/external/XNNPACK/src/qs8-igemm/gen/ |
D | 3x16c8-minmax-neon-mull-padal.c | 98 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() local 211 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() 269 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() 358 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 98 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 266 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 400 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 458 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() 547 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 98 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 251 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() 317 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() 406 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 102 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local 259 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() 326 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() 427 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 102 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 327 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 498 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 565 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() 666 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 102 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 311 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() 390 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() 491 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x16c8-minmax-neon-mull-padal.c | 99 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() local 196 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() 251 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() 340 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
D | 3x16c8-minmax-neon-mlal-padal.c | 99 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local 251 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 385 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 440 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() 529 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
D | 3x16c16-minmax-neon-mlal-padal.c | 99 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() local 236 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() 299 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() 388 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
D | 4x16c8-minmax-neon-mull-padal.c | 105 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local 242 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() 306 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() 407 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
D | 4x16c8-minmax-neon-mlal-padal.c | 105 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local 310 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 481 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 545 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() 646 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
D | 4x16c16-minmax-neon-mlal-padal.c | 105 int32x4_t vacc2x11 = vacc0x11; in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local 294 vacc2x11 = vpadalq_s16(vacc2x11, vprod2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() 370 const int32x4_t vsum2xAB = vpaddq_s32(vacc2x10, vacc2x11); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() 471 const int32x2_t vpsum2xB = vadd_s32(vget_low_s32(vacc2x11), vget_high_s32(vacc2x11)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|