/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x8c8-minmax-neon-mlal-padal.c | 110 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() local 114 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() 115 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() 116 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
|
D | 4x8c8-minmax-neon-mlal-padal.c | 129 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() local 134 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() 135 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() 136 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() 137 vprod3x1 = vmlal_s8(vprod3x1, vb1x1, va3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
D | 2x8c8-minmax-neon-mlal-padal.c | 91 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal() local 94 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal() 95 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
|
D | 1x8c8-minmax-neon-mlal-padal.c | 72 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal() local 74 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal()
|
D | 1x16c8-minmax-neon-mlal-padal.c | 88 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal() local 90 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal()
|
D | 2x16c8-minmax-neon-mlal-padal.c | 115 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal() local 118 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal() 119 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 142 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 146 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 147 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 148 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 169 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 174 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 175 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 176 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 177 vprod3x1 = vmlal_s8(vprod3x1, vb1x1, va3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 4x8c8-minmax-neon-mlal-padal.c | 146 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() local 151 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() 152 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() 153 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() 154 vprod3x1 = vmlal_s8(vprod3x1, vb1x1, va3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
D | 2x8c8-minmax-neon-mlal-padal.c | 104 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal() local 107 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal() 108 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
|
D | 3x8c8-minmax-neon-mlal-padal.c | 125 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() local 129 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() 130 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() 131 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
|
D | 1x8c8-minmax-neon-mlal-padal.c | 83 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal() local 85 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal()
|
D | 1x16c8-minmax-neon-mlal-padal.c | 99 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal() local 101 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal()
|
D | 2x16c8-minmax-neon-mlal-padal.c | 128 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_2x16c8__neon_mlal_padal() local 131 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_2x16c8__neon_mlal_padal() 132 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_2x16c8__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 186 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 191 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 192 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 193 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 194 vprod3x1 = vmlal_s8(vprod3x1, vb1x1, va3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 157 const int8x8_t vb1x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 161 vprod0x1 = vmlal_s8(vprod0x1, vb1x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 162 vprod1x1 = vmlal_s8(vprod1x1, vb1x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 163 vprod2x1 = vmlal_s8(vprod2x1, vb1x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|