35 inline void sum_offset(__m128i* X,  in sum_offset()  argument
42         __m128i tmp = _mm_and_si128(A[i], X[shift + i]);  in sum_offset()
43         A[i] = _mm_xor_si128(A[i], X[shift + i]);  in sum_offset()
49 inline void shift_left2D(__m128i* X, unsigned height, unsigned size_sse_row) {  in shift_left2D()  argument
52         unsigned x0 = X[ind].m128i_u32[0] & 1;  in shift_left2D()
54         X[ind] =  in shift_left2D()
55             _mm_or_si128(_mm_srli_epi16(X[ind], 1), _mm_slli_epi16(_mm_srli_si128(X[ind], 2), 15));  in shift_left2D()
57         unsigned x1 = X[ind + 1].m128i_u32[0] & 1;  in shift_left2D()
58         X[ind + 1] = _mm_or_si128(_mm_srli_epi16(X[ind + 1], 1),  in shift_left2D()
59                                   _mm_slli_epi16(_mm_srli_si128(X[ind + 1], 2), 15));  in shift_left2D()
60         X[ind].m128i_u32[3] |= x1 << 31;  in shift_left2D()
62         unsigned x2 = X[ind + 2].m128i_u32[0] & 1;  in shift_left2D()
63         X[ind + 2] = _mm_or_si128(_mm_srli_epi16(X[ind + 2], 1),  in shift_left2D()
64                                   _mm_slli_epi16(_mm_srli_si128(X[ind + 2], 2), 15));  in shift_left2D()
65         X[ind + 1].m128i_u32[3] |= x2 << 31;  in shift_left2D()
67         unsigned* dst = (unsigned*)&X[ind];  in shift_left2D()
72 inline void shift_right2D(__m128i* X, unsigned height, unsigned size_sse_row) {  in shift_right2D()  argument
76         unsigned x0 = X[ind].m128i_u32[3];  in shift_right2D()
78         X[ind] =  in shift_right2D()
79             _mm_or_si128(_mm_slli_epi16(X[ind], 1), _mm_srli_epi16(_mm_slli_si128(X[ind], 2), 15));  in shift_right2D()
81         unsigned x1 = X[ind + 1].m128i_u32[3];  in shift_right2D()
83         X[ind + 1] = _mm_or_si128(_mm_slli_epi16(X[ind + 1], 1),  in shift_right2D()
84                                   _mm_srli_epi16(_mm_slli_si128(X[ind + 1], 2), 15));  in shift_right2D()
85         X[ind + 1].m128i_u32[0] |= x0;  in shift_right2D()
87         unsigned* dst = (unsigned*)&X[ind];  in shift_right2D()
90         X[ind + 2] = _mm_or_si128(_mm_slli_epi16(X[ind + 2], 1),  in shift_right2D()
91                                   _mm_srli_epi16(_mm_slli_si128(X[ind + 2], 2), 15));  in shift_right2D()
92         X[ind + 2].m128i_u32[0] |= x1;  in shift_right2D()
93         X[ind].m128i_u32[0] |= x2;  in shift_right2D()
101     __m128i X[906], A[900], B[900], C[900];  in UpdateState()  local
120         X[i].m128i_u32[0] = 0;  in UpdateState()
121         X[i].m128i_u32[1] = 0;  in UpdateState()
122         X[i].m128i_u32[2] = 0;  in UpdateState()
123         X[i].m128i_u32[3] = 0;  in UpdateState()
131         unsigned* dst = (unsigned*)&X[(row + 1) * size_sse_row];  in UpdateState()
142     unsigned* dst = (unsigned*)X;  in UpdateState()
152     dst = (unsigned*)&X[(height + 1) * size_sse_row];  in UpdateState()
156     sum_offset(X, A, B, C, size_sse_ar, 0);  in UpdateState()
159     sum_offset(X, A, B, C, size_sse_ar, size_sse_row);  in UpdateState()
162     sum_offset(X, A, B, C, size_sse_ar, 2 * size_sse_row);  in UpdateState()
166     shift_left2D(X, height, size_sse_row);  in UpdateState()
169     sum_offset(X, A, B, C, size_sse_ar, 0);  in UpdateState()
172     sum_offset(X, A, B, C, size_sse_ar, 2 * size_sse_row);  in UpdateState()
176     shift_left2D(X, height, size_sse_row);  in UpdateState()
179     sum_offset(X, A, B, C, size_sse_ar, 0);  in UpdateState()
182     sum_offset(X, A, B, C, size_sse_ar, size_sse_row);  in UpdateState()
185     sum_offset(X, A, B, C, size_sse_ar, 2 * size_sse_row);  in UpdateState()
189     shift_right2D(X, height, size_sse_row);  in UpdateState()
198         X[shift + i] = _mm_and_si128(_mm_and_si128(_mm_or_si128(X[shift + i], A[i]), B[i]), C[i]);  in UpdateState()
205         unsigned* src = (unsigned*)&X[(row + 1) * size_sse_row];  in UpdateState()