/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x16c8-minmax-neon-mlal-padal.c | 275 int16x8_t vprod2x14 = vmull_s8(vb14x0, va2x0); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 278 vprod2x14 = vmlal_s8(vprod2x14, vb14x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 281 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 403 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 406 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 257 int16x8_t vprod2x14 = vmull_s8(vget_low_s8(vb14), vget_low_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 260 vprod2x14 = vmlal_s8(vprod2x14, vget_high_s8(vb14), vget_high_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() 263 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 341 int16x8_t vprod2x14 = vmull_s8(vb14x0, va2x0); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 345 vprod2x14 = vmlal_s8(vprod2x14, vb14x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 349 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 504 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 508 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 322 int16x8_t vprod2x14 = vmull_s8(vget_low_s8(vb14), vget_low_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 326 vprod2x14 = vmlal_s8(vprod2x14, vget_high_s8(vb14), vget_high_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() 330 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
D | 3x16c8-minmax-neon-mull-padal.c | 214 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() local 217 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 265 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local 269 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 3x16c8-minmax-neon-mlal-padal.c | 290 int16x8_t vprod2x14 = vmull_s8(vb14x0, va2x0); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 293 vprod2x14 = vmlal_s8(vprod2x14, vb14x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 296 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 418 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 421 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 272 int16x8_t vprod2x14 = vmull_s8(vget_low_s8(vb14), vget_low_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 275 vprod2x14 = vmlal_s8(vprod2x14, vget_high_s8(vb14), vget_high_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() 278 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 358 int16x8_t vprod2x14 = vmull_s8(vb14x0, va2x0); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 362 vprod2x14 = vmlal_s8(vprod2x14, vb14x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 366 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 521 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 525 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 339 int16x8_t vprod2x14 = vmull_s8(vget_low_s8(vb14), vget_low_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 343 vprod2x14 = vmlal_s8(vprod2x14, vget_high_s8(vb14), vget_high_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() 347 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
D | 3x16c8-minmax-neon-mull-padal.c | 229 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() local 232 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 282 const int16x8_t vprod2x14 = vmull_s8(vb14, va2); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local 286 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|