/external/XNNPACK/src/qs8-igemm/gen/ |
D | 2x8c2-minmax-fp32-neonv8-mlal-ld2r.c | 76 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() local 129 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() 148 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r()
|
D | 2x8c2-minmax-rndnu-neon-mlal-ld2r.c | 75 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_igemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r() local 128 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_igemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r() 147 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_igemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r()
|
D | 2x8c2-minmax-fp32-neon-mlal-ld2r.c | 75 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() local 128 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() 147 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r()
|
D | 3x8c2-minmax-rndnu-neon-mlal-ld2r.c | 85 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_igemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r() local 158 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_igemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r() 185 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_igemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r()
|
D | 2x8c4-minmax-rndnu-neon-mlal-ld1r.c | 80 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_igemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_ld1r() local 128 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qs8_igemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_ld1r()
|
D | 2x8c4-minmax-fp32-neon-mlal-ld1r.c | 80 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r() local 128 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qs8_igemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r()
|
/external/XNNPACK/src/qc8-igemm/gen/ |
D | 2x8c2-minmax-fp32-neon-mlal-ld2r.c | 75 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() local 128 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() 147 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r()
|
D | 2x8c2-minmax-fp32-neonv8-mlal-ld2r.c | 76 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() local 129 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() 148 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r()
|
D | 2x8c4-minmax-fp32-neon-mlal-ld1r.c | 80 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r() local 128 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r()
|
D | 2x8c4-minmax-fp32-neonv8-mlal-ld1r.c | 81 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r() local 129 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qc8_igemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r()
|
/external/XNNPACK/src/qc8-gemm/gen/ |
D | 2x8c2-minmax-fp32-neon-mlal-ld2r.c | 62 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() local 114 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() 133 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r()
|
D | 2x8c2-minmax-fp32-neonv8-mlal-ld2r.c | 63 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() local 115 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() 134 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r()
|
D | 2x8c2-minmax-fp32-neon-mlal-ld1r.c | 66 const int16x4_t va11x0 = vld1_dup_s16((const void*)(a1 + 2)); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld1r() local 103 const int8x8_t va1c1x0 = vreinterpret_s8_s16(va11x0); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld1r()
|
D | 2x8c4-minmax-fp32-neon-mlal-ld1r.c | 67 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r() local 115 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r()
|
D | 2x8c2-minmax-fp32-neonv8-mlal-ld1r.c | 67 const int16x4_t va11x0 = vld1_dup_s16((const void*)(a1 + 2)); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld1r() local 104 const int8x8_t va1c1x0 = vreinterpret_s8_s16(va11x0); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld1r()
|
D | 2x8c4-minmax-fp32-neonv8-mlal-ld1r.c | 68 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r() local 116 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qc8_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 2x8c2-minmax-fp32-neonv8-mlal-ld2r.c | 63 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() local 115 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r() 134 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r()
|
D | 2x8c2-minmax-rndnu-neon-mlal-ld2r.c | 62 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r() local 114 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r() 133 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r()
|
D | 2x8c2-minmax-fp32-neon-mlal-ld2r.c | 62 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() local 114 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r() 133 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r()
|
D | 3x8c2-minmax-rndnu-neon-mlal-ld2r.c | 70 const int16x4x2_t va11x0 = vld2_dup_s16((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r() local 142 const int8x8_t va1c2x0 = vreinterpret_s8_s16(va11x0.val[0]); in xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r() 169 const int8x8_t va1c3x0 = vreinterpret_s8_s16(va11x0.val[1]); in xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r()
|
D | 2x8c2-minmax-fp32-neon-mlal-ld1r.c | 66 const int16x4_t va11x0 = vld1_dup_s16((const void*)(a1 + 2)); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld1r() local 103 const int8x8_t va1c1x0 = vreinterpret_s8_s16(va11x0); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld1r()
|
D | 2x8c2-minmax-rndnu-neon-mlal-ld1r.c | 66 const int16x4_t va11x0 = vld1_dup_s16((const void*)(a1 + 2)); in xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld1r() local 103 const int8x8_t va1c1x0 = vreinterpret_s8_s16(va11x0); in xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld1r()
|
D | 2x8c2-minmax-fp32-neonv8-mlal-ld1r.c | 67 const int16x4_t va11x0 = vld1_dup_s16((const void*)(a1 + 2)); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld1r() local 104 const int8x8_t va1c1x0 = vreinterpret_s8_s16(va11x0); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld1r()
|
D | 2x8c4-minmax-fp32-neonv8-mlal-ld1r.c | 68 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r() local 116 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r()
|
D | 2x8c4-minmax-fp32-neon-mlal-ld1r.c | 67 const int32x2_t va11x0 = vld1_dup_s32((const void*)(a1 + 4)); a1 += 8; in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r() local 115 const int8x8_t va1c1x0 = vreinterpret_s8_s32(va11x0); in xnn_qs8_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r()
|