2337    __m128i row0, row1, row2, row3, row4, row5, row6, row7;  in stbi__idct_simd()  local
2405          dct_rot(y1o,y3o, row5,row1, rot3_0,rot3_1); \  in stbi__idct_simd()
2407          __m128i sum35 = _mm_add_epi16(row3, row5); \  in stbi__idct_simd()
2415          dct_bfly32o(row2,row5, x2,x5,bias,shift); \  in stbi__idct_simd()
2438    row5 = _mm_load_si128((const __m128i *) (data + 5*8));  in stbi__idct_simd()
2448       dct_interleave16(row1, row5);  in stbi__idct_simd()
2456       dct_interleave16(row5, row7);  in stbi__idct_simd()
2461       dct_interleave16(row4, row5);  in stbi__idct_simd()
2472       __m128i p2 = _mm_packus_epi16(row4, row5);  in stbi__idct_simd()
2517    int16x8_t row0, row1, row2, row3, row4, row5, row6, row7;  in stbi__idct_simd()  local
2579       int16x8_t sum15 = vaddq_s16(row1, row5); \  in stbi__idct_simd()
2581       int16x8_t sum35 = vaddq_s16(row3, row5); \  in stbi__idct_simd()
2594       dct_long_mac(x5, sump24o, row5, rot3_1); \  in stbi__idct_simd()
2599       dct_bfly32o(row2,row5, x2,x5,shiftop,shift); \  in stbi__idct_simd()
2609    row5 = vld1q_s16(data + 5*8);  in stbi__idct_simd()
2630       dct_trn16(row4, row5);  in stbi__idct_simd()
2637       dct_trn32(row5, row7);  in stbi__idct_simd()
2641       dct_trn64(row1, row5);  in stbi__idct_simd()
2663       uint8x8_t p5 = vqrshrun_n_s16(row5, 1);  in stbi__idct_simd()