/external/XNNPACK/src/qs8-igemm/gen/ |
D | 3x16c8-minmax-neon-mull-padal.c | 102 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() local 239 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 365 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 102 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 306 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 428 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 554 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 102 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 287 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 413 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 106 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local 295 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 434 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 106 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 379 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 534 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 673 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 106 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 359 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 498 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x16c8-minmax-neon-mull-padal.c | 103 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() local 224 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 347 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 103 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 291 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 413 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 536 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 103 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 272 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 395 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 109 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local 278 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 414 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 109 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 362 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 517 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 653 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 109 int32x4_t vacc2x15 = vacc0x15; in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 342 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); 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() 478 const int32x2_t vpsum2xF = vadd_s32(vget_low_s32(vacc2x15), vget_high_s32(vacc2x15)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|