From 76a3d3fcc518b48a07aacd04f36bbcf902da9405 Mon Sep 17 00:00:00 2001 From: Linfeng Zhang Date: Wed, 13 Sep 2017 13:05:47 -0700 Subject: [PATCH] Remove the unnecessary upcasts of (int)cospi_{1...31}_64 BUG=webm:1450 Change-Id: Ib046fe28caec5b9ebdc9d0152df7c54ff4266858 --- vpx_dsp/arm/fwd_txfm_neon.c | 8 +- vpx_dsp/x86/highbd_idct16x16_add_sse2.c | 90 +++++------ vpx_dsp/x86/highbd_idct16x16_add_sse4.c | 96 +++++------ vpx_dsp/x86/highbd_idct32x32_add_sse2.c | 248 ++++++++++++++-------------- vpx_dsp/x86/highbd_idct32x32_add_sse4.c | 276 ++++++++++++++++---------------- vpx_dsp/x86/highbd_idct4x4_add_sse2.c | 13 +- vpx_dsp/x86/highbd_idct4x4_add_sse4.c | 8 +- vpx_dsp/x86/highbd_idct8x8_add_sse2.c | 26 +-- vpx_dsp/x86/highbd_idct8x8_add_sse4.c | 24 +-- vpx_dsp/x86/highbd_inv_txfm_sse2.h | 4 +- vpx_dsp/x86/highbd_inv_txfm_sse4.h | 4 +- vpx_dsp/x86/inv_txfm_sse2.c | 103 +++++------- vpx_dsp/x86/inv_txfm_sse2.h | 120 ++++++-------- vpx_dsp/x86/inv_txfm_ssse3.c | 95 ++++++----- vpx_dsp/x86/inv_txfm_ssse3.h | 3 +- 15 files changed, 530 insertions(+), 588 deletions(-) diff --git a/vpx_dsp/arm/fwd_txfm_neon.c b/vpx_dsp/arm/fwd_txfm_neon.c index 918686e..bdee4d8 100644 --- a/vpx_dsp/arm/fwd_txfm_neon.c +++ b/vpx_dsp/arm/fwd_txfm_neon.c @@ -56,10 +56,10 @@ void vpx_fdct8x8_neon(const int16_t *input, tran_low_t *final_output, v_t2_hi = vmlal_n_s16(v_t2_hi, vget_high_s16(v_x3), (int16_t)cospi_8_64); v_t3_lo = vmlsl_n_s16(v_t3_lo, vget_low_s16(v_x2), (int16_t)cospi_8_64); v_t3_hi = vmlsl_n_s16(v_t3_hi, vget_high_s16(v_x2), (int16_t)cospi_8_64); - v_t0_lo = vmulq_n_s32(v_t0_lo, (int32_t)cospi_16_64); - v_t0_hi = vmulq_n_s32(v_t0_hi, (int32_t)cospi_16_64); - v_t1_lo = vmulq_n_s32(v_t1_lo, (int32_t)cospi_16_64); - v_t1_hi = vmulq_n_s32(v_t1_hi, (int32_t)cospi_16_64); + v_t0_lo = vmulq_n_s32(v_t0_lo, cospi_16_64); + v_t0_hi = vmulq_n_s32(v_t0_hi, cospi_16_64); + v_t1_lo = vmulq_n_s32(v_t1_lo, cospi_16_64); + v_t1_hi = vmulq_n_s32(v_t1_hi, cospi_16_64); { const int16x4_t a = vrshrn_n_s32(v_t0_lo, DCT_CONST_BITS); const int16x4_t b = vrshrn_n_s32(v_t0_hi, DCT_CONST_BITS); diff --git a/vpx_dsp/x86/highbd_idct16x16_add_sse2.c b/vpx_dsp/x86/highbd_idct16x16_add_sse2.c index c5d82d9..f4f7235 100644 --- a/vpx_dsp/x86/highbd_idct16x16_add_sse2.c +++ b/vpx_dsp/x86/highbd_idct16x16_add_sse2.c @@ -56,20 +56,20 @@ static INLINE void highbd_idct16_4col(__m128i *const io /*io[16]*/) { __m128i step1[16], step2[16]; // stage 2 - highbd_butterfly_sse2(io[1], io[15], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_butterfly_sse2(io[9], io[7], (int)cospi_14_64, (int)cospi_18_64, - &step2[9], &step2[14]); - highbd_butterfly_sse2(io[5], io[11], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_butterfly_sse2(io[13], io[3], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_butterfly_sse2(io[1], io[15], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_butterfly_sse2(io[9], io[7], cospi_14_64, cospi_18_64, &step2[9], + &step2[14]); + highbd_butterfly_sse2(io[5], io[11], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_butterfly_sse2(io[13], io[3], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 - highbd_butterfly_sse2(io[2], io[14], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_butterfly_sse2(io[10], io[6], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_butterfly_sse2(io[2], io[14], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_butterfly_sse2(io[10], io[6], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); step1[8] = _mm_add_epi32(step2[8], step2[9]); step1[9] = _mm_sub_epi32(step2[8], step2[9]); step1[10] = _mm_sub_epi32(step2[10], step2[11]); // step1[10] = -step1[10] @@ -81,11 +81,11 @@ static INLINE void highbd_idct16_4col(__m128i *const io /*io[16]*/) { // stage 4 highbd_butterfly_cospi16_sse2(io[0], io[8], &step2[0], &step2[1]); - highbd_butterfly_sse2(io[4], io[12], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); - highbd_butterfly_sse2(step1[14], step1[9], (int)cospi_24_64, (int)cospi_8_64, - &step2[9], &step2[14]); - highbd_butterfly_sse2(step1[10], step1[13], (int)cospi_8_64, (int)cospi_24_64, + highbd_butterfly_sse2(io[4], io[12], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); + highbd_butterfly_sse2(step1[14], step1[9], cospi_24_64, cospi_8_64, &step2[9], + &step2[14]); + highbd_butterfly_sse2(step1[10], step1[13], cospi_8_64, cospi_24_64, &step2[13], &step2[10]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step1[4] = _mm_add_epi32(step1[4], step1[5]); @@ -106,20 +106,20 @@ static INLINE void highbd_idct16x16_38_4col(__m128i *const io /*io[16]*/) { __m128i temp1[2], sign[2]; // stage 2 - highbd_partial_butterfly_sse2(io[1], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_neg_sse2(io[7], (int)cospi_14_64, (int)cospi_18_64, - &step2[9], &step2[14]); - highbd_partial_butterfly_sse2(io[5], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_partial_butterfly_neg_sse2(io[3], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse2(io[1], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_neg_sse2(io[7], cospi_14_64, cospi_18_64, &step2[9], + &step2[14]); + highbd_partial_butterfly_sse2(io[5], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_partial_butterfly_neg_sse2(io[3], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 - highbd_partial_butterfly_sse2(io[2], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_partial_butterfly_neg_sse2(io[6], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_partial_butterfly_sse2(io[2], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_partial_butterfly_neg_sse2(io[6], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); step1[8] = _mm_add_epi32(step2[8], step2[9]); step1[9] = _mm_sub_epi32(step2[8], step2[9]); step1[10] = _mm_sub_epi32(step2[10], step2[11]); // step1[10] = -step1[10] @@ -131,13 +131,13 @@ static INLINE void highbd_idct16x16_38_4col(__m128i *const io /*io[16]*/) { // stage 4 abs_extend_64bit_sse2(io[0], temp1, sign); - step2[0] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_16_64); + step2[0] = multiplication_round_shift_sse2(temp1, sign, cospi_16_64); step2[1] = step2[0]; - highbd_partial_butterfly_sse2(io[4], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); - highbd_butterfly_sse2(step1[14], step1[9], (int)cospi_24_64, (int)cospi_8_64, - &step2[9], &step2[14]); - highbd_butterfly_sse2(step1[10], step1[13], (int)cospi_8_64, (int)cospi_24_64, + highbd_partial_butterfly_sse2(io[4], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); + highbd_butterfly_sse2(step1[14], step1[9], cospi_24_64, cospi_8_64, &step2[9], + &step2[14]); + highbd_butterfly_sse2(step1[10], step1[13], cospi_8_64, cospi_24_64, &step2[13], &step2[10]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step1[4] = _mm_add_epi32(step1[4], step1[5]); @@ -158,14 +158,14 @@ static INLINE void highbd_idct16x16_10_4col(__m128i *const io /*io[16]*/) { __m128i temp[2], sign[2]; // stage 2 - highbd_partial_butterfly_sse2(io[1], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_neg_sse2(io[3], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse2(io[1], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_neg_sse2(io[3], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 - highbd_partial_butterfly_sse2(io[2], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); + highbd_partial_butterfly_sse2(io[2], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); step1[8] = step2[8]; step1[9] = step2[8]; step1[10] = @@ -179,13 +179,13 @@ static INLINE void highbd_idct16x16_10_4col(__m128i *const io /*io[16]*/) { // stage 4 abs_extend_64bit_sse2(io[0], temp, sign); - step2[0] = multiplication_round_shift_sse2(temp, sign, (int)cospi_16_64); + step2[0] = multiplication_round_shift_sse2(temp, sign, cospi_16_64); step2[1] = step2[0]; step2[2] = _mm_setzero_si128(); step2[3] = _mm_setzero_si128(); - highbd_butterfly_sse2(step1[14], step1[9], (int)cospi_24_64, (int)cospi_8_64, - &step2[9], &step2[14]); - highbd_butterfly_sse2(step1[10], step1[13], (int)cospi_8_64, (int)cospi_24_64, + highbd_butterfly_sse2(step1[14], step1[9], cospi_24_64, cospi_8_64, &step2[9], + &step2[14]); + highbd_butterfly_sse2(step1[10], step1[13], cospi_8_64, cospi_24_64, &step2[13], &step2[10]); step2[5] = step1[4]; step2[6] = step1[7]; diff --git a/vpx_dsp/x86/highbd_idct16x16_add_sse4.c b/vpx_dsp/x86/highbd_idct16x16_add_sse4.c index 4c5452c..de097c6 100644 --- a/vpx_dsp/x86/highbd_idct16x16_add_sse4.c +++ b/vpx_dsp/x86/highbd_idct16x16_add_sse4.c @@ -57,20 +57,20 @@ static INLINE void highbd_idct16_4col(__m128i *const io /*io[16]*/) { __m128i step1[16], step2[16]; // stage 2 - highbd_butterfly_sse4_1(io[1], io[15], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_butterfly_sse4_1(io[9], io[7], (int)cospi_14_64, (int)cospi_18_64, - &step2[9], &step2[14]); - highbd_butterfly_sse4_1(io[5], io[11], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_butterfly_sse4_1(io[13], io[3], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_butterfly_sse4_1(io[1], io[15], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_butterfly_sse4_1(io[9], io[7], cospi_14_64, cospi_18_64, &step2[9], + &step2[14]); + highbd_butterfly_sse4_1(io[5], io[11], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_butterfly_sse4_1(io[13], io[3], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 - highbd_butterfly_sse4_1(io[2], io[14], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_butterfly_sse4_1(io[10], io[6], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_butterfly_sse4_1(io[2], io[14], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_butterfly_sse4_1(io[10], io[6], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); step1[8] = _mm_add_epi32(step2[8], step2[9]); step1[9] = _mm_sub_epi32(step2[8], step2[9]); step1[10] = _mm_sub_epi32(step2[11], step2[10]); @@ -82,12 +82,12 @@ static INLINE void highbd_idct16_4col(__m128i *const io /*io[16]*/) { // stage 4 highbd_butterfly_cospi16_sse4_1(io[0], io[8], &step2[0], &step2[1]); - highbd_butterfly_sse4_1(io[4], io[12], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); - highbd_butterfly_sse4_1(step1[14], step1[9], (int)cospi_24_64, - (int)cospi_8_64, &step2[9], &step2[14]); - highbd_butterfly_sse4_1(step1[10], step1[13], -(int)cospi_8_64, - -(int)cospi_24_64, &step2[13], &step2[10]); + highbd_butterfly_sse4_1(io[4], io[12], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); + highbd_butterfly_sse4_1(step1[14], step1[9], cospi_24_64, cospi_8_64, + &step2[9], &step2[14]); + highbd_butterfly_sse4_1(step1[10], step1[13], -cospi_8_64, -cospi_24_64, + &step2[13], &step2[10]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step1[4] = _mm_add_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -107,20 +107,20 @@ static INLINE void highbd_idct16x16_38_4col(__m128i *const io /*io[16]*/) { __m128i temp1[2]; // stage 2 - highbd_partial_butterfly_sse4_1(io[1], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_sse4_1(io[7], -(int)cospi_18_64, (int)cospi_14_64, - &step2[9], &step2[14]); - highbd_partial_butterfly_sse4_1(io[5], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_partial_butterfly_sse4_1(io[3], -(int)cospi_26_64, (int)cospi_6_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse4_1(io[1], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_sse4_1(io[7], -cospi_18_64, cospi_14_64, &step2[9], + &step2[14]); + highbd_partial_butterfly_sse4_1(io[5], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_partial_butterfly_sse4_1(io[3], -cospi_26_64, cospi_6_64, &step2[11], + &step2[12]); // stage 3 - highbd_partial_butterfly_sse4_1(io[2], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_partial_butterfly_sse4_1(io[6], -(int)cospi_20_64, (int)cospi_12_64, - &step1[5], &step1[6]); + highbd_partial_butterfly_sse4_1(io[2], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_partial_butterfly_sse4_1(io[6], -cospi_20_64, cospi_12_64, &step1[5], + &step1[6]); step1[8] = _mm_add_epi32(step2[8], step2[9]); step1[9] = _mm_sub_epi32(step2[8], step2[9]); step1[10] = _mm_sub_epi32(step2[11], step2[10]); @@ -132,14 +132,14 @@ static INLINE void highbd_idct16x16_38_4col(__m128i *const io /*io[16]*/) { // stage 4 extend_64bit(io[0], temp1); - step2[0] = multiplication_round_shift_sse4_1(temp1, (int)cospi_16_64); + step2[0] = multiplication_round_shift_sse4_1(temp1, cospi_16_64); step2[1] = step2[0]; - highbd_partial_butterfly_sse4_1(io[4], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); - highbd_butterfly_sse4_1(step1[14], step1[9], (int)cospi_24_64, - (int)cospi_8_64, &step2[9], &step2[14]); - highbd_butterfly_sse4_1(step1[10], step1[13], -(int)cospi_8_64, - -(int)cospi_24_64, &step2[13], &step2[10]); + highbd_partial_butterfly_sse4_1(io[4], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); + highbd_butterfly_sse4_1(step1[14], step1[9], cospi_24_64, cospi_8_64, + &step2[9], &step2[14]); + highbd_butterfly_sse4_1(step1[10], step1[13], -cospi_8_64, -cospi_24_64, + &step2[13], &step2[10]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step1[4] = _mm_add_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -159,14 +159,14 @@ static INLINE void highbd_idct16x16_10_4col(__m128i *const io /*io[16]*/) { __m128i temp[2]; // stage 2 - highbd_partial_butterfly_sse4_1(io[1], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_sse4_1(io[3], -(int)cospi_26_64, (int)cospi_6_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse4_1(io[1], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_sse4_1(io[3], -cospi_26_64, cospi_6_64, &step2[11], + &step2[12]); // stage 3 - highbd_partial_butterfly_sse4_1(io[2], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); + highbd_partial_butterfly_sse4_1(io[2], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); step1[8] = step2[8]; step1[9] = step2[8]; step1[10] = step2[11]; @@ -178,14 +178,14 @@ static INLINE void highbd_idct16x16_10_4col(__m128i *const io /*io[16]*/) { // stage 4 extend_64bit(io[0], temp); - step2[0] = multiplication_round_shift_sse4_1(temp, (int)cospi_16_64); + step2[0] = multiplication_round_shift_sse4_1(temp, cospi_16_64); step2[1] = step2[0]; step2[2] = _mm_setzero_si128(); step2[3] = _mm_setzero_si128(); - highbd_butterfly_sse4_1(step1[14], step1[9], (int)cospi_24_64, - (int)cospi_8_64, &step2[9], &step2[14]); - highbd_butterfly_sse4_1(step1[10], step1[13], -(int)cospi_8_64, - -(int)cospi_24_64, &step2[13], &step2[10]); + highbd_butterfly_sse4_1(step1[14], step1[9], cospi_24_64, cospi_8_64, + &step2[9], &step2[14]); + highbd_butterfly_sse4_1(step1[10], step1[13], -cospi_8_64, -cospi_24_64, + &step2[13], &step2[10]); step2[5] = step1[4]; step2[6] = step1[7]; step2[8] = step1[8]; diff --git a/vpx_dsp/x86/highbd_idct32x32_add_sse2.c b/vpx_dsp/x86/highbd_idct32x32_add_sse2.c index 69ba02e..c710e89 100644 --- a/vpx_dsp/x86/highbd_idct32x32_add_sse2.c +++ b/vpx_dsp/x86/highbd_idct32x32_add_sse2.c @@ -21,9 +21,9 @@ static INLINE void highbd_idct32_4x32_quarter_2_stage_4_to_6( // stage 4 step2[8] = step1[8]; step2[15] = step1[15]; - highbd_butterfly_sse2(step1[14], step1[9], (int)cospi_24_64, (int)cospi_8_64, - &step2[9], &step2[14]); - highbd_butterfly_sse2(step1[10], step1[13], (int)cospi_8_64, (int)cospi_24_64, + highbd_butterfly_sse2(step1[14], step1[9], cospi_24_64, cospi_8_64, &step2[9], + &step2[14]); + highbd_butterfly_sse2(step1[10], step1[13], cospi_8_64, cospi_24_64, &step2[13], &step2[10]); step2[11] = step1[11]; step2[12] = step1[12]; @@ -41,10 +41,10 @@ static INLINE void highbd_idct32_4x32_quarter_2_stage_4_to_6( // stage 6 out[8] = step1[8]; out[9] = step1[9]; - highbd_butterfly_sse2(step1[13], step1[10], (int)cospi_16_64, - (int)cospi_16_64, &out[10], &out[13]); - highbd_butterfly_sse2(step1[12], step1[11], (int)cospi_16_64, - (int)cospi_16_64, &out[11], &out[12]); + highbd_butterfly_sse2(step1[13], step1[10], cospi_16_64, cospi_16_64, + &out[10], &out[13]); + highbd_butterfly_sse2(step1[12], step1[11], cospi_16_64, cospi_16_64, + &out[11], &out[12]); out[14] = step1[14]; out[15] = step1[15]; } @@ -75,13 +75,13 @@ static INLINE void highbd_idct32_4x32_quarter_3_4_stage_4_to_7( // stage 5 step1[16] = step2[16]; step1[17] = step2[17]; - highbd_butterfly_sse2(step2[29], step2[18], (int)cospi_24_64, (int)cospi_8_64, + highbd_butterfly_sse2(step2[29], step2[18], cospi_24_64, cospi_8_64, &step1[18], &step1[29]); - highbd_butterfly_sse2(step2[28], step2[19], (int)cospi_24_64, (int)cospi_8_64, + highbd_butterfly_sse2(step2[28], step2[19], cospi_24_64, cospi_8_64, &step1[19], &step1[28]); - highbd_butterfly_sse2(step2[20], step2[27], (int)cospi_8_64, (int)cospi_24_64, + highbd_butterfly_sse2(step2[20], step2[27], cospi_8_64, cospi_24_64, &step1[27], &step1[20]); - highbd_butterfly_sse2(step2[21], step2[26], (int)cospi_8_64, (int)cospi_24_64, + highbd_butterfly_sse2(step2[21], step2[26], cospi_8_64, cospi_24_64, &step1[26], &step1[21]); step1[22] = step2[22]; step1[23] = step2[23]; @@ -114,14 +114,14 @@ static INLINE void highbd_idct32_4x32_quarter_3_4_stage_4_to_7( out[17] = step2[17]; out[18] = step2[18]; out[19] = step2[19]; - highbd_butterfly_sse2(step2[27], step2[20], (int)cospi_16_64, - (int)cospi_16_64, &out[20], &out[27]); - highbd_butterfly_sse2(step2[26], step2[21], (int)cospi_16_64, - (int)cospi_16_64, &out[21], &out[26]); - highbd_butterfly_sse2(step2[25], step2[22], (int)cospi_16_64, - (int)cospi_16_64, &out[22], &out[25]); - highbd_butterfly_sse2(step2[24], step2[23], (int)cospi_16_64, - (int)cospi_16_64, &out[23], &out[24]); + highbd_butterfly_sse2(step2[27], step2[20], cospi_16_64, cospi_16_64, + &out[20], &out[27]); + highbd_butterfly_sse2(step2[26], step2[21], cospi_16_64, cospi_16_64, + &out[21], &out[26]); + highbd_butterfly_sse2(step2[25], step2[22], cospi_16_64, cospi_16_64, + &out[22], &out[25]); + highbd_butterfly_sse2(step2[24], step2[23], cospi_16_64, cospi_16_64, + &out[23], &out[24]); out[28] = step2[28]; out[29] = step2[29]; out[30] = step2[30]; @@ -142,16 +142,16 @@ static INLINE void highbd_idct32_1024_4x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - highbd_butterfly_sse2(in[4], in[28], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_butterfly_sse2(in[20], in[12], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_butterfly_sse2(in[4], in[28], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_butterfly_sse2(in[20], in[12], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); // stage 4 - highbd_butterfly_sse2(in[0], in[16], (int)cospi_16_64, (int)cospi_16_64, - &step2[1], &step2[0]); - highbd_butterfly_sse2(in[8], in[24], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); + highbd_butterfly_sse2(in[0], in[16], cospi_16_64, cospi_16_64, &step2[1], + &step2[0]); + highbd_butterfly_sse2(in[8], in[24], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -163,8 +163,8 @@ static INLINE void highbd_idct32_1024_4x32_quarter_1( step1[2] = _mm_sub_epi32(step2[1], step2[2]); step1[3] = _mm_sub_epi32(step2[0], step2[3]); step1[4] = step2[4]; - highbd_butterfly_sse2(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, - &step1[5], &step1[6]); + highbd_butterfly_sse2(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], + &step1[6]); step1[7] = step2[7]; // stage 6 @@ -186,14 +186,14 @@ static INLINE void highbd_idct32_1024_4x32_quarter_2( __m128i step1[32], step2[32]; // stage 2 - highbd_butterfly_sse2(in[2], in[30], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_butterfly_sse2(in[18], in[14], (int)cospi_14_64, (int)cospi_18_64, - &step2[9], &step2[14]); - highbd_butterfly_sse2(in[10], in[22], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_butterfly_sse2(in[26], in[6], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_butterfly_sse2(in[2], in[30], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_butterfly_sse2(in[18], in[14], cospi_14_64, cospi_18_64, &step2[9], + &step2[14]); + highbd_butterfly_sse2(in[10], in[22], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_butterfly_sse2(in[26], in[6], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 step1[8] = _mm_add_epi32(step2[8], step2[9]); @@ -226,24 +226,24 @@ static INLINE void highbd_idct32_1024_4x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - highbd_butterfly_sse2(in[1], in[31], (int)cospi_31_64, (int)cospi_1_64, - &step1[16], &step1[31]); - highbd_butterfly_sse2(in[17], in[15], (int)cospi_15_64, (int)cospi_17_64, - &step1[17], &step1[30]); - highbd_butterfly_sse2(in[9], in[23], (int)cospi_23_64, (int)cospi_9_64, - &step1[18], &step1[29]); - highbd_butterfly_sse2(in[25], in[7], (int)cospi_7_64, (int)cospi_25_64, - &step1[19], &step1[28]); - - highbd_butterfly_sse2(in[5], in[27], (int)cospi_27_64, (int)cospi_5_64, - &step1[20], &step1[27]); - highbd_butterfly_sse2(in[21], in[11], (int)cospi_11_64, (int)cospi_21_64, - &step1[21], &step1[26]); - - highbd_butterfly_sse2(in[13], in[19], (int)cospi_19_64, (int)cospi_13_64, - &step1[22], &step1[25]); - highbd_butterfly_sse2(in[29], in[3], (int)cospi_3_64, (int)cospi_29_64, - &step1[23], &step1[24]); + highbd_butterfly_sse2(in[1], in[31], cospi_31_64, cospi_1_64, &step1[16], + &step1[31]); + highbd_butterfly_sse2(in[17], in[15], cospi_15_64, cospi_17_64, &step1[17], + &step1[30]); + highbd_butterfly_sse2(in[9], in[23], cospi_23_64, cospi_9_64, &step1[18], + &step1[29]); + highbd_butterfly_sse2(in[25], in[7], cospi_7_64, cospi_25_64, &step1[19], + &step1[28]); + + highbd_butterfly_sse2(in[5], in[27], cospi_27_64, cospi_5_64, &step1[20], + &step1[27]); + highbd_butterfly_sse2(in[21], in[11], cospi_11_64, cospi_21_64, &step1[21], + &step1[26]); + + highbd_butterfly_sse2(in[13], in[19], cospi_19_64, cospi_13_64, &step1[22], + &step1[25]); + highbd_butterfly_sse2(in[29], in[3], cospi_3_64, cospi_29_64, &step1[23], + &step1[24]); // stage 2 step2[16] = _mm_add_epi32(step1[16], step1[17]); @@ -267,16 +267,16 @@ static INLINE void highbd_idct32_1024_4x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - highbd_butterfly_sse2(step2[30], step2[17], (int)cospi_28_64, (int)cospi_4_64, + highbd_butterfly_sse2(step2[30], step2[17], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - highbd_butterfly_sse2(step2[18], step2[29], (int)cospi_4_64, (int)cospi_28_64, + highbd_butterfly_sse2(step2[18], step2[29], cospi_4_64, cospi_28_64, &step1[29], &step1[18]); step1[19] = step2[19]; step1[20] = step2[20]; - highbd_butterfly_sse2(step2[26], step2[21], (int)cospi_12_64, - (int)cospi_20_64, &step1[21], &step1[26]); - highbd_butterfly_sse2(step2[22], step2[25], (int)cospi_20_64, - (int)cospi_12_64, &step1[25], &step1[22]); + highbd_butterfly_sse2(step2[26], step2[21], cospi_12_64, cospi_20_64, + &step1[21], &step1[26]); + highbd_butterfly_sse2(step2[22], step2[25], cospi_20_64, cospi_12_64, + &step1[25], &step1[22]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; @@ -366,16 +366,16 @@ static INLINE void highbd_idct32_135_4x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - highbd_partial_butterfly_sse2(in[4], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_partial_butterfly_neg_sse2(in[12], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_partial_butterfly_sse2(in[4], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_partial_butterfly_neg_sse2(in[12], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); // stage 4 - highbd_partial_butterfly_sse2(in[0], (int)cospi_16_64, (int)cospi_16_64, - &step2[1], &step2[0]); - highbd_partial_butterfly_sse2(in[8], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); + highbd_partial_butterfly_sse2(in[0], cospi_16_64, cospi_16_64, &step2[1], + &step2[0]); + highbd_partial_butterfly_sse2(in[8], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -387,8 +387,8 @@ static INLINE void highbd_idct32_135_4x32_quarter_1( step1[2] = _mm_sub_epi32(step2[1], step2[2]); step1[3] = _mm_sub_epi32(step2[0], step2[3]); step1[4] = step2[4]; - highbd_butterfly_sse2(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, - &step1[5], &step1[6]); + highbd_butterfly_sse2(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], + &step1[6]); step1[7] = step2[7]; // stage 6 @@ -410,14 +410,14 @@ static INLINE void highbd_idct32_135_4x32_quarter_2( __m128i step1[32], step2[32]; // stage 2 - highbd_partial_butterfly_sse2(in[2], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_neg_sse2(in[14], (int)cospi_14_64, (int)cospi_18_64, - &step2[9], &step2[14]); - highbd_partial_butterfly_sse2(in[10], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_partial_butterfly_neg_sse2(in[6], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse2(in[2], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_neg_sse2(in[14], cospi_14_64, cospi_18_64, &step2[9], + &step2[14]); + highbd_partial_butterfly_sse2(in[10], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_partial_butterfly_neg_sse2(in[6], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 step1[8] = _mm_add_epi32(step2[8], step2[9]); @@ -450,24 +450,24 @@ static INLINE void highbd_idct32_135_4x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - highbd_partial_butterfly_sse2(in[1], (int)cospi_31_64, (int)cospi_1_64, - &step1[16], &step1[31]); - highbd_partial_butterfly_neg_sse2(in[15], (int)cospi_15_64, (int)cospi_17_64, + highbd_partial_butterfly_sse2(in[1], cospi_31_64, cospi_1_64, &step1[16], + &step1[31]); + highbd_partial_butterfly_neg_sse2(in[15], cospi_15_64, cospi_17_64, &step1[17], &step1[30]); - highbd_partial_butterfly_sse2(in[9], (int)cospi_23_64, (int)cospi_9_64, - &step1[18], &step1[29]); - highbd_partial_butterfly_neg_sse2(in[7], (int)cospi_7_64, (int)cospi_25_64, - &step1[19], &step1[28]); - - highbd_partial_butterfly_sse2(in[5], (int)cospi_27_64, (int)cospi_5_64, - &step1[20], &step1[27]); - highbd_partial_butterfly_neg_sse2(in[11], (int)cospi_11_64, (int)cospi_21_64, + highbd_partial_butterfly_sse2(in[9], cospi_23_64, cospi_9_64, &step1[18], + &step1[29]); + highbd_partial_butterfly_neg_sse2(in[7], cospi_7_64, cospi_25_64, &step1[19], + &step1[28]); + + highbd_partial_butterfly_sse2(in[5], cospi_27_64, cospi_5_64, &step1[20], + &step1[27]); + highbd_partial_butterfly_neg_sse2(in[11], cospi_11_64, cospi_21_64, &step1[21], &step1[26]); - highbd_partial_butterfly_sse2(in[13], (int)cospi_19_64, (int)cospi_13_64, - &step1[22], &step1[25]); - highbd_partial_butterfly_neg_sse2(in[3], (int)cospi_3_64, (int)cospi_29_64, - &step1[23], &step1[24]); + highbd_partial_butterfly_sse2(in[13], cospi_19_64, cospi_13_64, &step1[22], + &step1[25]); + highbd_partial_butterfly_neg_sse2(in[3], cospi_3_64, cospi_29_64, &step1[23], + &step1[24]); // stage 2 step2[16] = _mm_add_epi32(step1[16], step1[17]); @@ -491,16 +491,16 @@ static INLINE void highbd_idct32_135_4x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - highbd_butterfly_sse2(step2[30], step2[17], (int)cospi_28_64, (int)cospi_4_64, + highbd_butterfly_sse2(step2[30], step2[17], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - highbd_butterfly_sse2(step2[18], step2[29], (int)cospi_4_64, (int)cospi_28_64, + highbd_butterfly_sse2(step2[18], step2[29], cospi_4_64, cospi_28_64, &step1[29], &step1[18]); step1[19] = step2[19]; step1[20] = step2[20]; - highbd_butterfly_sse2(step2[26], step2[21], (int)cospi_12_64, - (int)cospi_20_64, &step1[21], &step1[26]); - highbd_butterfly_sse2(step2[22], step2[25], (int)cospi_20_64, - (int)cospi_12_64, &step1[25], &step1[22]); + highbd_butterfly_sse2(step2[26], step2[21], cospi_12_64, cospi_20_64, + &step1[21], &step1[26]); + highbd_butterfly_sse2(step2[22], step2[25], cospi_20_64, cospi_12_64, + &step1[25], &step1[22]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; @@ -583,12 +583,12 @@ static INLINE void highbd_idct32_34_4x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - highbd_partial_butterfly_sse2(in[4], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); + highbd_partial_butterfly_sse2(in[4], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); // stage 4 - highbd_partial_butterfly_sse2(in[0], (int)cospi_16_64, (int)cospi_16_64, - &step2[1], &step2[0]); + highbd_partial_butterfly_sse2(in[0], cospi_16_64, cospi_16_64, &step2[1], + &step2[0]); step2[4] = step1[4]; step2[5] = step1[4]; step2[6] = step1[7]; @@ -600,8 +600,8 @@ static INLINE void highbd_idct32_34_4x32_quarter_1( step1[2] = step2[1]; step1[3] = step2[0]; step1[4] = step2[4]; - highbd_butterfly_sse2(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, - &step1[5], &step1[6]); + highbd_butterfly_sse2(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], + &step1[6]); step1[7] = step2[7]; // stage 6 @@ -623,10 +623,10 @@ static INLINE void highbd_idct32_34_4x32_quarter_2(const __m128i *in /*in[32]*/, __m128i step1[32], step2[32]; // stage 2 - highbd_partial_butterfly_sse2(in[2], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_neg_sse2(in[6], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse2(in[2], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_neg_sse2(in[6], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 step1[8] = step2[8]; @@ -663,15 +663,15 @@ static INLINE void highbd_idct32_34_4x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - highbd_partial_butterfly_sse2(in[1], (int)cospi_31_64, (int)cospi_1_64, - &step1[16], &step1[31]); - highbd_partial_butterfly_neg_sse2(in[7], (int)cospi_7_64, (int)cospi_25_64, - &step1[19], &step1[28]); + highbd_partial_butterfly_sse2(in[1], cospi_31_64, cospi_1_64, &step1[16], + &step1[31]); + highbd_partial_butterfly_neg_sse2(in[7], cospi_7_64, cospi_25_64, &step1[19], + &step1[28]); - highbd_partial_butterfly_sse2(in[5], (int)cospi_27_64, (int)cospi_5_64, - &step1[20], &step1[27]); - highbd_partial_butterfly_neg_sse2(in[3], (int)cospi_3_64, (int)cospi_29_64, - &step1[23], &step1[24]); + highbd_partial_butterfly_sse2(in[5], cospi_27_64, cospi_5_64, &step1[20], + &step1[27]); + highbd_partial_butterfly_neg_sse2(in[3], cospi_3_64, cospi_29_64, &step1[23], + &step1[24]); // stage 2 step2[16] = step1[16]; @@ -703,16 +703,16 @@ static INLINE void highbd_idct32_34_4x32_quarter_3_4( _mm_sub_epi32(_mm_setzero_si128(), step2[29]); // step2[29] = -step2[29] step1[16] = step2[16]; step1[31] = step2[31]; - highbd_butterfly_sse2(step2[30], step2[17], (int)cospi_28_64, (int)cospi_4_64, + highbd_butterfly_sse2(step2[30], step2[17], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - highbd_butterfly_sse2(step2[18], step2[29], (int)cospi_4_64, (int)cospi_28_64, + highbd_butterfly_sse2(step2[18], step2[29], cospi_4_64, cospi_28_64, &step1[29], &step1[18]); step1[19] = step2[19]; step1[20] = step2[20]; - highbd_butterfly_sse2(step2[26], step2[21], (int)cospi_12_64, - (int)cospi_20_64, &step1[21], &step1[26]); - highbd_butterfly_sse2(step2[22], step2[25], (int)cospi_20_64, - (int)cospi_12_64, &step1[25], &step1[22]); + highbd_butterfly_sse2(step2[26], step2[21], cospi_12_64, cospi_20_64, + &step1[21], &step1[26]); + highbd_butterfly_sse2(step2[22], step2[25], cospi_20_64, cospi_12_64, + &step1[25], &step1[22]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; diff --git a/vpx_dsp/x86/highbd_idct32x32_add_sse4.c b/vpx_dsp/x86/highbd_idct32x32_add_sse4.c index ed6a11f..2d0a53a 100644 --- a/vpx_dsp/x86/highbd_idct32x32_add_sse4.c +++ b/vpx_dsp/x86/highbd_idct32x32_add_sse4.c @@ -25,10 +25,10 @@ static INLINE void highbd_idct32_4x32_quarter_2_stage_4_to_6( // stage 4 step2[8] = step1[8]; step2[15] = step1[15]; - highbd_butterfly_sse4_1(step1[14], step1[9], (int)cospi_24_64, - (int)cospi_8_64, &step2[9], &step2[14]); - highbd_butterfly_sse4_1(step1[13], step1[10], -(int)cospi_8_64, - (int)cospi_24_64, &step2[10], &step2[13]); + highbd_butterfly_sse4_1(step1[14], step1[9], cospi_24_64, cospi_8_64, + &step2[9], &step2[14]); + highbd_butterfly_sse4_1(step1[13], step1[10], -cospi_8_64, cospi_24_64, + &step2[10], &step2[13]); step2[11] = step1[11]; step2[12] = step1[12]; @@ -45,10 +45,10 @@ static INLINE void highbd_idct32_4x32_quarter_2_stage_4_to_6( // stage 6 out[8] = step1[8]; out[9] = step1[9]; - highbd_butterfly_sse4_1(step1[13], step1[10], (int)cospi_16_64, - (int)cospi_16_64, &out[10], &out[13]); - highbd_butterfly_sse4_1(step1[12], step1[11], (int)cospi_16_64, - (int)cospi_16_64, &out[11], &out[12]); + highbd_butterfly_sse4_1(step1[13], step1[10], cospi_16_64, cospi_16_64, + &out[10], &out[13]); + highbd_butterfly_sse4_1(step1[12], step1[11], cospi_16_64, cospi_16_64, + &out[11], &out[12]); out[14] = step1[14]; out[15] = step1[15]; } @@ -79,14 +79,14 @@ static INLINE void highbd_idct32_4x32_quarter_3_4_stage_4_to_7( // stage 5 step1[16] = step2[16]; step1[17] = step2[17]; - highbd_butterfly_sse4_1(step2[29], step2[18], (int)cospi_24_64, - (int)cospi_8_64, &step1[18], &step1[29]); - highbd_butterfly_sse4_1(step2[28], step2[19], (int)cospi_24_64, - (int)cospi_8_64, &step1[19], &step1[28]); - highbd_butterfly_sse4_1(step2[27], step2[20], -(int)cospi_8_64, - (int)cospi_24_64, &step1[20], &step1[27]); - highbd_butterfly_sse4_1(step2[26], step2[21], -(int)cospi_8_64, - (int)cospi_24_64, &step1[21], &step1[26]); + highbd_butterfly_sse4_1(step2[29], step2[18], cospi_24_64, cospi_8_64, + &step1[18], &step1[29]); + highbd_butterfly_sse4_1(step2[28], step2[19], cospi_24_64, cospi_8_64, + &step1[19], &step1[28]); + highbd_butterfly_sse4_1(step2[27], step2[20], -cospi_8_64, cospi_24_64, + &step1[20], &step1[27]); + highbd_butterfly_sse4_1(step2[26], step2[21], -cospi_8_64, cospi_24_64, + &step1[21], &step1[26]); step1[22] = step2[22]; step1[23] = step2[23]; step1[24] = step2[24]; @@ -118,14 +118,14 @@ static INLINE void highbd_idct32_4x32_quarter_3_4_stage_4_to_7( out[17] = step2[17]; out[18] = step2[18]; out[19] = step2[19]; - highbd_butterfly_sse4_1(step2[27], step2[20], (int)cospi_16_64, - (int)cospi_16_64, &out[20], &out[27]); - highbd_butterfly_sse4_1(step2[26], step2[21], (int)cospi_16_64, - (int)cospi_16_64, &out[21], &out[26]); - highbd_butterfly_sse4_1(step2[25], step2[22], (int)cospi_16_64, - (int)cospi_16_64, &out[22], &out[25]); - highbd_butterfly_sse4_1(step2[24], step2[23], (int)cospi_16_64, - (int)cospi_16_64, &out[23], &out[24]); + highbd_butterfly_sse4_1(step2[27], step2[20], cospi_16_64, cospi_16_64, + &out[20], &out[27]); + highbd_butterfly_sse4_1(step2[26], step2[21], cospi_16_64, cospi_16_64, + &out[21], &out[26]); + highbd_butterfly_sse4_1(step2[25], step2[22], cospi_16_64, cospi_16_64, + &out[22], &out[25]); + highbd_butterfly_sse4_1(step2[24], step2[23], cospi_16_64, cospi_16_64, + &out[23], &out[24]); out[28] = step2[28]; out[29] = step2[29]; out[30] = step2[30]; @@ -146,16 +146,16 @@ static INLINE void highbd_idct32_1024_4x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - highbd_butterfly_sse4_1(in[4], in[28], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_butterfly_sse4_1(in[20], in[12], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_butterfly_sse4_1(in[4], in[28], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_butterfly_sse4_1(in[20], in[12], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); // stage 4 - highbd_butterfly_sse4_1(in[0], in[16], (int)cospi_16_64, (int)cospi_16_64, - &step2[1], &step2[0]); - highbd_butterfly_sse4_1(in[8], in[24], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); + highbd_butterfly_sse4_1(in[0], in[16], cospi_16_64, cospi_16_64, &step2[1], + &step2[0]); + highbd_butterfly_sse4_1(in[8], in[24], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -167,8 +167,8 @@ static INLINE void highbd_idct32_1024_4x32_quarter_1( step1[2] = _mm_sub_epi32(step2[1], step2[2]); step1[3] = _mm_sub_epi32(step2[0], step2[3]); step1[4] = step2[4]; - highbd_butterfly_sse4_1(step2[6], step2[5], (int)cospi_16_64, - (int)cospi_16_64, &step1[5], &step1[6]); + highbd_butterfly_sse4_1(step2[6], step2[5], cospi_16_64, cospi_16_64, + &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -190,14 +190,14 @@ static INLINE void highbd_idct32_1024_4x32_quarter_2( __m128i step1[32], step2[32]; // stage 2 - highbd_butterfly_sse4_1(in[2], in[30], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_butterfly_sse4_1(in[18], in[14], (int)cospi_14_64, (int)cospi_18_64, - &step2[9], &step2[14]); - highbd_butterfly_sse4_1(in[10], in[22], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_butterfly_sse4_1(in[26], in[6], (int)cospi_6_64, (int)cospi_26_64, - &step2[11], &step2[12]); + highbd_butterfly_sse4_1(in[2], in[30], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_butterfly_sse4_1(in[18], in[14], cospi_14_64, cospi_18_64, &step2[9], + &step2[14]); + highbd_butterfly_sse4_1(in[10], in[22], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_butterfly_sse4_1(in[26], in[6], cospi_6_64, cospi_26_64, &step2[11], + &step2[12]); // stage 3 step1[8] = _mm_add_epi32(step2[8], step2[9]); @@ -230,24 +230,24 @@ static INLINE void highbd_idct32_1024_4x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - highbd_butterfly_sse4_1(in[1], in[31], (int)cospi_31_64, (int)cospi_1_64, - &step1[16], &step1[31]); - highbd_butterfly_sse4_1(in[17], in[15], (int)cospi_15_64, (int)cospi_17_64, - &step1[17], &step1[30]); - highbd_butterfly_sse4_1(in[9], in[23], (int)cospi_23_64, (int)cospi_9_64, - &step1[18], &step1[29]); - highbd_butterfly_sse4_1(in[25], in[7], (int)cospi_7_64, (int)cospi_25_64, - &step1[19], &step1[28]); - - highbd_butterfly_sse4_1(in[5], in[27], (int)cospi_27_64, (int)cospi_5_64, - &step1[20], &step1[27]); - highbd_butterfly_sse4_1(in[21], in[11], (int)cospi_11_64, (int)cospi_21_64, - &step1[21], &step1[26]); - - highbd_butterfly_sse4_1(in[13], in[19], (int)cospi_19_64, (int)cospi_13_64, - &step1[22], &step1[25]); - highbd_butterfly_sse4_1(in[29], in[3], (int)cospi_3_64, (int)cospi_29_64, - &step1[23], &step1[24]); + highbd_butterfly_sse4_1(in[1], in[31], cospi_31_64, cospi_1_64, &step1[16], + &step1[31]); + highbd_butterfly_sse4_1(in[17], in[15], cospi_15_64, cospi_17_64, &step1[17], + &step1[30]); + highbd_butterfly_sse4_1(in[9], in[23], cospi_23_64, cospi_9_64, &step1[18], + &step1[29]); + highbd_butterfly_sse4_1(in[25], in[7], cospi_7_64, cospi_25_64, &step1[19], + &step1[28]); + + highbd_butterfly_sse4_1(in[5], in[27], cospi_27_64, cospi_5_64, &step1[20], + &step1[27]); + highbd_butterfly_sse4_1(in[21], in[11], cospi_11_64, cospi_21_64, &step1[21], + &step1[26]); + + highbd_butterfly_sse4_1(in[13], in[19], cospi_19_64, cospi_13_64, &step1[22], + &step1[25]); + highbd_butterfly_sse4_1(in[29], in[3], cospi_3_64, cospi_29_64, &step1[23], + &step1[24]); // stage 2 step2[16] = _mm_add_epi32(step1[16], step1[17]); @@ -271,16 +271,16 @@ static INLINE void highbd_idct32_1024_4x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - highbd_butterfly_sse4_1(step2[30], step2[17], (int)cospi_28_64, - (int)cospi_4_64, &step1[17], &step1[30]); - highbd_butterfly_sse4_1(step2[29], step2[18], -(int)cospi_4_64, - (int)cospi_28_64, &step1[18], &step1[29]); + highbd_butterfly_sse4_1(step2[30], step2[17], cospi_28_64, cospi_4_64, + &step1[17], &step1[30]); + highbd_butterfly_sse4_1(step2[29], step2[18], -cospi_4_64, cospi_28_64, + &step1[18], &step1[29]); step1[19] = step2[19]; step1[20] = step2[20]; - highbd_butterfly_sse4_1(step2[26], step2[21], (int)cospi_12_64, - (int)cospi_20_64, &step1[21], &step1[26]); - highbd_butterfly_sse4_1(step2[25], step2[22], -(int)cospi_20_64, - (int)cospi_12_64, &step1[22], &step1[25]); + highbd_butterfly_sse4_1(step2[26], step2[21], cospi_12_64, cospi_20_64, + &step1[21], &step1[26]); + highbd_butterfly_sse4_1(step2[25], step2[22], -cospi_20_64, cospi_12_64, + &step1[22], &step1[25]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; @@ -370,16 +370,16 @@ static INLINE void highbd_idct32_135_4x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - highbd_partial_butterfly_sse4_1(in[4], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_partial_butterfly_sse4_1(in[12], -(int)cospi_20_64, (int)cospi_12_64, - &step1[5], &step1[6]); + highbd_partial_butterfly_sse4_1(in[4], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_partial_butterfly_sse4_1(in[12], -cospi_20_64, cospi_12_64, &step1[5], + &step1[6]); // stage 4 - highbd_partial_butterfly_sse4_1(in[0], (int)cospi_16_64, (int)cospi_16_64, - &step2[1], &step2[0]); - highbd_partial_butterfly_sse4_1(in[8], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); + highbd_partial_butterfly_sse4_1(in[0], cospi_16_64, cospi_16_64, &step2[1], + &step2[0]); + highbd_partial_butterfly_sse4_1(in[8], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -391,8 +391,8 @@ static INLINE void highbd_idct32_135_4x32_quarter_1( step1[2] = _mm_sub_epi32(step2[1], step2[2]); step1[3] = _mm_sub_epi32(step2[0], step2[3]); step1[4] = step2[4]; - highbd_butterfly_sse4_1(step2[6], step2[5], (int)cospi_16_64, - (int)cospi_16_64, &step1[5], &step1[6]); + highbd_butterfly_sse4_1(step2[6], step2[5], cospi_16_64, cospi_16_64, + &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -414,14 +414,14 @@ static INLINE void highbd_idct32_135_4x32_quarter_2( __m128i step1[32], step2[32]; // stage 2 - highbd_partial_butterfly_sse4_1(in[2], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_sse4_1(in[14], -(int)cospi_18_64, (int)cospi_14_64, - &step2[9], &step2[14]); - highbd_partial_butterfly_sse4_1(in[10], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - highbd_partial_butterfly_sse4_1(in[6], -(int)cospi_26_64, (int)cospi_6_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse4_1(in[2], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_sse4_1(in[14], -cospi_18_64, cospi_14_64, &step2[9], + &step2[14]); + highbd_partial_butterfly_sse4_1(in[10], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + highbd_partial_butterfly_sse4_1(in[6], -cospi_26_64, cospi_6_64, &step2[11], + &step2[12]); // stage 3 step1[8] = _mm_add_epi32(step2[8], step2[9]); @@ -454,24 +454,24 @@ static INLINE void highbd_idct32_135_4x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - highbd_partial_butterfly_sse4_1(in[1], (int)cospi_31_64, (int)cospi_1_64, - &step1[16], &step1[31]); - highbd_partial_butterfly_sse4_1(in[15], -(int)cospi_17_64, (int)cospi_15_64, - &step1[17], &step1[30]); - highbd_partial_butterfly_sse4_1(in[9], (int)cospi_23_64, (int)cospi_9_64, - &step1[18], &step1[29]); - highbd_partial_butterfly_sse4_1(in[7], -(int)cospi_25_64, (int)cospi_7_64, - &step1[19], &step1[28]); - - highbd_partial_butterfly_sse4_1(in[5], (int)cospi_27_64, (int)cospi_5_64, - &step1[20], &step1[27]); - highbd_partial_butterfly_sse4_1(in[11], -(int)cospi_21_64, (int)cospi_11_64, - &step1[21], &step1[26]); - - highbd_partial_butterfly_sse4_1(in[13], (int)cospi_19_64, (int)cospi_13_64, - &step1[22], &step1[25]); - highbd_partial_butterfly_sse4_1(in[3], -(int)cospi_29_64, (int)cospi_3_64, - &step1[23], &step1[24]); + highbd_partial_butterfly_sse4_1(in[1], cospi_31_64, cospi_1_64, &step1[16], + &step1[31]); + highbd_partial_butterfly_sse4_1(in[15], -cospi_17_64, cospi_15_64, &step1[17], + &step1[30]); + highbd_partial_butterfly_sse4_1(in[9], cospi_23_64, cospi_9_64, &step1[18], + &step1[29]); + highbd_partial_butterfly_sse4_1(in[7], -cospi_25_64, cospi_7_64, &step1[19], + &step1[28]); + + highbd_partial_butterfly_sse4_1(in[5], cospi_27_64, cospi_5_64, &step1[20], + &step1[27]); + highbd_partial_butterfly_sse4_1(in[11], -cospi_21_64, cospi_11_64, &step1[21], + &step1[26]); + + highbd_partial_butterfly_sse4_1(in[13], cospi_19_64, cospi_13_64, &step1[22], + &step1[25]); + highbd_partial_butterfly_sse4_1(in[3], -cospi_29_64, cospi_3_64, &step1[23], + &step1[24]); // stage 2 step2[16] = _mm_add_epi32(step1[16], step1[17]); @@ -495,16 +495,16 @@ static INLINE void highbd_idct32_135_4x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - highbd_butterfly_sse4_1(step2[30], step2[17], (int)cospi_28_64, - (int)cospi_4_64, &step1[17], &step1[30]); - highbd_butterfly_sse4_1(step2[29], step2[18], -(int)cospi_4_64, - (int)cospi_28_64, &step1[18], &step1[29]); + highbd_butterfly_sse4_1(step2[30], step2[17], cospi_28_64, cospi_4_64, + &step1[17], &step1[30]); + highbd_butterfly_sse4_1(step2[29], step2[18], -cospi_4_64, cospi_28_64, + &step1[18], &step1[29]); step1[19] = step2[19]; step1[20] = step2[20]; - highbd_butterfly_sse4_1(step2[26], step2[21], (int)cospi_12_64, - (int)cospi_20_64, &step1[21], &step1[26]); - highbd_butterfly_sse4_1(step2[25], step2[22], -(int)cospi_20_64, - (int)cospi_12_64, &step1[22], &step1[25]); + highbd_butterfly_sse4_1(step2[26], step2[21], cospi_12_64, cospi_20_64, + &step1[21], &step1[26]); + highbd_butterfly_sse4_1(step2[25], step2[22], -cospi_20_64, cospi_12_64, + &step1[22], &step1[25]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; @@ -583,12 +583,12 @@ static INLINE void highbd_idct32_34_4x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - highbd_partial_butterfly_sse4_1(in[4], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); + highbd_partial_butterfly_sse4_1(in[4], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); // stage 4 - highbd_partial_butterfly_sse4_1(in[0], (int)cospi_16_64, (int)cospi_16_64, - &step2[1], &step2[0]); + highbd_partial_butterfly_sse4_1(in[0], cospi_16_64, cospi_16_64, &step2[1], + &step2[0]); step2[4] = step1[4]; step2[5] = step1[4]; step2[6] = step1[7]; @@ -600,8 +600,8 @@ static INLINE void highbd_idct32_34_4x32_quarter_1( step1[2] = step2[1]; step1[3] = step2[0]; step1[4] = step2[4]; - highbd_butterfly_sse4_1(step2[6], step2[5], (int)cospi_16_64, - (int)cospi_16_64, &step1[5], &step1[6]); + highbd_butterfly_sse4_1(step2[6], step2[5], cospi_16_64, cospi_16_64, + &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -623,10 +623,10 @@ static INLINE void highbd_idct32_34_4x32_quarter_2(const __m128i *in /*in[32]*/, __m128i step1[32], step2[32]; // stage 2 - highbd_partial_butterfly_sse4_1(in[2], (int)cospi_30_64, (int)cospi_2_64, - &step2[8], &step2[15]); - highbd_partial_butterfly_sse4_1(in[6], -(int)cospi_26_64, (int)cospi_6_64, - &step2[11], &step2[12]); + highbd_partial_butterfly_sse4_1(in[2], cospi_30_64, cospi_2_64, &step2[8], + &step2[15]); + highbd_partial_butterfly_sse4_1(in[6], -cospi_26_64, cospi_6_64, &step2[11], + &step2[12]); // stage 3 step1[8] = step2[8]; @@ -659,15 +659,15 @@ static INLINE void highbd_idct32_34_4x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - highbd_partial_butterfly_sse4_1(in[1], (int)cospi_31_64, (int)cospi_1_64, - &step1[16], &step1[31]); - highbd_partial_butterfly_sse4_1(in[7], -(int)cospi_25_64, (int)cospi_7_64, - &step1[19], &step1[28]); + highbd_partial_butterfly_sse4_1(in[1], cospi_31_64, cospi_1_64, &step1[16], + &step1[31]); + highbd_partial_butterfly_sse4_1(in[7], -cospi_25_64, cospi_7_64, &step1[19], + &step1[28]); - highbd_partial_butterfly_sse4_1(in[5], (int)cospi_27_64, (int)cospi_5_64, - &step1[20], &step1[27]); - highbd_partial_butterfly_sse4_1(in[3], -(int)cospi_29_64, (int)cospi_3_64, - &step1[23], &step1[24]); + highbd_partial_butterfly_sse4_1(in[5], cospi_27_64, cospi_5_64, &step1[20], + &step1[27]); + highbd_partial_butterfly_sse4_1(in[3], -cospi_29_64, cospi_3_64, &step1[23], + &step1[24]); // stage 2 step2[16] = step1[16]; @@ -691,16 +691,16 @@ static INLINE void highbd_idct32_34_4x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - highbd_butterfly_sse4_1(step2[30], step2[17], (int)cospi_28_64, - (int)cospi_4_64, &step1[17], &step1[30]); - highbd_butterfly_sse4_1(step2[29], step2[18], -(int)cospi_4_64, - (int)cospi_28_64, &step1[18], &step1[29]); + highbd_butterfly_sse4_1(step2[30], step2[17], cospi_28_64, cospi_4_64, + &step1[17], &step1[30]); + highbd_butterfly_sse4_1(step2[29], step2[18], -cospi_4_64, cospi_28_64, + &step1[18], &step1[29]); step1[19] = step2[19]; step1[20] = step2[20]; - highbd_butterfly_sse4_1(step2[26], step2[21], (int)cospi_12_64, - (int)cospi_20_64, &step1[21], &step1[26]); - highbd_butterfly_sse4_1(step2[25], step2[22], -(int)cospi_20_64, - (int)cospi_12_64, &step1[22], &step1[25]); + highbd_butterfly_sse4_1(step2[26], step2[21], cospi_12_64, cospi_20_64, + &step1[21], &step1[26]); + highbd_butterfly_sse4_1(step2[25], step2[22], -cospi_20_64, cospi_12_64, + &step1[22], &step1[25]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; diff --git a/vpx_dsp/x86/highbd_idct4x4_add_sse2.c b/vpx_dsp/x86/highbd_idct4x4_add_sse2.c index d54a036..2e54d24 100644 --- a/vpx_dsp/x86/highbd_idct4x4_add_sse2.c +++ b/vpx_dsp/x86/highbd_idct4x4_add_sse2.c @@ -24,12 +24,9 @@ static INLINE __m128i dct_const_round_shift_4_sse2(const __m128i in0, } static INLINE void highbd_idct4_small_sse2(__m128i *const io) { - const __m128i cospi_p16_p16 = - _mm_setr_epi32((int)cospi_16_64, 0, (int)cospi_16_64, 0); - const __m128i cospi_p08_p08 = - _mm_setr_epi32((int)cospi_8_64, 0, (int)cospi_8_64, 0); - const __m128i cospi_p24_p24 = - _mm_setr_epi32((int)cospi_24_64, 0, (int)cospi_24_64, 0); + const __m128i cospi_p16_p16 = _mm_setr_epi32(cospi_16_64, 0, cospi_16_64, 0); + const __m128i cospi_p08_p08 = _mm_setr_epi32(cospi_8_64, 0, cospi_8_64, 0); + const __m128i cospi_p24_p24 = _mm_setr_epi32(cospi_24_64, 0, cospi_24_64, 0); __m128i temp1[4], temp2[4], step[4]; transpose_32bit_4x4(io, io); @@ -81,8 +78,8 @@ static INLINE void highbd_idct4_large_sse2(__m128i *const io) { // stage 1 highbd_butterfly_cospi16_sse2(io[0], io[2], &step[0], &step[1]); - highbd_butterfly_sse2(io[1], io[3], (int)cospi_24_64, (int)cospi_8_64, - &step[2], &step[3]); + highbd_butterfly_sse2(io[1], io[3], cospi_24_64, cospi_8_64, &step[2], + &step[3]); // stage 2 io[0] = _mm_add_epi32(step[0], step[3]); // step[0] + step[3] diff --git a/vpx_dsp/x86/highbd_idct4x4_add_sse4.c b/vpx_dsp/x86/highbd_idct4x4_add_sse4.c index 015824f..38e64f3 100644 --- a/vpx_dsp/x86/highbd_idct4x4_add_sse4.c +++ b/vpx_dsp/x86/highbd_idct4x4_add_sse4.c @@ -24,12 +24,12 @@ static INLINE void highbd_idct4(__m128i *const io) { // stage 1 temp[0] = _mm_add_epi32(io[0], io[2]); // input[0] + input[2] extend_64bit(temp[0], temp); - step[0] = multiplication_round_shift_sse4_1(temp, (int)cospi_16_64); + step[0] = multiplication_round_shift_sse4_1(temp, cospi_16_64); temp[0] = _mm_sub_epi32(io[0], io[2]); // input[0] - input[2] extend_64bit(temp[0], temp); - step[1] = multiplication_round_shift_sse4_1(temp, (int)cospi_16_64); - highbd_butterfly_sse4_1(io[1], io[3], (int)cospi_24_64, (int)cospi_8_64, - &step[2], &step[3]); + step[1] = multiplication_round_shift_sse4_1(temp, cospi_16_64); + highbd_butterfly_sse4_1(io[1], io[3], cospi_24_64, cospi_8_64, &step[2], + &step[3]); // stage 2 io[0] = _mm_add_epi32(step[0], step[3]); // step[0] + step[3] diff --git a/vpx_dsp/x86/highbd_idct8x8_add_sse2.c b/vpx_dsp/x86/highbd_idct8x8_add_sse2.c index ff73341..909a6b7 100644 --- a/vpx_dsp/x86/highbd_idct8x8_add_sse2.c +++ b/vpx_dsp/x86/highbd_idct8x8_add_sse2.c @@ -25,15 +25,15 @@ static void highbd_idct8x8_half1d(__m128i *const io) { step1[2] = io[4]; step1[1] = io[2]; step1[3] = io[6]; - highbd_butterfly_sse2(io[1], io[7], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_butterfly_sse2(io[5], io[3], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_butterfly_sse2(io[1], io[7], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_butterfly_sse2(io[5], io[3], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); // stage 2 highbd_butterfly_cospi16_sse2(step1[0], step1[2], &step2[0], &step2[1]); - highbd_butterfly_sse2(step1[1], step1[3], (int)cospi_24_64, (int)cospi_8_64, - &step2[2], &step2[3]); + highbd_butterfly_sse2(step1[1], step1[3], cospi_24_64, cospi_8_64, &step2[2], + &step2[3]); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); @@ -61,18 +61,18 @@ static void highbd_idct8x8_12_half1d(__m128i *const io) { step1[0] = io[0]; step1[1] = io[2]; abs_extend_64bit_sse2(io[1], temp1, sign); - step1[4] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_28_64); - step1[7] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_4_64); + step1[4] = multiplication_round_shift_sse2(temp1, sign, cospi_28_64); + step1[7] = multiplication_round_shift_sse2(temp1, sign, cospi_4_64); abs_extend_64bit_sse2(io[3], temp1, sign); - step1[5] = multiplication_neg_round_shift_sse2(temp1, sign, (int)cospi_20_64); - step1[6] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_12_64); + step1[5] = multiplication_neg_round_shift_sse2(temp1, sign, cospi_20_64); + step1[6] = multiplication_round_shift_sse2(temp1, sign, cospi_12_64); // stage 2 abs_extend_64bit_sse2(step1[0], temp1, sign); - step2[0] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_16_64); + step2[0] = multiplication_round_shift_sse2(temp1, sign, cospi_16_64); abs_extend_64bit_sse2(step1[1], temp1, sign); - step2[2] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_24_64); - step2[3] = multiplication_round_shift_sse2(temp1, sign, (int)cospi_8_64); + step2[2] = multiplication_round_shift_sse2(temp1, sign, cospi_24_64); + step2[3] = multiplication_round_shift_sse2(temp1, sign, cospi_8_64); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); diff --git a/vpx_dsp/x86/highbd_idct8x8_add_sse4.c b/vpx_dsp/x86/highbd_idct8x8_add_sse4.c index 1e9bde0..ae391b2 100644 --- a/vpx_dsp/x86/highbd_idct8x8_add_sse4.c +++ b/vpx_dsp/x86/highbd_idct8x8_add_sse4.c @@ -27,14 +27,14 @@ static void highbd_idct8x8_half1d(__m128i *const io) { step1[2] = io[4]; step1[1] = io[2]; step1[3] = io[6]; - highbd_butterfly_sse4_1(io[1], io[7], (int)cospi_28_64, (int)cospi_4_64, - &step1[4], &step1[7]); - highbd_butterfly_sse4_1(io[5], io[3], (int)cospi_12_64, (int)cospi_20_64, - &step1[5], &step1[6]); + highbd_butterfly_sse4_1(io[1], io[7], cospi_28_64, cospi_4_64, &step1[4], + &step1[7]); + highbd_butterfly_sse4_1(io[5], io[3], cospi_12_64, cospi_20_64, &step1[5], + &step1[6]); // stage 2 highbd_butterfly_cospi16_sse4_1(step1[0], step1[2], &step2[0], &step2[1]); - highbd_butterfly_sse4_1(step1[1], step1[3], (int)cospi_24_64, (int)cospi_8_64, + highbd_butterfly_sse4_1(step1[1], step1[3], cospi_24_64, cospi_8_64, &step2[2], &step2[3]); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); @@ -63,18 +63,18 @@ static void highbd_idct8x8_12_half1d(__m128i *const io) { step1[0] = io[0]; step1[1] = io[2]; extend_64bit(io[1], temp1); - step1[4] = multiplication_round_shift_sse4_1(temp1, (int)cospi_28_64); - step1[7] = multiplication_round_shift_sse4_1(temp1, (int)cospi_4_64); + step1[4] = multiplication_round_shift_sse4_1(temp1, cospi_28_64); + step1[7] = multiplication_round_shift_sse4_1(temp1, cospi_4_64); extend_64bit(io[3], temp1); - step1[5] = multiplication_round_shift_sse4_1(temp1, -(int)cospi_20_64); - step1[6] = multiplication_round_shift_sse4_1(temp1, (int)cospi_12_64); + step1[5] = multiplication_round_shift_sse4_1(temp1, -cospi_20_64); + step1[6] = multiplication_round_shift_sse4_1(temp1, cospi_12_64); // stage 2 extend_64bit(step1[0], temp1); - step2[0] = multiplication_round_shift_sse4_1(temp1, (int)cospi_16_64); + step2[0] = multiplication_round_shift_sse4_1(temp1, cospi_16_64); extend_64bit(step1[1], temp1); - step2[2] = multiplication_round_shift_sse4_1(temp1, (int)cospi_24_64); - step2[3] = multiplication_round_shift_sse4_1(temp1, (int)cospi_8_64); + step2[2] = multiplication_round_shift_sse4_1(temp1, cospi_24_64); + step2[3] = multiplication_round_shift_sse4_1(temp1, cospi_8_64); step2[4] = _mm_add_epi32(step1[4], step1[5]); step2[5] = _mm_sub_epi32(step1[4], step1[5]); step2[6] = _mm_sub_epi32(step1[7], step1[6]); diff --git a/vpx_dsp/x86/highbd_inv_txfm_sse2.h b/vpx_dsp/x86/highbd_inv_txfm_sse2.h index cbe49d1..e0f7495 100644 --- a/vpx_dsp/x86/highbd_inv_txfm_sse2.h +++ b/vpx_dsp/x86/highbd_inv_txfm_sse2.h @@ -181,10 +181,10 @@ static INLINE void highbd_butterfly_cospi16_sse2(const __m128i in0, temp2 = _mm_add_epi32(in0, in1); abs_extend_64bit_sse2(temp2, temp1, sign); - *out0 = multiplication_round_shift_sse2(temp1, sign, (int)cospi_16_64); + *out0 = multiplication_round_shift_sse2(temp1, sign, cospi_16_64); temp2 = _mm_sub_epi32(in0, in1); abs_extend_64bit_sse2(temp2, temp1, sign); - *out1 = multiplication_round_shift_sse2(temp1, sign, (int)cospi_16_64); + *out1 = multiplication_round_shift_sse2(temp1, sign, cospi_16_64); } // Only do addition and subtraction butterfly, size = 16, 32 diff --git a/vpx_dsp/x86/highbd_inv_txfm_sse4.h b/vpx_dsp/x86/highbd_inv_txfm_sse4.h index 68344ff..9c8eef4 100644 --- a/vpx_dsp/x86/highbd_inv_txfm_sse4.h +++ b/vpx_dsp/x86/highbd_inv_txfm_sse4.h @@ -67,10 +67,10 @@ static INLINE void highbd_butterfly_cospi16_sse4_1(const __m128i in0, temp2 = _mm_add_epi32(in0, in1); extend_64bit(temp2, temp1); - *out0 = multiplication_round_shift_sse4_1(temp1, (int)cospi_16_64); + *out0 = multiplication_round_shift_sse4_1(temp1, cospi_16_64); temp2 = _mm_sub_epi32(in0, in1); extend_64bit(temp2, temp1); - *out1 = multiplication_round_shift_sse4_1(temp1, (int)cospi_16_64); + *out1 = multiplication_round_shift_sse4_1(temp1, cospi_16_64); } static INLINE void highbd_partial_butterfly_sse4_1(const __m128i in, diff --git a/vpx_dsp/x86/inv_txfm_sse2.c b/vpx_dsp/x86/inv_txfm_sse2.c index 38670b0..19bd0f1 100644 --- a/vpx_dsp/x86/inv_txfm_sse2.c +++ b/vpx_dsp/x86/inv_txfm_sse2.c @@ -1006,8 +1006,7 @@ static INLINE void idct32_34_8x32_quarter_1(const __m128i *const in /*in[32]*/, __m128i step1[8], step2[8]; // stage 3 - butterfly(in[4], zero, (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); + butterfly(in[4], zero, cospi_28_64, cospi_4_64, &step1[4], &step1[7]); // stage 4 step2[0] = butterfly_cospi16(in[0]); @@ -1022,8 +1021,7 @@ static INLINE void idct32_34_8x32_quarter_1(const __m128i *const in /*in[32]*/, step1[2] = step2[0]; step1[3] = step2[0]; step1[4] = step2[4]; - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -1046,10 +1044,8 @@ static INLINE void idct32_34_8x32_quarter_2(const __m128i *const in /*in[32]*/, __m128i step1[16], step2[16]; // stage 2 - butterfly(in[2], zero, (int)cospi_30_64, (int)cospi_2_64, &step2[8], - &step2[15]); - butterfly(zero, in[6], (int)cospi_6_64, (int)cospi_26_64, &step2[11], - &step2[12]); + butterfly(in[2], zero, cospi_30_64, cospi_2_64, &step2[8], &step2[15]); + butterfly(zero, in[6], cospi_6_64, cospi_26_64, &step2[11], &step2[12]); // stage 3 step1[8] = step2[8]; @@ -1082,24 +1078,20 @@ static INLINE void idct32_34_8x32_quarter_3_4( __m128i step1[32]; // stage 1 - butterfly(in[1], zero, (int)cospi_31_64, (int)cospi_1_64, &step1[16], - &step1[31]); - butterfly(zero, in[7], (int)cospi_7_64, (int)cospi_25_64, &step1[19], - &step1[28]); - butterfly(in[5], zero, (int)cospi_27_64, (int)cospi_5_64, &step1[20], - &step1[27]); - butterfly(zero, in[3], (int)cospi_3_64, (int)cospi_29_64, &step1[23], - &step1[24]); + butterfly(in[1], zero, cospi_31_64, cospi_1_64, &step1[16], &step1[31]); + butterfly(zero, in[7], cospi_7_64, cospi_25_64, &step1[19], &step1[28]); + butterfly(in[5], zero, cospi_27_64, cospi_5_64, &step1[20], &step1[27]); + butterfly(zero, in[3], cospi_3_64, cospi_29_64, &step1[23], &step1[24]); // stage 3 - butterfly(step1[31], step1[16], (int)cospi_28_64, (int)cospi_4_64, &step1[17], + butterfly(step1[31], step1[16], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - butterfly(step1[28], step1[19], -(int)cospi_4_64, (int)cospi_28_64, - &step1[18], &step1[29]); - butterfly(step1[27], step1[20], (int)cospi_12_64, (int)cospi_20_64, - &step1[21], &step1[26]); - butterfly(step1[24], step1[23], -(int)cospi_20_64, (int)cospi_12_64, - &step1[22], &step1[25]); + butterfly(step1[28], step1[19], -cospi_4_64, cospi_28_64, &step1[18], + &step1[29]); + butterfly(step1[27], step1[20], cospi_12_64, cospi_20_64, &step1[21], + &step1[26]); + butterfly(step1[24], step1[23], -cospi_20_64, cospi_12_64, &step1[22], + &step1[25]); idct32_8x32_quarter_3_4_stage_4_to_7(step1, out); } @@ -1145,16 +1137,12 @@ static INLINE void idct32_1024_8x32_quarter_1( __m128i step1[8], step2[8]; // stage 3 - butterfly(in[4], in[28], (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); - butterfly(in[20], in[12], (int)cospi_12_64, (int)cospi_20_64, &step1[5], - &step1[6]); + butterfly(in[4], in[28], cospi_28_64, cospi_4_64, &step1[4], &step1[7]); + butterfly(in[20], in[12], cospi_12_64, cospi_20_64, &step1[5], &step1[6]); // stage 4 - butterfly(in[0], in[16], (int)cospi_16_64, (int)cospi_16_64, &step2[1], - &step2[0]); - butterfly(in[8], in[24], (int)cospi_24_64, (int)cospi_8_64, &step2[2], - &step2[3]); + butterfly(in[0], in[16], cospi_16_64, cospi_16_64, &step2[1], &step2[0]); + butterfly(in[8], in[24], cospi_24_64, cospi_8_64, &step2[2], &step2[3]); step2[4] = _mm_add_epi16(step1[4], step1[5]); step2[5] = _mm_sub_epi16(step1[4], step1[5]); step2[6] = _mm_sub_epi16(step1[7], step1[6]); @@ -1166,8 +1154,7 @@ static INLINE void idct32_1024_8x32_quarter_1( step1[2] = _mm_sub_epi16(step2[1], step2[2]); step1[3] = _mm_sub_epi16(step2[0], step2[3]); step1[4] = step2[4]; - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -1189,14 +1176,10 @@ static INLINE void idct32_1024_8x32_quarter_2( __m128i step1[16], step2[16]; // stage 2 - butterfly(in[2], in[30], (int)cospi_30_64, (int)cospi_2_64, &step2[8], - &step2[15]); - butterfly(in[18], in[14], (int)cospi_14_64, (int)cospi_18_64, &step2[9], - &step2[14]); - butterfly(in[10], in[22], (int)cospi_22_64, (int)cospi_10_64, &step2[10], - &step2[13]); - butterfly(in[26], in[6], (int)cospi_6_64, (int)cospi_26_64, &step2[11], - &step2[12]); + butterfly(in[2], in[30], cospi_30_64, cospi_2_64, &step2[8], &step2[15]); + butterfly(in[18], in[14], cospi_14_64, cospi_18_64, &step2[9], &step2[14]); + butterfly(in[10], in[22], cospi_22_64, cospi_10_64, &step2[10], &step2[13]); + butterfly(in[26], in[6], cospi_6_64, cospi_26_64, &step2[11], &step2[12]); // stage 3 step1[8] = _mm_add_epi16(step2[8], step2[9]); @@ -1229,24 +1212,16 @@ static INLINE void idct32_1024_8x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - butterfly(in[1], in[31], (int)cospi_31_64, (int)cospi_1_64, &step1[16], - &step1[31]); - butterfly(in[17], in[15], (int)cospi_15_64, (int)cospi_17_64, &step1[17], - &step1[30]); - butterfly(in[9], in[23], (int)cospi_23_64, (int)cospi_9_64, &step1[18], - &step1[29]); - butterfly(in[25], in[7], (int)cospi_7_64, (int)cospi_25_64, &step1[19], - &step1[28]); + butterfly(in[1], in[31], cospi_31_64, cospi_1_64, &step1[16], &step1[31]); + butterfly(in[17], in[15], cospi_15_64, cospi_17_64, &step1[17], &step1[30]); + butterfly(in[9], in[23], cospi_23_64, cospi_9_64, &step1[18], &step1[29]); + butterfly(in[25], in[7], cospi_7_64, cospi_25_64, &step1[19], &step1[28]); - butterfly(in[5], in[27], (int)cospi_27_64, (int)cospi_5_64, &step1[20], - &step1[27]); - butterfly(in[21], in[11], (int)cospi_11_64, (int)cospi_21_64, &step1[21], - &step1[26]); + butterfly(in[5], in[27], cospi_27_64, cospi_5_64, &step1[20], &step1[27]); + butterfly(in[21], in[11], cospi_11_64, cospi_21_64, &step1[21], &step1[26]); - butterfly(in[13], in[19], (int)cospi_19_64, (int)cospi_13_64, &step1[22], - &step1[25]); - butterfly(in[29], in[3], (int)cospi_3_64, (int)cospi_29_64, &step1[23], - &step1[24]); + butterfly(in[13], in[19], cospi_19_64, cospi_13_64, &step1[22], &step1[25]); + butterfly(in[29], in[3], cospi_3_64, cospi_29_64, &step1[23], &step1[24]); // stage 2 step2[16] = _mm_add_epi16(step1[16], step1[17]); @@ -1270,16 +1245,16 @@ static INLINE void idct32_1024_8x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - butterfly(step2[30], step2[17], (int)cospi_28_64, (int)cospi_4_64, &step1[17], + butterfly(step2[30], step2[17], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - butterfly(step2[29], step2[18], -(int)cospi_4_64, (int)cospi_28_64, - &step1[18], &step1[29]); + butterfly(step2[29], step2[18], -cospi_4_64, cospi_28_64, &step1[18], + &step1[29]); step1[19] = step2[19]; step1[20] = step2[20]; - butterfly(step2[26], step2[21], (int)cospi_12_64, (int)cospi_20_64, - &step1[21], &step1[26]); - butterfly(step2[25], step2[22], -(int)cospi_20_64, (int)cospi_12_64, - &step1[22], &step1[25]); + butterfly(step2[26], step2[21], cospi_12_64, cospi_20_64, &step1[21], + &step1[26]); + butterfly(step2[25], step2[22], -cospi_20_64, cospi_12_64, &step1[22], + &step1[25]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; diff --git a/vpx_dsp/x86/inv_txfm_sse2.h b/vpx_dsp/x86/inv_txfm_sse2.h index 8c69d90..5cd5098 100644 --- a/vpx_dsp/x86/inv_txfm_sse2.h +++ b/vpx_dsp/x86/inv_txfm_sse2.h @@ -90,7 +90,7 @@ static INLINE void butterfly(const __m128i in0, const __m128i in1, const int c0, } static INLINE __m128i butterfly_cospi16(const __m128i in) { - const __m128i cst = pair_set_epi16((int)cospi_16_64, (int)cospi_16_64); + const __m128i cst = pair_set_epi16(cospi_16_64, cospi_16_64); const __m128i lo = _mm_unpacklo_epi16(in, _mm_setzero_si128()); const __m128i hi = _mm_unpackhi_epi16(in, _mm_setzero_si128()); return idct_calc_wraplow_sse2(lo, hi, cst); @@ -252,16 +252,12 @@ static INLINE void idct8(const __m128i *const in /*in[8]*/, __m128i step1[8], step2[8]; // stage 1 - butterfly(in[1], in[7], (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); - butterfly(in[5], in[3], (int)cospi_12_64, (int)cospi_20_64, &step1[5], - &step1[6]); + butterfly(in[1], in[7], cospi_28_64, cospi_4_64, &step1[4], &step1[7]); + butterfly(in[5], in[3], cospi_12_64, cospi_20_64, &step1[5], &step1[6]); // stage 2 - butterfly(in[0], in[4], (int)cospi_16_64, (int)cospi_16_64, &step2[1], - &step2[0]); - butterfly(in[2], in[6], (int)cospi_24_64, (int)cospi_8_64, &step2[2], - &step2[3]); + butterfly(in[0], in[4], cospi_16_64, cospi_16_64, &step2[1], &step2[0]); + butterfly(in[2], in[6], cospi_24_64, cospi_8_64, &step2[2], &step2[3]); step2[4] = _mm_add_epi16(step1[4], step1[5]); step2[5] = _mm_sub_epi16(step1[4], step1[5]); @@ -273,8 +269,7 @@ static INLINE void idct8(const __m128i *const in /*in[8]*/, step1[1] = _mm_add_epi16(step2[1], step2[2]); step1[2] = _mm_sub_epi16(step2[1], step2[2]); step1[3] = _mm_sub_epi16(step2[0], step2[3]); - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); // stage 4 out[0] = _mm_add_epi16(step1[0], step2[7]); @@ -350,20 +345,14 @@ static INLINE void idct16_8col(const __m128i *const in /*in[16]*/, __m128i step1[16], step2[16]; // stage 2 - butterfly(in[1], in[15], (int)cospi_30_64, (int)cospi_2_64, &step2[8], - &step2[15]); - butterfly(in[9], in[7], (int)cospi_14_64, (int)cospi_18_64, &step2[9], - &step2[14]); - butterfly(in[5], in[11], (int)cospi_22_64, (int)cospi_10_64, &step2[10], - &step2[13]); - butterfly(in[13], in[3], (int)cospi_6_64, (int)cospi_26_64, &step2[11], - &step2[12]); + butterfly(in[1], in[15], cospi_30_64, cospi_2_64, &step2[8], &step2[15]); + butterfly(in[9], in[7], cospi_14_64, cospi_18_64, &step2[9], &step2[14]); + butterfly(in[5], in[11], cospi_22_64, cospi_10_64, &step2[10], &step2[13]); + butterfly(in[13], in[3], cospi_6_64, cospi_26_64, &step2[11], &step2[12]); // stage 3 - butterfly(in[2], in[14], (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); - butterfly(in[10], in[6], (int)cospi_12_64, (int)cospi_20_64, &step1[5], - &step1[6]); + butterfly(in[2], in[14], cospi_28_64, cospi_4_64, &step1[4], &step1[7]); + butterfly(in[10], in[6], cospi_12_64, cospi_20_64, &step1[5], &step1[6]); step1[8] = _mm_add_epi16(step2[8], step2[9]); step1[9] = _mm_sub_epi16(step2[8], step2[9]); step1[10] = _mm_sub_epi16(step2[11], step2[10]); @@ -374,14 +363,12 @@ static INLINE void idct16_8col(const __m128i *const in /*in[16]*/, step1[15] = _mm_add_epi16(step2[14], step2[15]); // stage 4 - butterfly(in[0], in[8], (int)cospi_16_64, (int)cospi_16_64, &step2[1], - &step2[0]); - butterfly(in[4], in[12], (int)cospi_24_64, (int)cospi_8_64, &step2[2], - &step2[3]); - butterfly(step1[14], step1[9], (int)cospi_24_64, (int)cospi_8_64, &step2[9], + butterfly(in[0], in[8], cospi_16_64, cospi_16_64, &step2[1], &step2[0]); + butterfly(in[4], in[12], cospi_24_64, cospi_8_64, &step2[2], &step2[3]); + butterfly(step1[14], step1[9], cospi_24_64, cospi_8_64, &step2[9], &step2[14]); - butterfly(step1[10], step1[13], -(int)cospi_8_64, -(int)cospi_24_64, - &step2[13], &step2[10]); + butterfly(step1[10], step1[13], -cospi_8_64, -cospi_24_64, &step2[13], + &step2[10]); step2[5] = _mm_sub_epi16(step1[4], step1[5]); step1[4] = _mm_add_epi16(step1[4], step1[5]); step2[6] = _mm_sub_epi16(step1[7], step1[6]); @@ -396,8 +383,7 @@ static INLINE void idct16_8col(const __m128i *const in /*in[16]*/, step1[1] = _mm_add_epi16(step2[1], step2[2]); step1[2] = _mm_sub_epi16(step2[1], step2[2]); step1[3] = _mm_sub_epi16(step2[0], step2[3]); - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); step1[8] = _mm_add_epi16(step2[8], step2[11]); step1[9] = _mm_add_epi16(step2[9], step2[10]); step1[10] = _mm_sub_epi16(step2[9], step2[10]); @@ -416,10 +402,10 @@ static INLINE void idct16_8col(const __m128i *const in /*in[16]*/, step2[5] = _mm_sub_epi16(step1[2], step1[5]); step2[6] = _mm_sub_epi16(step1[1], step1[6]); step2[7] = _mm_sub_epi16(step1[0], step1[7]); - butterfly(step1[13], step1[10], (int)cospi_16_64, (int)cospi_16_64, - &step2[10], &step2[13]); - butterfly(step1[12], step1[11], (int)cospi_16_64, (int)cospi_16_64, - &step2[11], &step2[12]); + butterfly(step1[13], step1[10], cospi_16_64, cospi_16_64, &step2[10], + &step2[13]); + butterfly(step1[12], step1[11], cospi_16_64, cospi_16_64, &step2[11], + &step2[12]); // stage 7 out[0] = _mm_add_epi16(step2[0], step1[15]); @@ -553,25 +539,21 @@ static INLINE void idct16x16_10_pass2(__m128i *const l /*l[8]*/, transpose_16bit_4x8(l, io); // stage 2 - butterfly(io[1], zero, (int)cospi_30_64, (int)cospi_2_64, &step2[8], - &step2[15]); - butterfly(zero, io[3], (int)cospi_6_64, (int)cospi_26_64, &step2[11], - &step2[12]); + butterfly(io[1], zero, cospi_30_64, cospi_2_64, &step2[8], &step2[15]); + butterfly(zero, io[3], cospi_6_64, cospi_26_64, &step2[11], &step2[12]); // stage 3 - butterfly(io[2], zero, (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); + butterfly(io[2], zero, cospi_28_64, cospi_4_64, &step1[4], &step1[7]); // stage 4 step1[0] = butterfly_cospi16(io[0]); - butterfly(step2[15], step2[8], (int)cospi_24_64, (int)cospi_8_64, &step2[9], + butterfly(step2[15], step2[8], cospi_24_64, cospi_8_64, &step2[9], &step2[14]); - butterfly(step2[11], step2[12], -(int)cospi_8_64, -(int)cospi_24_64, - &step2[13], &step2[10]); + butterfly(step2[11], step2[12], -cospi_8_64, -cospi_24_64, &step2[13], + &step2[10]); // stage 5 - butterfly(step1[7], step1[4], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step1[7], step1[4], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); step1[8] = _mm_add_epi16(step2[8], step2[11]); step1[9] = _mm_add_epi16(step2[9], step2[10]); step1[10] = _mm_sub_epi16(step2[9], step2[10]); @@ -590,10 +572,10 @@ static INLINE void idct16x16_10_pass2(__m128i *const l /*l[8]*/, step2[5] = _mm_sub_epi16(step1[0], step1[5]); step2[6] = _mm_sub_epi16(step1[0], step1[6]); step2[7] = _mm_sub_epi16(step1[0], step1[7]); - butterfly(step1[13], step1[10], (int)cospi_16_64, (int)cospi_16_64, - &step2[10], &step2[13]); - butterfly(step1[12], step1[11], (int)cospi_16_64, (int)cospi_16_64, - &step2[11], &step2[12]); + butterfly(step1[13], step1[10], cospi_16_64, cospi_16_64, &step2[10], + &step2[13]); + butterfly(step1[12], step1[11], cospi_16_64, cospi_16_64, &step2[11], + &step2[12]); // stage 7 io[0] = _mm_add_epi16(step2[0], step1[15]); @@ -621,10 +603,10 @@ static INLINE void idct32_8x32_quarter_2_stage_4_to_6( // stage 4 step2[8] = step1[8]; step2[15] = step1[15]; - butterfly(step1[14], step1[9], (int)cospi_24_64, (int)cospi_8_64, &step2[9], + butterfly(step1[14], step1[9], cospi_24_64, cospi_8_64, &step2[9], &step2[14]); - butterfly(step1[13], step1[10], -(int)cospi_8_64, (int)cospi_24_64, - &step2[10], &step2[13]); + butterfly(step1[13], step1[10], -cospi_8_64, cospi_24_64, &step2[10], + &step2[13]); step2[11] = step1[11]; step2[12] = step1[12]; @@ -641,10 +623,8 @@ static INLINE void idct32_8x32_quarter_2_stage_4_to_6( // stage 6 out[8] = step1[8]; out[9] = step1[9]; - butterfly(step1[13], step1[10], (int)cospi_16_64, (int)cospi_16_64, &out[10], - &out[13]); - butterfly(step1[12], step1[11], (int)cospi_16_64, (int)cospi_16_64, &out[11], - &out[12]); + butterfly(step1[13], step1[10], cospi_16_64, cospi_16_64, &out[10], &out[13]); + butterfly(step1[12], step1[11], cospi_16_64, cospi_16_64, &out[11], &out[12]); out[14] = step1[14]; out[15] = step1[15]; } @@ -675,14 +655,14 @@ static INLINE void idct32_8x32_quarter_3_4_stage_4_to_7( // stage 5 step1[16] = step2[16]; step1[17] = step2[17]; - butterfly(step2[29], step2[18], (int)cospi_24_64, (int)cospi_8_64, &step1[18], + butterfly(step2[29], step2[18], cospi_24_64, cospi_8_64, &step1[18], &step1[29]); - butterfly(step2[28], step2[19], (int)cospi_24_64, (int)cospi_8_64, &step1[19], + butterfly(step2[28], step2[19], cospi_24_64, cospi_8_64, &step1[19], &step1[28]); - butterfly(step2[27], step2[20], -(int)cospi_8_64, (int)cospi_24_64, - &step1[20], &step1[27]); - butterfly(step2[26], step2[21], -(int)cospi_8_64, (int)cospi_24_64, - &step1[21], &step1[26]); + butterfly(step2[27], step2[20], -cospi_8_64, cospi_24_64, &step1[20], + &step1[27]); + butterfly(step2[26], step2[21], -cospi_8_64, cospi_24_64, &step1[21], + &step1[26]); step1[22] = step2[22]; step1[23] = step2[23]; step1[24] = step2[24]; @@ -710,14 +690,10 @@ static INLINE void idct32_8x32_quarter_3_4_stage_4_to_7( out[31] = _mm_add_epi16(step1[24], step1[31]); // stage 7 - butterfly(step2[27], step2[20], (int)cospi_16_64, (int)cospi_16_64, &out[20], - &out[27]); - butterfly(step2[26], step2[21], (int)cospi_16_64, (int)cospi_16_64, &out[21], - &out[26]); - butterfly(step2[25], step2[22], (int)cospi_16_64, (int)cospi_16_64, &out[22], - &out[25]); - butterfly(step2[24], step2[23], (int)cospi_16_64, (int)cospi_16_64, &out[23], - &out[24]); + butterfly(step2[27], step2[20], cospi_16_64, cospi_16_64, &out[20], &out[27]); + butterfly(step2[26], step2[21], cospi_16_64, cospi_16_64, &out[21], &out[26]); + butterfly(step2[25], step2[22], cospi_16_64, cospi_16_64, &out[22], &out[25]); + butterfly(step2[24], step2[23], cospi_16_64, cospi_16_64, &out[23], &out[24]); } void idct4_sse2(__m128i *const in); diff --git a/vpx_dsp/x86/inv_txfm_ssse3.c b/vpx_dsp/x86/inv_txfm_ssse3.c index be1fdc7..6e99469 100644 --- a/vpx_dsp/x86/inv_txfm_ssse3.c +++ b/vpx_dsp/x86/inv_txfm_ssse3.c @@ -26,7 +26,7 @@ static INLINE void partial_butterfly_ssse3(const __m128i in, const int c0, } static INLINE __m128i partial_butterfly_cospi16_ssse3(const __m128i in) { - const __m128i coef_pair = _mm_set1_epi16(2 * (int)cospi_16_64); + const __m128i coef_pair = _mm_set1_epi16(2 * cospi_16_64); return _mm_mulhrs_epi16(in, coef_pair); } @@ -57,8 +57,7 @@ static INLINE void idct32_34_8x32_quarter_1(const __m128i *const in /*in[32]*/, __m128i step1[8], step2[8]; // stage 3 - partial_butterfly_ssse3(in[4], (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); + partial_butterfly_ssse3(in[4], cospi_28_64, cospi_4_64, &step1[4], &step1[7]); // stage 4 step2[0] = partial_butterfly_cospi16_ssse3(in[0]); @@ -73,8 +72,7 @@ static INLINE void idct32_34_8x32_quarter_1(const __m128i *const in /*in[32]*/, step1[2] = step2[0]; step1[3] = step2[0]; step1[4] = step2[4]; - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -96,9 +94,9 @@ static INLINE void idct32_34_8x32_quarter_2(const __m128i *const in /*in[32]*/, __m128i step1[16], step2[16]; // stage 2 - partial_butterfly_ssse3(in[2], (int)cospi_30_64, (int)cospi_2_64, &step2[8], + partial_butterfly_ssse3(in[2], cospi_30_64, cospi_2_64, &step2[8], &step2[15]); - partial_butterfly_ssse3(in[6], -(int)cospi_26_64, (int)cospi_6_64, &step2[11], + partial_butterfly_ssse3(in[6], -cospi_26_64, cospi_6_64, &step2[11], &step2[12]); // stage 3 @@ -131,24 +129,24 @@ static INLINE void idct32_34_8x32_quarter_3_4( __m128i step1[32]; // stage 1 - partial_butterfly_ssse3(in[1], (int)cospi_31_64, (int)cospi_1_64, &step1[16], + partial_butterfly_ssse3(in[1], cospi_31_64, cospi_1_64, &step1[16], &step1[31]); - partial_butterfly_ssse3(in[7], -(int)cospi_25_64, (int)cospi_7_64, &step1[19], + partial_butterfly_ssse3(in[7], -cospi_25_64, cospi_7_64, &step1[19], &step1[28]); - partial_butterfly_ssse3(in[5], (int)cospi_27_64, (int)cospi_5_64, &step1[20], + partial_butterfly_ssse3(in[5], cospi_27_64, cospi_5_64, &step1[20], &step1[27]); - partial_butterfly_ssse3(in[3], -(int)cospi_29_64, (int)cospi_3_64, &step1[23], + partial_butterfly_ssse3(in[3], -cospi_29_64, cospi_3_64, &step1[23], &step1[24]); // stage 3 - butterfly(step1[31], step1[16], (int)cospi_28_64, (int)cospi_4_64, &step1[17], + butterfly(step1[31], step1[16], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - butterfly(step1[28], step1[19], -(int)cospi_4_64, (int)cospi_28_64, - &step1[18], &step1[29]); - butterfly(step1[27], step1[20], (int)cospi_12_64, (int)cospi_20_64, - &step1[21], &step1[26]); - butterfly(step1[24], step1[23], -(int)cospi_20_64, (int)cospi_12_64, - &step1[22], &step1[25]); + butterfly(step1[28], step1[19], -cospi_4_64, cospi_28_64, &step1[18], + &step1[29]); + butterfly(step1[27], step1[20], cospi_12_64, cospi_20_64, &step1[21], + &step1[26]); + butterfly(step1[24], step1[23], -cospi_20_64, cospi_12_64, &step1[22], + &step1[25]); idct32_8x32_quarter_3_4_stage_4_to_7(step1, out); } @@ -194,15 +192,13 @@ static INLINE void idct32_135_8x32_quarter_1(const __m128i *const in /*in[32]*/, __m128i step1[8], step2[8]; // stage 3 - partial_butterfly_ssse3(in[4], (int)cospi_28_64, (int)cospi_4_64, &step1[4], - &step1[7]); - partial_butterfly_ssse3(in[12], -(int)cospi_20_64, (int)cospi_12_64, - &step1[5], &step1[6]); + partial_butterfly_ssse3(in[4], cospi_28_64, cospi_4_64, &step1[4], &step1[7]); + partial_butterfly_ssse3(in[12], -cospi_20_64, cospi_12_64, &step1[5], + &step1[6]); // stage 4 step2[0] = partial_butterfly_cospi16_ssse3(in[0]); - partial_butterfly_ssse3(in[8], (int)cospi_24_64, (int)cospi_8_64, &step2[2], - &step2[3]); + partial_butterfly_ssse3(in[8], cospi_24_64, cospi_8_64, &step2[2], &step2[3]); step2[4] = _mm_add_epi16(step1[4], step1[5]); step2[5] = _mm_sub_epi16(step1[4], step1[5]); step2[6] = _mm_sub_epi16(step1[7], step1[6]); @@ -214,8 +210,7 @@ static INLINE void idct32_135_8x32_quarter_1(const __m128i *const in /*in[32]*/, step1[2] = _mm_sub_epi16(step2[0], step2[2]); step1[3] = _mm_sub_epi16(step2[0], step2[3]); step1[4] = step2[4]; - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); step1[7] = step2[7]; // stage 6 @@ -237,13 +232,13 @@ static INLINE void idct32_135_8x32_quarter_2(const __m128i *const in /*in[32]*/, __m128i step1[16], step2[16]; // stage 2 - partial_butterfly_ssse3(in[2], (int)cospi_30_64, (int)cospi_2_64, &step2[8], + partial_butterfly_ssse3(in[2], cospi_30_64, cospi_2_64, &step2[8], &step2[15]); - partial_butterfly_ssse3(in[14], -(int)cospi_18_64, (int)cospi_14_64, - &step2[9], &step2[14]); - partial_butterfly_ssse3(in[10], (int)cospi_22_64, (int)cospi_10_64, - &step2[10], &step2[13]); - partial_butterfly_ssse3(in[6], -(int)cospi_26_64, (int)cospi_6_64, &step2[11], + partial_butterfly_ssse3(in[14], -cospi_18_64, cospi_14_64, &step2[9], + &step2[14]); + partial_butterfly_ssse3(in[10], cospi_22_64, cospi_10_64, &step2[10], + &step2[13]); + partial_butterfly_ssse3(in[6], -cospi_26_64, cospi_6_64, &step2[11], &step2[12]); // stage 3 @@ -277,23 +272,23 @@ static INLINE void idct32_135_8x32_quarter_3_4( __m128i step1[32], step2[32]; // stage 1 - partial_butterfly_ssse3(in[1], (int)cospi_31_64, (int)cospi_1_64, &step1[16], + partial_butterfly_ssse3(in[1], cospi_31_64, cospi_1_64, &step1[16], &step1[31]); - partial_butterfly_ssse3(in[15], -(int)cospi_17_64, (int)cospi_15_64, - &step1[17], &step1[30]); - partial_butterfly_ssse3(in[9], (int)cospi_23_64, (int)cospi_9_64, &step1[18], + partial_butterfly_ssse3(in[15], -cospi_17_64, cospi_15_64, &step1[17], + &step1[30]); + partial_butterfly_ssse3(in[9], cospi_23_64, cospi_9_64, &step1[18], &step1[29]); - partial_butterfly_ssse3(in[7], -(int)cospi_25_64, (int)cospi_7_64, &step1[19], + partial_butterfly_ssse3(in[7], -cospi_25_64, cospi_7_64, &step1[19], &step1[28]); - partial_butterfly_ssse3(in[5], (int)cospi_27_64, (int)cospi_5_64, &step1[20], + partial_butterfly_ssse3(in[5], cospi_27_64, cospi_5_64, &step1[20], &step1[27]); - partial_butterfly_ssse3(in[11], -(int)cospi_21_64, (int)cospi_11_64, - &step1[21], &step1[26]); + partial_butterfly_ssse3(in[11], -cospi_21_64, cospi_11_64, &step1[21], + &step1[26]); - partial_butterfly_ssse3(in[13], (int)cospi_19_64, (int)cospi_13_64, - &step1[22], &step1[25]); - partial_butterfly_ssse3(in[3], -(int)cospi_29_64, (int)cospi_3_64, &step1[23], + partial_butterfly_ssse3(in[13], cospi_19_64, cospi_13_64, &step1[22], + &step1[25]); + partial_butterfly_ssse3(in[3], -cospi_29_64, cospi_3_64, &step1[23], &step1[24]); // stage 2 @@ -318,16 +313,16 @@ static INLINE void idct32_135_8x32_quarter_3_4( // stage 3 step1[16] = step2[16]; step1[31] = step2[31]; - butterfly(step2[30], step2[17], (int)cospi_28_64, (int)cospi_4_64, &step1[17], + butterfly(step2[30], step2[17], cospi_28_64, cospi_4_64, &step1[17], &step1[30]); - butterfly(step2[29], step2[18], -(int)cospi_4_64, (int)cospi_28_64, - &step1[18], &step1[29]); + butterfly(step2[29], step2[18], -cospi_4_64, cospi_28_64, &step1[18], + &step1[29]); step1[19] = step2[19]; step1[20] = step2[20]; - butterfly(step2[26], step2[21], (int)cospi_12_64, (int)cospi_20_64, - &step1[21], &step1[26]); - butterfly(step2[25], step2[22], -(int)cospi_20_64, (int)cospi_12_64, - &step1[22], &step1[25]); + butterfly(step2[26], step2[21], cospi_12_64, cospi_20_64, &step1[21], + &step1[26]); + butterfly(step2[25], step2[22], -cospi_20_64, cospi_12_64, &step1[22], + &step1[25]); step1[23] = step2[23]; step1[24] = step2[24]; step1[27] = step2[27]; diff --git a/vpx_dsp/x86/inv_txfm_ssse3.h b/vpx_dsp/x86/inv_txfm_ssse3.h index 52fbf45..d0bd5a5 100644 --- a/vpx_dsp/x86/inv_txfm_ssse3.h +++ b/vpx_dsp/x86/inv_txfm_ssse3.h @@ -92,8 +92,7 @@ static INLINE void idct8x8_12_add_kernel_ssse3(__m128i *const io /* io[8] */) { step1[1] = _mm_add_epi16(step2[0], step2[2]); step1[2] = _mm_sub_epi16(step2[0], step2[2]); step1[3] = _mm_sub_epi16(step2[0], step2[3]); - butterfly(step2[6], step2[5], (int)cospi_16_64, (int)cospi_16_64, &step1[5], - &step1[6]); + butterfly(step2[6], step2[5], cospi_16_64, cospi_16_64, &step1[5], &step1[6]); // stage 4 io[0] = _mm_add_epi16(step1[0], step2[7]); -- 2.7.4