/external/XNNPACK/src/qs8-igemm/gen/ |
D | 3x16c8-minmax-neon-mull-padal.c | 101 int32x4_t vacc2x14 = vacc0x14; 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() 271 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() 364 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 101 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 296 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 421 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 460 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 553 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 101 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 278 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() 319 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() 412 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 105 int32x4_t vacc2x14 = vacc0x14; 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() 328 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() 433 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 105 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 366 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 525 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 567 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 672 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 105 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 347 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() 392 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() 497 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x16c8-minmax-neon-mull-padal.c | 102 int32x4_t vacc2x14 = vacc0x14; 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() 253 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() 346 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 102 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 281 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 406 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 442 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 535 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 102 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 263 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() 301 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() 394 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 108 int32x4_t vacc2x14 = vacc0x14; 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() 308 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() 413 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 108 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 349 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 508 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 547 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 652 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 108 int32x4_t vacc2x14 = vacc0x14; in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 330 vacc2x14 = vpadalq_s16(vacc2x14, vprod2x14); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() 372 const int32x4_t vsum2xEF = vpaddq_s32(vacc2x14, vacc2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() 477 const int32x2_t vpsum2xE = vadd_s32(vget_low_s32(vacc2x14), vget_high_s32(vacc2x14)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|