summaryrefslogtreecommitdiff
path: root/vp9/encoder/x86/vp9_dct_ssse3.c
diff options
context:
space:
mode:
Diffstat (limited to 'vp9/encoder/x86/vp9_dct_ssse3.c')
-rw-r--r--vp9/encoder/x86/vp9_dct_ssse3.c80
1 files changed, 38 insertions, 42 deletions
diff --git a/vp9/encoder/x86/vp9_dct_ssse3.c b/vp9/encoder/x86/vp9_dct_ssse3.c
index 1a1d4eabc..fb2a92541 100644
--- a/vp9/encoder/x86/vp9_dct_ssse3.c
+++ b/vp9/encoder/x86/vp9_dct_ssse3.c
@@ -15,16 +15,12 @@
#include "vpx_dsp/x86/inv_txfm_sse2.h"
#include "vpx_dsp/x86/txfm_common_sse2.h"
-void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
- int16_t* coeff_ptr, intptr_t n_coeffs,
- int skip_block, const int16_t* zbin_ptr,
- const int16_t* round_ptr, const int16_t* quant_ptr,
- const int16_t* quant_shift_ptr,
- int16_t* qcoeff_ptr,
- int16_t* dqcoeff_ptr, const int16_t* dequant_ptr,
- uint16_t* eob_ptr,
- const int16_t* scan_ptr,
- const int16_t* iscan_ptr) {
+void vp9_fdct8x8_quant_ssse3(
+ const int16_t *input, int stride, int16_t *coeff_ptr, intptr_t n_coeffs,
+ int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr,
+ const int16_t *quant_ptr, const int16_t *quant_shift_ptr,
+ int16_t *qcoeff_ptr, int16_t *dqcoeff_ptr, const int16_t *dequant_ptr,
+ uint16_t *eob_ptr, const int16_t *scan_ptr, const int16_t *iscan_ptr) {
__m128i zero;
int pass;
// Constants
@@ -42,14 +38,14 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
const __m128i k__cospi_m20_p12 = pair_set_epi16(-cospi_20_64, cospi_12_64);
const __m128i k__DCT_CONST_ROUNDING = _mm_set1_epi32(DCT_CONST_ROUNDING);
// Load input
- __m128i in0 = _mm_load_si128((const __m128i *)(input + 0 * stride));
- __m128i in1 = _mm_load_si128((const __m128i *)(input + 1 * stride));
- __m128i in2 = _mm_load_si128((const __m128i *)(input + 2 * stride));
- __m128i in3 = _mm_load_si128((const __m128i *)(input + 3 * stride));
- __m128i in4 = _mm_load_si128((const __m128i *)(input + 4 * stride));
- __m128i in5 = _mm_load_si128((const __m128i *)(input + 5 * stride));
- __m128i in6 = _mm_load_si128((const __m128i *)(input + 6 * stride));
- __m128i in7 = _mm_load_si128((const __m128i *)(input + 7 * stride));
+ __m128i in0 = _mm_load_si128((const __m128i *)(input + 0 * stride));
+ __m128i in1 = _mm_load_si128((const __m128i *)(input + 1 * stride));
+ __m128i in2 = _mm_load_si128((const __m128i *)(input + 2 * stride));
+ __m128i in3 = _mm_load_si128((const __m128i *)(input + 3 * stride));
+ __m128i in4 = _mm_load_si128((const __m128i *)(input + 4 * stride));
+ __m128i in5 = _mm_load_si128((const __m128i *)(input + 5 * stride));
+ __m128i in6 = _mm_load_si128((const __m128i *)(input + 6 * stride));
+ __m128i in7 = _mm_load_si128((const __m128i *)(input + 7 * stride));
__m128i *in[8];
int index = 0;
@@ -298,9 +294,9 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
// Setup global values
{
- round = _mm_load_si128((const __m128i*)round_ptr);
- quant = _mm_load_si128((const __m128i*)quant_ptr);
- dequant = _mm_load_si128((const __m128i*)dequant_ptr);
+ round = _mm_load_si128((const __m128i *)round_ptr);
+ quant = _mm_load_si128((const __m128i *)quant_ptr);
+ dequant = _mm_load_si128((const __m128i *)dequant_ptr);
}
{
@@ -332,15 +328,15 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
qcoeff0 = _mm_sub_epi16(qcoeff0, coeff0_sign);
qcoeff1 = _mm_sub_epi16(qcoeff1, coeff1_sign);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs), qcoeff0);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs) + 1, qcoeff1);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs), qcoeff0);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs) + 1, qcoeff1);
coeff0 = _mm_mullo_epi16(qcoeff0, dequant);
dequant = _mm_unpackhi_epi64(dequant, dequant);
coeff1 = _mm_mullo_epi16(qcoeff1, dequant);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs), coeff0);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs) + 1, coeff1);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs), coeff0);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs) + 1, coeff1);
}
{
@@ -353,8 +349,8 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
zero_coeff1 = _mm_cmpeq_epi16(coeff1, zero);
nzero_coeff0 = _mm_cmpeq_epi16(zero_coeff0, zero);
nzero_coeff1 = _mm_cmpeq_epi16(zero_coeff1, zero);
- iscan0 = _mm_load_si128((const __m128i*)(iscan_ptr + n_coeffs));
- iscan1 = _mm_load_si128((const __m128i*)(iscan_ptr + n_coeffs) + 1);
+ iscan0 = _mm_load_si128((const __m128i *)(iscan_ptr + n_coeffs));
+ iscan1 = _mm_load_si128((const __m128i *)(iscan_ptr + n_coeffs) + 1);
// Add one to convert from indices to counts
iscan0 = _mm_sub_epi16(iscan0, nzero_coeff0);
iscan1 = _mm_sub_epi16(iscan1, nzero_coeff1);
@@ -388,7 +384,7 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
qcoeff1 = _mm_sub_epi16(qcoeff1, coeff1_sign);
nzflag = _mm_movemask_epi8(_mm_cmpgt_epi16(qcoeff0, thr)) |
- _mm_movemask_epi8(_mm_cmpgt_epi16(qcoeff1, thr));
+ _mm_movemask_epi8(_mm_cmpgt_epi16(qcoeff1, thr));
if (nzflag) {
qcoeff0 = _mm_adds_epi16(qcoeff0, round);
@@ -402,20 +398,20 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
qcoeff0 = _mm_sub_epi16(qcoeff0, coeff0_sign);
qcoeff1 = _mm_sub_epi16(qcoeff1, coeff1_sign);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs), qcoeff0);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs) + 1, qcoeff1);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs), qcoeff0);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs) + 1, qcoeff1);
coeff0 = _mm_mullo_epi16(qcoeff0, dequant);
coeff1 = _mm_mullo_epi16(qcoeff1, dequant);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs), coeff0);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs) + 1, coeff1);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs), coeff0);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs) + 1, coeff1);
} else {
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs), zero);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs) + 1, zero);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs), zero);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs) + 1, zero);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs), zero);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs) + 1, zero);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs), zero);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs) + 1, zero);
}
}
@@ -429,8 +425,8 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
zero_coeff1 = _mm_cmpeq_epi16(coeff1, zero);
nzero_coeff0 = _mm_cmpeq_epi16(zero_coeff0, zero);
nzero_coeff1 = _mm_cmpeq_epi16(zero_coeff1, zero);
- iscan0 = _mm_load_si128((const __m128i*)(iscan_ptr + n_coeffs));
- iscan1 = _mm_load_si128((const __m128i*)(iscan_ptr + n_coeffs) + 1);
+ iscan0 = _mm_load_si128((const __m128i *)(iscan_ptr + n_coeffs));
+ iscan1 = _mm_load_si128((const __m128i *)(iscan_ptr + n_coeffs) + 1);
// Add one to convert from indices to counts
iscan0 = _mm_sub_epi16(iscan0, nzero_coeff0);
iscan1 = _mm_sub_epi16(iscan1, nzero_coeff1);
@@ -456,10 +452,10 @@ void vp9_fdct8x8_quant_ssse3(const int16_t *input, int stride,
}
} else {
do {
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs), zero);
- _mm_store_si128((__m128i*)(dqcoeff_ptr + n_coeffs) + 1, zero);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs), zero);
- _mm_store_si128((__m128i*)(qcoeff_ptr + n_coeffs) + 1, zero);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs), zero);
+ _mm_store_si128((__m128i *)(dqcoeff_ptr + n_coeffs) + 1, zero);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs), zero);
+ _mm_store_si128((__m128i *)(qcoeff_ptr + n_coeffs) + 1, zero);
n_coeffs += 8 * 2;
} while (n_coeffs < 0);
*eob_ptr = 0;
generated by cgit v1.2.3 (git 2.25.1) at 2024-10-10 01:24:33 +0000