diff options
Diffstat (limited to 'vpx_dsp/x86')
-rw-r--r-- | vpx_dsp/x86/highbd_idct16x16_add_sse4.c | 6 | ||||
-rw-r--r-- | vpx_dsp/x86/highbd_inv_txfm_sse4.h | 1 | ||||
-rw-r--r-- | vpx_dsp/x86/inv_txfm_sse2.c | 6 | ||||
-rw-r--r-- | vpx_dsp/x86/inv_txfm_sse2.h | 1 |
4 files changed, 8 insertions, 6 deletions
diff --git a/vpx_dsp/x86/highbd_idct16x16_add_sse4.c b/vpx_dsp/x86/highbd_idct16x16_add_sse4.c index de097c66a..7898ee12c 100644 --- a/vpx_dsp/x86/highbd_idct16x16_add_sse4.c +++ b/vpx_dsp/x86/highbd_idct16x16_add_sse4.c @@ -53,7 +53,7 @@ static INLINE void highbd_idct16_4col_stage6(const __m128i *const in, out[15] = in[15]; } -static INLINE void highbd_idct16_4col(__m128i *const io /*io[16]*/) { +void vpx_highbd_idct16_4col_sse4_1(__m128i *const io /*io[16]*/) { __m128i step1[16], step2[16]; // stage 2 @@ -233,7 +233,7 @@ void vpx_highbd_idct16x16_256_add_sse4_1(const tran_low_t *input, in = all[i]; highbd_load_transpose_32bit_8x4(&input[0], 16, &in[0]); highbd_load_transpose_32bit_8x4(&input[8], 16, &in[8]); - highbd_idct16_4col(in); + vpx_highbd_idct16_4col_sse4_1(in); input += 4 * 16; } @@ -243,7 +243,7 @@ void vpx_highbd_idct16x16_256_add_sse4_1(const tran_low_t *input, transpose_32bit_4x4(all[1] + i, out + 4); transpose_32bit_4x4(all[2] + i, out + 8); transpose_32bit_4x4(all[3] + i, out + 12); - highbd_idct16_4col(out); + vpx_highbd_idct16_4col_sse4_1(out); for (j = 0; j < 16; ++j) { highbd_write_buffer_4(dest + j * stride, out[j], bd); diff --git a/vpx_dsp/x86/highbd_inv_txfm_sse4.h b/vpx_dsp/x86/highbd_inv_txfm_sse4.h index e8ec335e3..5a7fd1d39 100644 --- a/vpx_dsp/x86/highbd_inv_txfm_sse4.h +++ b/vpx_dsp/x86/highbd_inv_txfm_sse4.h @@ -107,5 +107,6 @@ static INLINE void highbd_idct4_sse4_1(__m128i *const io) { } void vpx_highbd_idct8x8_half1d_sse4_1(__m128i *const io); +void vpx_highbd_idct16_4col_sse4_1(__m128i *const io /*io[16]*/); #endif // VPX_DSP_X86_HIGHBD_INV_TXFM_SSE4_H_ diff --git a/vpx_dsp/x86/inv_txfm_sse2.c b/vpx_dsp/x86/inv_txfm_sse2.c index e17d2627d..4b02da966 100644 --- a/vpx_dsp/x86/inv_txfm_sse2.c +++ b/vpx_dsp/x86/inv_txfm_sse2.c @@ -514,7 +514,7 @@ void vpx_idct16x16_1_add_sse2(const tran_low_t *input, uint8_t *dest, } } -static void iadst16_8col(__m128i *const in) { +void vpx_iadst16_8col_sse2(__m128i *const in) { // perform 16x16 1-D ADST for 8 columns __m128i s[16], x[16], u[32], v[32]; const __m128i k__cospi_p01_p31 = pair_set_epi16(cospi_1_64, cospi_31_64); @@ -874,8 +874,8 @@ void idct16_sse2(__m128i *const in0, __m128i *const in1) { void iadst16_sse2(__m128i *const in0, __m128i *const in1) { transpose_16bit_16x16(in0, in1); - iadst16_8col(in0); - iadst16_8col(in1); + vpx_iadst16_8col_sse2(in0); + vpx_iadst16_8col_sse2(in1); } // Group the coefficient calculation into smaller functions to prevent stack diff --git a/vpx_dsp/x86/inv_txfm_sse2.h b/vpx_dsp/x86/inv_txfm_sse2.h index 4cb1f0789..d573f66c9 100644 --- a/vpx_dsp/x86/inv_txfm_sse2.h +++ b/vpx_dsp/x86/inv_txfm_sse2.h @@ -701,6 +701,7 @@ void vpx_idct8_sse2(__m128i *const in); void idct16_sse2(__m128i *const in0, __m128i *const in1); void iadst4_sse2(__m128i *const in); void iadst8_sse2(__m128i *const in); +void vpx_iadst16_8col_sse2(__m128i *const in); void iadst16_sse2(__m128i *const in0, __m128i *const in1); void idct32_1024_8x32(const __m128i *const in, __m128i *const out); void idct32_34_8x32_sse2(const __m128i *const in, __m128i *const out); |