/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x8c8-minmax-neon-mlal-padal.c | 150 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() local 154 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() 155 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() 156 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
|
D | 4x8c8-minmax-neon-mlal-padal.c | 181 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() local 186 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() 187 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() 188 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() 189 vprod3x5 = vmlal_s8(vprod3x5, vb5x1, va3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
D | 2x8c8-minmax-neon-mlal-padal.c | 119 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal() local 122 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal() 123 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
|
D | 1x8c8-minmax-neon-mlal-padal.c | 88 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal() local 90 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal()
|
D | 1x16c8-minmax-neon-mlal-padal.c | 104 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal() local 106 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal()
|
D | 2x16c8-minmax-neon-mlal-padal.c | 143 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal() local 146 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal() 147 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 182 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 186 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 187 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 188 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 221 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 226 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 227 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 228 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 229 vprod3x5 = vmlal_s8(vprod3x5, vb5x1, va3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 4x8c8-minmax-neon-mlal-padal.c | 198 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() local 203 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() 204 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() 205 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() 206 vprod3x5 = vmlal_s8(vprod3x5, vb5x1, va3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
D | 2x8c8-minmax-neon-mlal-padal.c | 132 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal() local 135 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal() 136 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
|
D | 3x8c8-minmax-neon-mlal-padal.c | 165 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() local 169 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() 170 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() 171 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
|
D | 1x8c8-minmax-neon-mlal-padal.c | 99 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal() local 101 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal()
|
D | 1x16c8-minmax-neon-mlal-padal.c | 115 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal() local 117 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal()
|
D | 2x16c8-minmax-neon-mlal-padal.c | 156 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_2x16c8__neon_mlal_padal() local 159 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_2x16c8__neon_mlal_padal() 160 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_2x16c8__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 238 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 243 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 244 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 245 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 246 vprod3x5 = vmlal_s8(vprod3x5, vb5x1, va3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 197 const int8x8_t vb5x1 = vld1_s8(w); w = (const void*) ((uintptr_t) w + 8 * sizeof( int8_t)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 201 vprod0x5 = vmlal_s8(vprod0x5, vb5x1, va0x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 202 vprod1x5 = vmlal_s8(vprod1x5, vb5x1, va1x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 203 vprod2x5 = vmlal_s8(vprod2x5, vb5x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|