd); - -#if CONFIG_DEBLOCK_13TAP highbd_flat_mask4_13(&one, p, q, &flat2, bd); -#else - highbd_flat_mask5(&one, p, q, &flat2, bd); -#endif flat = _mm_and_si128(flat, mask); flat2 = _mm_and_si128(flat2, flat); @@ -253,16 +241,11 @@ // flat and wide flat calculations __m128i flat_p[3], flat_q[3]; -#if CONFIG_DEBLOCK_13TAP __m128i flat2_p[6], flat2_q[6]; -#else - __m128i flat2_p[7], flat2_q[7]; -#endif { const __m128i eight = _mm_set1_epi16(8); const __m128i four = _mm_set1_epi16(4); -#if CONFIG_DEBLOCK_13TAP __m128i sum_p = _mm_add_epi16(p[5], _mm_add_epi16(p[4], p[3])); __m128i sum_q = _mm_add_epi16(q[5], _mm_add_epi16(q[4], q[3])); @@ -379,79 +362,6 @@ sum_q, _mm_add_epi16( sum_q6, _mm_add_epi16(q[5], _mm_add_epi16(q[4], q[6])))), 4); -#else - __m128i sum_p = - _mm_add_epi16(_mm_add_epi16(p[6], p[5]), _mm_add_epi16(p[4], p[3])); - __m128i sum_q = - _mm_add_epi16(_mm_add_epi16(q[6], q[5]), _mm_add_epi16(q[4], q[3])); - - __m128i sum_lp = _mm_add_epi16(p[0], _mm_add_epi16(p[2], p[1])); - sum_p = _mm_add_epi16(sum_p, sum_lp); - - __m128i sum_lq = _mm_add_epi16(q[0], _mm_add_epi16(q[2], q[1])); - sum_q = _mm_add_epi16(sum_q, sum_lq); - sum_p = _mm_add_epi16(eight, _mm_add_epi16(sum_p, sum_q)); - sum_lp = _mm_add_epi16(four, _mm_add_epi16(sum_lp, sum_lq)); - - flat2_p[0] = - _mm_srli_epi16(_mm_add_epi16(sum_p, _mm_add_epi16(p[7], p[0])), 4); - flat2_q[0] = - _mm_srli_epi16(_mm_add_epi16(sum_p, _mm_add_epi16(q[7], q[0])), 4); - flat_p[0] = - _mm_srli_epi16(_mm_add_epi16(sum_lp, _mm_add_epi16(p[3], p[0])), 3); - flat_q[0] = - _mm_srli_epi16(_mm_add_epi16(sum_lp, _mm_add_epi16(q[3], q[0])), 3); - - __m128i sum_p7 = _mm_add_epi16(p[7], p[7]); - __m128i sum_q7 = _mm_add_epi16(q[7], q[7]); - __m128i sum_p3 = _mm_add_epi16(p[3], p[3]); - __m128i sum_q3 = _mm_add_epi16(q[3], q[3]); - - sum_q = _mm_sub_epi16(sum_p, p[6]); - sum_p = _mm_sub_epi16(sum_p, q[6]); - flat2_p[1] = - _mm_srli_epi16(_mm_add_epi16(sum_p, _mm_add_epi16(sum_p7, p[1])), 4); - flat2_q[1] = - _mm_srli_epi16(_mm_add_epi16(sum_q, _mm_add_epi16(sum_q7, q[1])), 4); - - sum_lq = _mm_sub_epi16(sum_lp, p[2]); - sum_lp = _mm_sub_epi16(sum_lp, q[2]); - flat_p[1] = - _mm_srli_epi16(_mm_add_epi16(sum_lp, _mm_add_epi16(sum_p3, p[1])), 3); - flat_q[1] = - _mm_srli_epi16(_mm_add_epi16(sum_lq, _mm_add_epi16(sum_q3, q[1])), 3); - - sum_p7 = _mm_add_epi16(sum_p7, p[7]); - sum_q7 = _mm_add_epi16(sum_q7, q[7]); - sum_p3 = _mm_add_epi16(sum_p3, p[3]); - sum_q3 = _mm_add_epi16(sum_q3, q[3]); - - sum_p = _mm_sub_epi16(sum_p, q[5]); - sum_q = _mm_sub_epi16(sum_q, p[5]); - flat2_p[2] = - _mm_srli_epi16(_mm_add_epi16(sum_p, _mm_add_epi16(sum_p7, p[2])), 4); - flat2_q[2] = - _mm_srli_epi16(_mm_add_epi16(sum_q, _mm_add_epi16(sum_q7, q[2])), 4); - - sum_lp = _mm_sub_epi16(sum_lp, q[1]); - sum_lq = _mm_sub_epi16(sum_lq, p[1]); - flat_p[2] = - _mm_srli_epi16(_mm_add_epi16(sum_lp, _mm_add_epi16(sum_p3, p[2])), 3); - flat_q[2] = - _mm_srli_epi16(_mm_add_epi16(sum_lq, _mm_add_epi16(sum_q3, q[2])), 3); - - int i; - for (i = 3; i < 7; ++i) { - sum_p7 = _mm_add_epi16(sum_p7, p[7]); - sum_q7 = _mm_add_epi16(sum_q7, q[7]); - sum_p = _mm_sub_epi16(sum_p, q[7 - i]); - sum_q = _mm_sub_epi16(sum_q, p[7 - i]); - flat2_p[i] = - _mm_srli_epi16(_mm_add_epi16(sum_p, _mm_add_epi16(sum_p7, p[i])), 4); - flat2_q[i] = - _mm_srli_epi16(_mm_add_epi16(sum_q, _mm_add_epi16(sum_q7, q[i])), 4); - } -#endif } // highbd_filter8 @@ -474,12 +384,8 @@ q[i] = _mm_or_si128(qs[i], flat_q[i]); } -// highbd_filter16 -#if CONFIG_DEBLOCK_13TAP + // highbd_filter16 for (i = 5; i >= 0; i--) { -#else - for (i = 6; i >= 0; i--) { -#endif // p[i] remains unchanged if !(flat2 && flat && mask) p[i] = _mm_andnot_si128(flat2, p[i]); flat2_p[i] = _mm_and_si128(flat2, flat2_p[i]);
diff --git a/aom_dsp/x86/loopfilter_sse2.c b/aom_dsp/x86/loopfilter_sse2.c index 64f7756..a0e7ded 100644 --- a/aom_dsp/x86/loopfilter_sse2.c +++ b/aom_dsp/x86/loopfilter_sse2.c
@@ -251,9 +251,6 @@ const __m128i limit = _mm_load_si128((const __m128i *)_limit); const __m128i thresh = _mm_load_si128((const __m128i *)_thresh); __m128i mask, hev, flat, flat2; -#if !CONFIG_DEBLOCK_13TAP - __m128i q7p7; -#endif __m128i q6p6, q5p5, q4p4, q3p3, q2p2, q1p1, q0p0, p0q0, p1q1; __m128i abs_p1p0; @@ -316,9 +313,6 @@ __m128i filt; __m128i work_a; __m128i filter1, filter2; -#if !CONFIG_DEBLOCK_13TAP - __m128i flat2_q6p6; -#endif __m128i flat2_q5p5, flat2_q4p4, flat2_q3p3, flat2_q2p2; __m128i flat2_q1p1, flat2_q0p0, flat_q2p2, flat_q1p1, flat_q0p0; @@ -368,14 +362,7 @@ q6p6 = _mm_castps_si128( _mm_loadh_pi(_mm_castsi128_ps(q6p6), (__m64 *)(s + 6 * p))); flat2 = _mm_max_epu8(abs_diff(q4p4, q0p0), abs_diff(q5p5, q0p0)); -#if !CONFIG_DEBLOCK_13TAP - q7p7 = _mm_loadl_epi64((__m128i *)(s - 8 * p)); - q7p7 = _mm_castps_si128( - _mm_loadh_pi(_mm_castsi128_ps(q7p7), (__m64 *)(s + 7 * p))); - work = _mm_max_epu8(abs_diff(q6p6, q0p0), abs_diff(q7p7, q0p0)); -#else work = abs_diff(q6p6, q0p0); -#endif flat2 = _mm_max_epu8(work, flat2); flat2 = _mm_max_epu8(flat2, _mm_srli_si128(flat2, 8)); flat2 = _mm_subs_epu8(flat2, one); @@ -387,23 +374,13 @@ { const __m128i eight = _mm_set1_epi16(8); const __m128i four = _mm_set1_epi16(4); -#if !CONFIG_DEBLOCK_13TAP - __m128i p7_16, q7_16; -#endif __m128i p6_16, p5_16, p4_16, p3_16, p2_16, p1_16, p0_16; __m128i q6_16, q5_16, q4_16, q3_16, q2_16, q1_16, q0_16; __m128i pixelFilter_p, pixelFilter_q; __m128i pixetFilter_p2p1p0, pixetFilter_q2q1q0; -#if !CONFIG_DEBLOCK_13TAP - __m128i sum_p7, sum_q7; -#else __m128i sum_p6, sum_q6; -#endif __m128i sum_p3, sum_q3, res_p, res_q; -#if !CONFIG_DEBLOCK_13TAP - p7_16 = _mm_unpacklo_epi8(q7p7, zero); -#endif p6_16 = _mm_unpacklo_epi8(q6p6, zero); p5_16 = _mm_unpacklo_epi8(q5p5, zero); p4_16 = _mm_unpacklo_epi8(q4p4, zero); @@ -418,17 +395,8 @@ q4_16 = _mm_unpackhi_epi8(q4p4, zero); q5_16 = _mm_unpackhi_epi8(q5p5, zero); q6_16 = _mm_unpackhi_epi8(q6p6, zero); -#if !CONFIG_DEBLOCK_13TAP - q7_16 = _mm_unpackhi_epi8(q7p7, zero); - - pixelFilter_p = _mm_add_epi16(_mm_add_epi16(p6_16, p5_16), - _mm_add_epi16(p4_16, p3_16)); - pixelFilter_q = _mm_add_epi16(_mm_add_epi16(q6_16, q5_16), - _mm_add_epi16(q4_16, q3_16)); -#else pixelFilter_p = _mm_add_epi16(p5_16, _mm_add_epi16(p4_16, p3_16)); pixelFilter_q = _mm_add_epi16(q5_16, _mm_add_epi16(q4_16, q3_16)); -#endif pixetFilter_p2p1p0 = _mm_add_epi16(p0_16, _mm_add_epi16(p2_16, p1_16)); pixelFilter_p = _mm_add_epi16(pixelFilter_p, pixetFilter_p2p1p0); @@ -440,23 +408,15 @@ pixetFilter_p2p1p0 = _mm_add_epi16( four, _mm_add_epi16(pixetFilter_p2p1p0, pixetFilter_q2q1q0)); res_p = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(p7_16, p0_16)), 4); -#else _mm_add_epi16(pixelFilter_p, _mm_add_epi16(_mm_add_epi16(p6_16, p0_16), _mm_add_epi16(p1_16, q0_16))), 4); -#endif res_q = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(q7_16, q0_16)), 4); -#else _mm_add_epi16(pixelFilter_p, _mm_add_epi16(_mm_add_epi16(q6_16, q0_16), _mm_add_epi16(p0_16, q1_16))), 4); -#endif flat2_q0p0 = _mm_packus_epi16(res_p, res_q); res_p = _mm_srli_epi16( @@ -466,44 +426,26 @@ flat_q0p0 = _mm_packus_epi16(res_p, res_q); -#if !CONFIG_DEBLOCK_13TAP - sum_p7 = _mm_add_epi16(p7_16, p7_16); - sum_q7 = _mm_add_epi16(q7_16, q7_16); -#else sum_p6 = _mm_add_epi16(p6_16, p6_16); sum_q6 = _mm_add_epi16(q6_16, q6_16); -#endif sum_p3 = _mm_add_epi16(p3_16, p3_16); sum_q3 = _mm_add_epi16(q3_16, q3_16); -#if !CONFIG_DEBLOCK_13TAP - pixelFilter_q = _mm_sub_epi16(pixelFilter_p, p6_16); - pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q6_16); -#else pixelFilter_q = _mm_sub_epi16(pixelFilter_p, p5_16); pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q5_16); -#endif res_p = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(sum_p7, p1_16)), 4); -#else _mm_add_epi16( pixelFilter_p, _mm_add_epi16(sum_p6, _mm_add_epi16(p1_16, _mm_add_epi16(p2_16, p0_16)))), 4); -#endif res_q = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_q, _mm_add_epi16(sum_q7, q1_16)), 4); -#else _mm_add_epi16( pixelFilter_q, _mm_add_epi16(sum_q6, _mm_add_epi16(q1_16, _mm_add_epi16(q0_16, q2_16)))), 4); -#endif flat2_q1p1 = _mm_packus_epi16(res_p, res_q); pixetFilter_q2q1q0 = _mm_sub_epi16(pixetFilter_p2p1p0, p2_16); @@ -514,44 +456,26 @@ _mm_add_epi16(pixetFilter_q2q1q0, _mm_add_epi16(sum_q3, q1_16)), 3); flat_q1p1 = _mm_packus_epi16(res_p, res_q); -#if !CONFIG_DEBLOCK_13TAP - sum_p7 = _mm_add_epi16(sum_p7, p7_16); - sum_q7 = _mm_add_epi16(sum_q7, q7_16); -#else sum_p6 = _mm_add_epi16(sum_p6, p6_16); sum_q6 = _mm_add_epi16(sum_q6, q6_16); -#endif sum_p3 = _mm_add_epi16(sum_p3, p3_16); sum_q3 = _mm_add_epi16(sum_q3, q3_16); -#if !CONFIG_DEBLOCK_13TAP - pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q5_16); - pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p5_16); -#else pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q4_16); pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p4_16); -#endif res_p = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(sum_p7, p2_16)), 4); -#else _mm_add_epi16( pixelFilter_p, _mm_add_epi16(sum_p6, _mm_add_epi16(p2_16, _mm_add_epi16(p3_16, p1_16)))), 4); -#endif res_q = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_q, _mm_add_epi16(sum_q7, q2_16)), 4); -#else _mm_add_epi16( pixelFilter_q, _mm_add_epi16(sum_q6, _mm_add_epi16(q2_16, _mm_add_epi16(q1_16, q3_16)))), 4); -#endif flat2_q2p2 = _mm_packus_epi16(res_p, res_q); pixetFilter_p2p1p0 = _mm_sub_epi16(pixetFilter_p2p1p0, q1_16); @@ -563,131 +487,64 @@ _mm_add_epi16(pixetFilter_q2q1q0, _mm_add_epi16(sum_q3, q2_16)), 3); flat_q2p2 = _mm_packus_epi16(res_p, res_q); -#if !CONFIG_DEBLOCK_13TAP - sum_p7 = _mm_add_epi16(sum_p7, p7_16); - sum_q7 = _mm_add_epi16(sum_q7, q7_16); -#else sum_p6 = _mm_add_epi16(sum_p6, p6_16); sum_q6 = _mm_add_epi16(sum_q6, q6_16); -#endif -#if !CONFIG_DEBLOCK_13TAP - pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q4_16); - pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p4_16); -#else pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q3_16); pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p3_16); -#endif res_p = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(sum_p7, p3_16)), 4); -#else _mm_add_epi16( pixelFilter_p, _mm_add_epi16(sum_p6, _mm_add_epi16(p3_16, _mm_add_epi16(p4_16, p2_16)))), 4); -#endif res_q = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_q, _mm_add_epi16(sum_q7, q3_16)), 4); -#else _mm_add_epi16( pixelFilter_q, _mm_add_epi16(sum_q6, _mm_add_epi16(q3_16, _mm_add_epi16(q2_16, q4_16)))), 4); -#endif flat2_q3p3 = _mm_packus_epi16(res_p, res_q); -#if !CONFIG_DEBLOCK_13TAP - sum_p7 = _mm_add_epi16(sum_p7, p7_16); - sum_q7 = _mm_add_epi16(sum_q7, q7_16); -#else sum_p6 = _mm_add_epi16(sum_p6, p6_16); sum_q6 = _mm_add_epi16(sum_q6, q6_16); -#endif -#if !CONFIG_DEBLOCK_13TAP - pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q3_16); - pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p3_16); -#else pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q2_16); pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p2_16); -#endif res_p = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(sum_p7, p4_16)), 4); -#else _mm_add_epi16( pixelFilter_p, _mm_add_epi16(sum_p6, _mm_add_epi16(p4_16, _mm_add_epi16(p5_16, p3_16)))), 4); -#endif res_q = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_q, _mm_add_epi16(sum_q7, q4_16)), 4); -#else _mm_add_epi16( pixelFilter_q, _mm_add_epi16(sum_q6, _mm_add_epi16(q4_16, _mm_add_epi16(q3_16, q5_16)))), 4); -#endif flat2_q4p4 = _mm_packus_epi16(res_p, res_q); -#if !CONFIG_DEBLOCK_13TAP - sum_p7 = _mm_add_epi16(sum_p7, p7_16); - sum_q7 = _mm_add_epi16(sum_q7, q7_16); -#else sum_p6 = _mm_add_epi16(sum_p6, p6_16); sum_q6 = _mm_add_epi16(sum_q6, q6_16); -#endif - -#if !CONFIG_DEBLOCK_13TAP - pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q2_16); - pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p2_16); -#else pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q1_16); pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p1_16); -#endif res_p = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(sum_p7, p5_16)), 4); -#else _mm_add_epi16( pixelFilter_p, _mm_add_epi16(sum_p6, _mm_add_epi16(p5_16, _mm_add_epi16(p6_16, p4_16)))), 4); -#endif res_q = _mm_srli_epi16( -#if !CONFIG_DEBLOCK_13TAP - _mm_add_epi16(pixelFilter_q, _mm_add_epi16(sum_q7, q5_16)), 4); -#else _mm_add_epi16( pixelFilter_q, _mm_add_epi16(sum_q6, _mm_add_epi16(q5_16, _mm_add_epi16(q6_16, q4_16)))), 4); -#endif flat2_q5p5 = _mm_packus_epi16(res_p, res_q); - -#if !CONFIG_DEBLOCK_13TAP - sum_p7 = _mm_add_epi16(sum_p7, p7_16); - sum_q7 = _mm_add_epi16(sum_q7, q7_16); - pixelFilter_p = _mm_sub_epi16(pixelFilter_p, q1_16); - pixelFilter_q = _mm_sub_epi16(pixelFilter_q, p1_16); - res_p = _mm_srli_epi16( - _mm_add_epi16(pixelFilter_p, _mm_add_epi16(sum_p7, p6_16)), 4); - res_q = _mm_srli_epi16( - _mm_add_epi16(pixelFilter_q, _mm_add_epi16(sum_q7, q6_16)), 4); - flat2_q6p6 = _mm_packus_epi16(res_p, res_q); -#endif } // wide flat // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ @@ -707,13 +564,6 @@ flat_q0p0 = _mm_and_si128(flat, flat_q0p0); q0p0 = _mm_or_si128(qs0ps0, flat_q0p0); -#if !CONFIG_DEBLOCK_13TAP - q6p6 = _mm_andnot_si128(flat2, q6p6); - flat2_q6p6 = _mm_and_si128(flat2, flat2_q6p6); - q6p6 = _mm_or_si128(q6p6, flat2_q6p6); - store_buffer_horz_8(q6p6, p, 6, s); -#endif - q5p5 = _mm_andnot_si128(flat2, q5p5); flat2_q5p5 = _mm_and_si128(flat2, flat2_q5p5); q5p5 = _mm_or_si128(q5p5, flat2_q5p5);
diff --git a/av1/common/av1_loopfilter.c b/av1/common/av1_loopfilter.c index 398d707..33c787a 100644 --- a/av1/common/av1_loopfilter.c +++ b/av1/common/av1_loopfilter.c
@@ -35,13 +35,6 @@ #endif // CONFIG_EXT_DELTA_Q #endif // CONFIG_LOOPFILTER_LEVEL -#if CONFIG_DEBLOCK_13TAP -#define PARALLEL_DEBLOCKING_5_TAP_CHROMA 1 -#else -#define PARALLEL_DEBLOCKING_5_TAP_CHROMA 0 -#endif - -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA extern void aom_highbd_lpf_horizontal_6_c(uint16_t *s, int p, const uint8_t *blimit, const uint8_t *limit, @@ -51,7 +44,6 @@ const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd); -#endif // 64 bit masks for left transform size. Each 1 represents a position where // we should apply a loop filter across the left border of an 8x8 block @@ -2125,21 +2117,15 @@ if (TX_4X4 >= min_ts) { params->filter_length = 4; } else if (TX_8X8 == min_ts) { -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA if (plane != 0) params->filter_length = 6; else -#endif params->filter_length = 8; } else { params->filter_length = 16; // No wide filtering for chroma plane if (plane != 0) { -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA params->filter_length = 6; -#else - params->filter_length = 8; -#endif } } @@ -2199,7 +2185,6 @@ aom_lpf_vertical_4(p, dst_stride, params.mblim, params.lim, params.hev_thr); break; -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA case 6: // apply 6-tap filter for chroma plane only assert(plane != 0); if (cm->use_highbitdepth) @@ -2210,7 +2195,6 @@ aom_lpf_vertical_6(p, dst_stride, params.mblim, params.lim, params.hev_thr); break; -#endif // apply 8-tap filtering case 8: if (cm->use_highbitdepth) @@ -2224,23 +2208,12 @@ // apply 16-tap filtering case 16: if (cm->use_highbitdepth) -#if CONFIG_DEBLOCK_13TAP aom_highbd_lpf_vertical_16(CONVERT_TO_SHORTPTR(p), dst_stride, params.mblim, params.lim, params.hev_thr, cm->bit_depth); -#else - aom_highbd_lpf_vertical_16(CONVERT_TO_SHORTPTR(p), dst_stride, - params.mblim, params.lim, params.hev_thr, - cm->bit_depth); -#endif else -#if CONFIG_DEBLOCK_13TAP aom_lpf_vertical_16(p, dst_stride, params.mblim, params.lim, params.hev_thr); -#else - aom_lpf_vertical_16(p, dst_stride, params.mblim, params.lim, - params.hev_thr); -#endif break; // no filtering default: break; @@ -2289,7 +2262,6 @@ aom_lpf_horizontal_4(p, dst_stride, params.mblim, params.lim, params.hev_thr); break; -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA // apply 6-tap filtering case 6: assert(plane != 0); @@ -2301,7 +2273,6 @@ aom_lpf_horizontal_6(p, dst_stride, params.mblim, params.lim, params.hev_thr); break; -#endif // apply 8-tap filtering case 8: if (cm->use_highbitdepth) @@ -2315,23 +2286,12 @@ // apply 16-tap filtering case 16: if (cm->use_highbitdepth) -#if CONFIG_DEBLOCK_13TAP aom_highbd_lpf_horizontal_16(CONVERT_TO_SHORTPTR(p), dst_stride, params.mblim, params.lim, params.hev_thr, cm->bit_depth); -#else - aom_highbd_lpf_horizontal_16(CONVERT_TO_SHORTPTR(p), dst_stride, - params.mblim, params.lim, - params.hev_thr, cm->bit_depth); -#endif else -#if CONFIG_DEBLOCK_13TAP aom_lpf_horizontal_16(p, dst_stride, params.mblim, params.lim, params.hev_thr); -#else - aom_lpf_horizontal_16(p, dst_stride, params.mblim, params.lim, - params.hev_thr); -#endif break; // no filtering default: break;
diff --git a/build/cmake/aom_config_defaults.cmake b/build/cmake/aom_config_defaults.cmake index a69a355..d14274e 100644 --- a/build/cmake/aom_config_defaults.cmake +++ b/build/cmake/aom_config_defaults.cmake
@@ -95,7 +95,6 @@ set(CONFIG_CFL 1 CACHE NUMBER "AV1 experiment flag.") set(CONFIG_CICP 0 CACHE NUMBER "AV1 experiment flag.") set(CONFIG_COLORSPACE_HEADERS 0 CACHE NUMBER "AV1 experiment flag.") -set(CONFIG_DEBLOCK_13TAP 1 CACHE NUMBER "AV1 experiment flag.") set(CONFIG_DEPENDENT_HORZTILEGROUPS 0 CACHE NUMBER "AV1 experiment flag.") set(CONFIG_DEPENDENT_HORZTILES 0 CACHE NUMBER "AV1 experiment flag.") set(CONFIG_DIST_8X8 1 CACHE NUMBER "AV1 experiment flag.")
diff --git a/test/lpf_test.cc b/test/lpf_test.cc index ca8fbfd..f3d0aa1 100644 --- a/test/lpf_test.cc +++ b/test/lpf_test.cc
@@ -446,19 +446,10 @@ make_tuple(&aom_highbd_lpf_horizontal_4_sse2, &aom_highbd_lpf_horizontal_4_c, 8), make_tuple(&aom_highbd_lpf_vertical_4_sse2, &aom_highbd_lpf_vertical_4_c, 8), -#if PARALLEL_DEBLOCKING_5_TAP_CHROMA - make_tuple(&aom_highbd_lpf_horizontal_6_sse2, &aom_highbd_lpf_horizontal_6_c, - 8), - make_tuple(&aom_highbd_lpf_vertical_6_sse2, &aom_highbd_lpf_vertical_6_c, 8), -#endif make_tuple(&aom_highbd_lpf_horizontal_8_sse2, &aom_highbd_lpf_horizontal_8_c, 8), make_tuple(&aom_highbd_lpf_horizontal_16_sse2, &aom_highbd_lpf_horizontal_16_c, 8), -#if !CONFIG_DEBLOCK_13TAP // No SIMD implementation for deblock_13tap yet - make_tuple(&aom_highbd_lpf_horizontal_16_dual_sse2, - &aom_highbd_lpf_horizontal_16_dual_c, 8), -#endif make_tuple(&aom_highbd_lpf_vertical_8_sse2, &aom_highbd_lpf_vertical_8_c, 8), make_tuple(&aom_highbd_lpf_vertical_16_sse2, &aom_highbd_lpf_vertical_16_c, 8), @@ -469,10 +460,6 @@ 10), make_tuple(&aom_highbd_lpf_horizontal_16_sse2, &aom_highbd_lpf_horizontal_16_c, 10), -#if !CONFIG_DEBLOCK_13TAP // No SIMD implementation for deblock_13tap yet - make_tuple(&aom_highbd_lpf_horizontal_16_dual_sse2, - &aom_highbd_lpf_horizontal_16_dual_c, 10), -#endif make_tuple(&aom_highbd_lpf_vertical_8_sse2, &aom_highbd_lpf_vertical_8_c, 10), make_tuple(&aom_highbd_lpf_vertical_16_sse2, &aom_highbd_lpf_vertical_16_c, 10), @@ -483,18 +470,6 @@ 12), make_tuple(&aom_highbd_lpf_horizontal_16_sse2, &aom_highbd_lpf_horizontal_16_c, 12), -#if !CONFIG_DEBLOCK_13TAP // No SIMD implementation for deblock_13tap yet - make_tuple(&aom_highbd_lpf_horizontal_16_dual_sse2, - &aom_highbd_lpf_horizontal_16_dual_c, 12), - make_tuple(&aom_highbd_lpf_vertical_16_sse2, &aom_highbd_lpf_vertical_16_c, - 12), - make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2, - &aom_highbd_lpf_vertical_16_dual_c, 8), - make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2, - &aom_highbd_lpf_vertical_16_dual_c, 10), - make_tuple(&aom_highbd_lpf_vertical_16_dual_sse2, - &aom_highbd_lpf_vertical_16_dual_c, 12), -#endif make_tuple(&aom_highbd_lpf_vertical_8_sse2, &aom_highbd_lpf_vertical_8_c, 12) }; @@ -504,21 +479,12 @@ const loop_param_t kLoop8Test6[] = { make_tuple(&aom_lpf_horizontal_4_sse2, &aom_lpf_horizontal_4_c, 8), make_tuple(&aom_lpf_horizontal_8_sse2, &aom_lpf_horizontal_8_c, 8), -#if CONFIG_DEBLOCK_13TAP make_tuple(&aom_lpf_horizontal_6_sse2, &aom_lpf_horizontal_6_c, 8), make_tuple(&aom_lpf_vertical_6_sse2, &aom_lpf_vertical_6_c, 8), -#endif make_tuple(&aom_lpf_horizontal_16_sse2, &aom_lpf_horizontal_16_c, 8), -#if !CONFIG_DEBLOCK_13TAP // No SIMD implementation for deblock_13tap yet - make_tuple(&aom_lpf_horizontal_16_dual_sse2, &aom_lpf_horizontal_16_dual_c, - 8), -#endif make_tuple(&aom_lpf_vertical_4_sse2, &aom_lpf_vertical_4_c, 8), make_tuple(&aom_lpf_vertical_8_sse2, &aom_lpf_vertical_8_c, 8), make_tuple(&aom_lpf_vertical_16_sse2, &aom_lpf_vertical_16_c, 8), -#if !CONFIG_DEBLOCK_13TAP - make_tuple(&aom_lpf_vertical_16_dual_sse2, &aom_lpf_vertical_16_dual_c, 8) -#endif }; INSTANTIATE_TEST_CASE_P(SSE2, Loop8Test6Param_lbd, @@ -526,28 +492,6 @@ #endif // HAVE_SSE2 -#if HAVE_AVX2 -#if !CONFIG_DEBLOCK_13TAP // No SIMD implementation for deblock_13tap yet -const hbddual_loop_param_t kHbdLoop8Test9Avx2[] = { - make_tuple(&aom_highbd_lpf_horizontal_16_dual_avx2, - &aom_highbd_lpf_horizontal_16_dual_c, 8), - make_tuple(&aom_highbd_lpf_horizontal_16_dual_avx2, - &aom_highbd_lpf_horizontal_16_dual_c, 10), - make_tuple(&aom_highbd_lpf_horizontal_16_dual_avx2, - &aom_highbd_lpf_horizontal_16_dual_c, 12), - make_tuple(&aom_highbd_lpf_vertical_16_dual_avx2, - &aom_highbd_lpf_vertical_16_dual_c, 8), - make_tuple(&aom_highbd_lpf_vertical_16_dual_avx2, - &aom_highbd_lpf_vertical_16_dual_c, 10), - make_tuple(&aom_highbd_lpf_vertical_16_dual_avx2, - &aom_highbd_lpf_vertical_16_dual_c, 12) -}; - -INSTANTIATE_TEST_CASE_P(AVX2, Loop8Test9Param_hbd, - ::testing::ValuesIn(kHbdLoop8Test9Avx2)); -#endif -#endif - #if HAVE_SSE2 const hbddual_loop_param_t kHbdLoop8Test9[] = { make_tuple(&aom_highbd_lpf_horizontal_4_dual_sse2,