shithub: libvpx

Info  •  Files  •  Log  •  Branches

fdct_quant: resolve missing declarations

Store outputs using store_tran_low()

BUG=webm:1584

Change-Id: I213abe047e14625c5ef80df7fa6fdc2a31e40fb6

--- a/vp9/common/vp9_rtcd_defs.pl

+++ b/vp9/common/vp9_rtcd_defs.pl

@@ -141,7 +141,7 @@

   specialize qw/vp9_block_error_fp avx2 sse2/;

-  specialize qw/vp9_fdct8x8_quant neon ssse3/;

+  specialize qw/vp9_fdct8x8_quant sse2 ssse3 neon/;

   add_proto qw/int64_t vp9_highbd_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd";

   specialize qw/vp9_highbd_block_error sse2/;

--- a/vp9/encoder/x86/vp9_dct_intrin_sse2.c

+++ b/vp9/encoder/x86/vp9_dct_intrin_sse2.c

@@ -14,6 +14,7 @@

 #include "./vp9_rtcd.h"

 #include "./vpx_dsp_rtcd.h"

 #include "vpx_dsp/txfm_common.h"

+#include "vpx_dsp/x86/bitdepth_conversion_sse2.h"

 #include "vpx_dsp/x86/fwd_txfm_sse2.h"

 #include "vpx_dsp/x86/transpose_sse2.h"

 #include "vpx_dsp/x86/txfm_common_sse2.h"

@@ -181,10 +182,10 @@

 }

 void vp9_fdct8x8_quant_sse2(const int16_t *input, int stride,

-                            int16_t *coeff_ptr, intptr_t n_coeffs,

+                            tran_low_t *coeff_ptr, intptr_t n_coeffs,

                             int skip_block, const int16_t *round_ptr,

-                            const int16_t *quant_ptr, int16_t *qcoeff_ptr,

-                            int16_t *dqcoeff_ptr, const int16_t *dequant_ptr,

+                            const int16_t *quant_ptr, tran_low_t *qcoeff_ptr,

+                            tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr,

                             uint16_t *eob_ptr, const int16_t *scan,

                             const int16_t *iscan) {

   __m128i zero;

@@ -497,15 +498,15 @@

         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);

+        store_tran_low(qcoeff0, qcoeff_ptr + n_coeffs);

+        store_tran_low(qcoeff0, qcoeff_ptr + n_coeffs + 8);

         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);

+        store_tran_low(coeff0, dqcoeff_ptr + n_coeffs);

+        store_tran_low(coeff0, dqcoeff_ptr + n_coeffs + 8);

       }

       {

@@ -562,14 +563,14 @@

         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);

+        store_tran_low(qcoeff0, qcoeff_ptr + n_coeffs);

+        store_tran_low(qcoeff1, qcoeff_ptr + n_coeffs + 8);

         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);

+        store_tran_low(coeff0, dqcoeff_ptr + n_coeffs);

+        store_tran_low(coeff1, dqcoeff_ptr + n_coeffs + 8);

       }

       {

@@ -609,10 +610,10 @@

     }

   } 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);

+      store_tran_low(zero, qcoeff_ptr + n_coeffs);

+      store_tran_low(zero, qcoeff_ptr + n_coeffs + 8);

+      store_tran_low(zero, dqcoeff_ptr + n_coeffs);

+      store_tran_low(zero, dqcoeff_ptr + n_coeffs + 8);

       n_coeffs += 8 * 2;

     } while (n_coeffs < 0);

     *eob_ptr = 0;