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()
2407          __m128i sum35 = _mm_add_epi16(row3, row5); \  in stbi__idct_simd()
2416          dct_bfly32o(row3,row4, x3,x4,bias,shift); \  in stbi__idct_simd()
2436    row3 = _mm_load_si128((const __m128i *) (data + 3*8));  in stbi__idct_simd()
2450       dct_interleave16(row3, row7);  in stbi__idct_simd()
2454       dct_interleave16(row1, row3);  in stbi__idct_simd()
2460       dct_interleave16(row2, row3);  in stbi__idct_simd()
2471       __m128i p1 = _mm_packus_epi16(row2, row3);  in stbi__idct_simd()
2517    int16x8_t row0, row1, row2, row3, row4, row5, row6, row7;  in stbi__idct_simd()  local
2581       int16x8_t sum35 = vaddq_s16(row3, row5); \  in stbi__idct_simd()
2582       int16x8_t sum37 = vaddq_s16(row3, row7); \  in stbi__idct_simd()
2595       dct_long_mac(x6, sump23o, row3, rot3_2); \  in stbi__idct_simd()
2600       dct_bfly32o(row3,row4, x3,x4,shiftop,shift); \  in stbi__idct_simd()
2607    row3 = vld1q_s16(data + 3*8);  in stbi__idct_simd()
2629       dct_trn16(row2, row3);  in stbi__idct_simd()
2635       dct_trn32(row1, row3);  in stbi__idct_simd()
2643       dct_trn64(row3, row7);  in stbi__idct_simd()
2661       uint8x8_t p3 = vqrshrun_n_s16(row3, 1);  in stbi__idct_simd()