Searched refs:va01 (Results 1 – 10 of 10) sorted by relevance
| /external/XNNPACK/src/qu8-vadd/ |
| D | minmax-neon.c | 33 const uint8x16_t va01 = vld1q_u8(a); a += 16; in xnn_qu8_vadd_minmax_ukernel__neon() local
39 const int16x8_t vxa0 = vreinterpretq_s16_u16(vsubl_u8(vget_low_u8(va01), va_zero_point)); in xnn_qu8_vadd_minmax_ukernel__neon()
41 const int16x8_t vxa1 = vreinterpretq_s16_u16(vsubl_u8(vget_high_u8(va01), va_zero_point)); in xnn_qu8_vadd_minmax_ukernel__neon()
105 const uint8x16_t va01 = vld1q_u8(a); a += 16; in xnn_qu8_vadd_minmax_ukernel__neon() local
109 const int16x8_t vxa0 = vreinterpretq_s16_u16(vsubl_u8(vget_low_u8(va01), va_zero_point)); in xnn_qu8_vadd_minmax_ukernel__neon()
111 const int16x8_t vxa1 = vreinterpretq_s16_u16(vsubl_u8(vget_high_u8(va01), va_zero_point)); in xnn_qu8_vadd_minmax_ukernel__neon()
|
| /external/XNNPACK/src/f16-spmm/gen/ |
| D | 8x1-minmax-neonfp16arith.c | 106 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_8x1__neonfp16arith() local
109 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_8x1__neonfp16arith()
|
| D | 8x1-minmax-neonfp16arith-x2.c | 122 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_8x1__neonfp16arith_x2() local
125 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_8x1__neonfp16arith_x2()
|
| D | 16x1-minmax-neonfp16arith.c | 139 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_16x1__neonfp16arith() local
142 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_16x1__neonfp16arith()
|
| D | 16x1-minmax-neonfp16arith-x2.c | 162 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_16x1__neonfp16arith_x2() local
165 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_16x1__neonfp16arith_x2()
|
| D | 32x1-minmax-neonfp16arith.c | 185 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_32x1__neonfp16arith() local
188 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_32x1__neonfp16arith()
|
| D | 24x1-minmax-neonfp16arith.c | 178 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_24x1__neonfp16arith() local
181 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_24x1__neonfp16arith()
|
| D | 24x1-minmax-neonfp16arith-x2.c | 208 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_24x1__neonfp16arith_x2() local
211 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_24x1__neonfp16arith_x2()
|
| D | 32x1-minmax-neonfp16arith-x2.c | 222 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1))); in xnn_f16_spmm_minmax_ukernel_32x1__neonfp16arith_x2() local
225 vacc01 = vfma_f16(vacc01, va01, vb); in xnn_f16_spmm_minmax_ukernel_32x1__neonfp16arith_x2()
|
| /external/XNNPACK/src/f16-spmm/ |
| D | neonfp16arith.c.in | 128 … const float16x4_t va01 = vreinterpret_f16_f32(vld1_dup_f32(__builtin_assume_aligned(i, 1)));
|