2337    __m128i row0, row1, row2, row3, row4, row5, row6, row7;  in stbi__idct_simd()  local
2404          dct_rot(y0o,y2o, row7,row3, rot2_0,rot2_1); \  in stbi__idct_simd()
2406          __m128i sum17 = _mm_add_epi16(row1, row7); \  in stbi__idct_simd()
2413          dct_bfly32o(row0,row7, x0,x7,bias,shift); \  in stbi__idct_simd()
2440    row7 = _mm_load_si128((const __m128i *) (data + 7*8));  in stbi__idct_simd()
2450       dct_interleave16(row3, row7);  in stbi__idct_simd()
2456       dct_interleave16(row5, row7);  in stbi__idct_simd()
2462       dct_interleave16(row6, row7);  in stbi__idct_simd()
2473       __m128i p3 = _mm_packus_epi16(row6, row7);  in stbi__idct_simd()
2517    int16x8_t row0, row1, row2, row3, row4, row5, row6, row7;  in stbi__idct_simd()  local
2580       int16x8_t sum17 = vaddq_s16(row1, row7); \  in stbi__idct_simd()
2582       int16x8_t sum37 = vaddq_s16(row3, row7); \  in stbi__idct_simd()
2593       dct_long_mac(x4, sump13o, row7, rot3_0); \  in stbi__idct_simd()
2597       dct_bfly32o(row0,row7, x0,x7,shiftop,shift); \  in stbi__idct_simd()
2611    row7 = vld1q_s16(data + 7*8);  in stbi__idct_simd()
2631       dct_trn16(row6, row7);  in stbi__idct_simd()
2637       dct_trn32(row5, row7);  in stbi__idct_simd()
2643       dct_trn64(row3, row7);  in stbi__idct_simd()
2665       uint8x8_t p7 = vqrshrun_n_s16(row7, 1);  in stbi__idct_simd()