-void x265_calcRecons4_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons8_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons16_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons32_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons16_sse4(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons32_sse4(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-
-void x265_getResidual4_sse2(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-void x265_getResidual8_sse2(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-void x265_getResidual16_sse2(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-void x265_getResidual16_sse4(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-void x265_getResidual32_sse2(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-void x265_getResidual32_sse4(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-
-void x265_transpose4_sse2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose8_sse2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose16_sse2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose32_sse2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose64_sse2(pixel *dest, pixel *src, intptr_t stride);
-
-void x265_transpose8_avx2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose16_avx2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose32_avx2(pixel *dest, pixel *src, intptr_t stride);
-void x265_transpose64_avx2(pixel *dest, pixel *src, intptr_t stride);
-
-uint32_t x265_quant_sse4(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int16_t *qCoef, int qBits, int add, int numCoeff);
-uint32_t x265_quant_avx2(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int16_t *qCoef, int qBits, int add, int numCoeff);
-uint32_t x265_nquant_sse4(int32_t *coef, int32_t *quantCoeff, int16_t *qCoef, int qBits, int add, int numCoeff);
-uint32_t x265_nquant_avx2(int32_t *coef, int32_t *quantCoeff, int16_t *qCoef, int qBits, int add, int numCoeff);
-void x265_dequant_normal_sse4(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift);
-void x265_dequant_normal_avx2(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift);
-int x265_count_nonzero_ssse3(const int16_t *quantCoeff, int numCoeff);
-
-void x265_weight_pp_sse4(pixel *src, pixel *dst, intptr_t stride, int width, int height, int w0, int round, int shift, int offset);
-void x265_weight_pp_avx2(pixel *src, pixel *dst, intptr_t stride, int width, int height, int w0, int round, int shift, int offset);
-void x265_weight_sp_sse4(int16_t *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset);
-
-void x265_pixel_ssim_4x4x2_core_mmx2(const uint8_t * pix1, intptr_t stride1,
- const uint8_t * pix2, intptr_t stride2, int sums[2][4]);
-void x265_pixel_ssim_4x4x2_core_sse2(const pixel * pix1, intptr_t stride1,
- const pixel * pix2, intptr_t stride2, int sums[2][4]);
-void x265_pixel_ssim_4x4x2_core_avx(const pixel * pix1, intptr_t stride1,
- const pixel * pix2, intptr_t stride2, int sums[2][4]);
+void x265_getResidual4_sse2(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride);
+void x265_getResidual8_sse2(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride);
+void x265_getResidual16_sse2(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride);
+void x265_getResidual16_sse4(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride);
+void x265_getResidual32_sse2(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride);
+void x265_getResidual32_sse4(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride);
+
+void x265_transpose4_sse2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose8_sse2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose16_sse2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose32_sse2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose64_sse2(pixel* dest, const pixel* src, intptr_t stride);
+
+void x265_transpose8_avx2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose16_avx2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose32_avx2(pixel* dest, const pixel* src, intptr_t stride);
+void x265_transpose64_avx2(pixel* dest, const pixel* src, intptr_t stride);
+
+uint32_t x265_quant_sse4(const int16_t* coef, const int32_t* quantCoeff, int32_t* deltaU, int16_t* qCoef, int qBits, int add, int numCoeff);
+uint32_t x265_quant_avx2(const int16_t* coef, const int32_t* quantCoeff, int32_t* deltaU, int16_t* qCoef, int qBits, int add, int numCoeff);
+uint32_t x265_nquant_sse4(const int16_t* coef, const int32_t* quantCoeff, int16_t* qCoef, int qBits, int add, int numCoeff);
+uint32_t x265_nquant_avx2(const int16_t* coef, const int32_t* quantCoeff, int16_t* qCoef, int qBits, int add, int numCoeff);
+void x265_dequant_normal_sse4(const int16_t* quantCoef, int16_t* coef, int num, int scale, int shift);
+void x265_dequant_normal_avx2(const int16_t* quantCoef, int16_t* coef, int num, int scale, int shift);
+int x265_count_nonzero_ssse3(const int16_t* quantCoeff, int numCoeff);
+
+void x265_weight_pp_sse4(const pixel* src, pixel* dst, intptr_t stride, int width, int height, int w0, int round, int shift, int offset);
+void x265_weight_pp_avx2(const pixel* src, pixel* dst, intptr_t stride, int width, int height, int w0, int round, int shift, int offset);
+void x265_weight_sp_sse4(const int16_t* src, pixel* dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset);
+
+void x265_pixel_ssim_4x4x2_core_mmx2(const uint8_t* pix1, intptr_t stride1,
+ const uint8_t* pix2, intptr_t stride2, int sums[2][4]);
+void x265_pixel_ssim_4x4x2_core_sse2(const pixel* pix1, intptr_t stride1,
+ const pixel* pix2, intptr_t stride2, int sums[2][4]);
+void x265_pixel_ssim_4x4x2_core_avx(const pixel* pix1, intptr_t stride1,
+ const pixel* pix2, intptr_t stride2, int sums[2][4]);