Searched refs:vprod3x8 (Results 1 – 6 of 6) sorted by relevance
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 4x16c8-minmax-neon-mlal-padal.c | 264 int16x8_t vprod3x8 = vmull_s8(vb8x0, va3x0); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 268 vprod3x8 = vmlal_s8(vprod3x8, vb8x1, va3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 272 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 451 const int16x8_t vprod3x8 = vmull_s8(vb8, va3); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 455 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 251 int16x8_t vprod3x8 = vmull_s8(vget_low_s8(vb8), vget_low_s8(va3)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 255 vprod3x8 = vmlal_s8(vprod3x8, vget_high_s8(vb8), vget_high_s8(va3)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() 259 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 212 const int16x8_t vprod3x8 = vmull_s8(vb8, va3); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local 216 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 4x16c8-minmax-neon-mlal-padal.c | 281 int16x8_t vprod3x8 = vmull_s8(vb8x0, va3x0); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 285 vprod3x8 = vmlal_s8(vprod3x8, vb8x1, va3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 289 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 468 const int16x8_t vprod3x8 = vmull_s8(vb8, va3); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 472 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 268 int16x8_t vprod3x8 = vmull_s8(vget_low_s8(vb8), vget_low_s8(va3)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 272 vprod3x8 = vmlal_s8(vprod3x8, vget_high_s8(vb8), vget_high_s8(va3)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() 276 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 229 const int16x8_t vprod3x8 = vmull_s8(vb8, va3); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local 233 vacc3x8 = vpadalq_s16(vacc3x8, vprod3x8); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|