| /external/XNNPACK/src/f32-velu/gen/ |
| D | velu-avx-rr2-lut4-p4-perm-x16.c | 57 __m256 ven0 = _mm256_andnot_ps(vindex_mask, vn0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x16() local
59 …128 ven0_lo = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_castps256_ps128(ven0)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x16()
65 …8 ven0_hi = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_extractf128_ps(ven0, 1)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x16()
70 ven0 = _mm256_insertf128_ps(_mm256_castps128_ps256(ven0_lo), ven0_hi, 1); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x16()
75 __m256 vs0 = _mm256_mul_ps(vl0, ven0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x16()
|
| D | velu-avx-rr2-lut4-p4-perm-x24.c | 60 __m256 ven0 = _mm256_andnot_ps(vindex_mask, vn0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x24() local
62 …128 ven0_lo = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_castps256_ps128(ven0)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x24()
71 …8 ven0_hi = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_extractf128_ps(ven0, 1)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x24()
78 ven0 = _mm256_insertf128_ps(_mm256_castps128_ps256(ven0_lo), ven0_hi, 1); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x24()
85 __m256 vs0 = _mm256_mul_ps(vl0, ven0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x24()
|
| D | velu-avx-rr2-lut4-p4-perm-x32.c | 63 __m256 ven0 = _mm256_andnot_ps(vindex_mask, vn0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x32() local
65 …128 ven0_lo = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_castps256_ps128(ven0)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x32()
77 …8 ven0_hi = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_extractf128_ps(ven0, 1)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x32()
86 ven0 = _mm256_insertf128_ps(_mm256_castps128_ps256(ven0_lo), ven0_hi, 1); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x32()
95 __m256 vs0 = _mm256_mul_ps(vl0, ven0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x32()
|
| D | velu-wasm-rr2-lut16-p3-x2.c | 54 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2() local
62 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2()
|
| D | velu-scalar-rr2-lut16-p3-x2.c | 54 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2() local
62 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
|
| D | velu-avx-rr2-lut4-p4-perm-x40.c | 66 __m256 ven0 = _mm256_andnot_ps(vindex_mask, vn0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x40() local
68 …128 ven0_lo = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_castps256_ps128(ven0)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x40()
83 …8 ven0_hi = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_extractf128_ps(ven0, 1)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x40()
94 ven0 = _mm256_insertf128_ps(_mm256_castps128_ps256(ven0_lo), ven0_hi, 1); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x40()
105 __m256 vs0 = _mm256_mul_ps(vl0, ven0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x40()
|
| D | velu-scalar-rr2-lut16-p3-x3.c | 57 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3() local
68 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
|
| D | velu-wasm-rr2-lut16-p3-x3.c | 57 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x3() local
68 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x3()
|
| D | velu-avx-rr2-lut4-p4-perm-x48.c | 69 __m256 ven0 = _mm256_andnot_ps(vindex_mask, vn0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x48() local
71 …128 ven0_lo = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_castps256_ps128(ven0)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x48()
89 …8 ven0_hi = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(_mm256_extractf128_ps(ven0, 1)), 21)); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x48()
102 ven0 = _mm256_insertf128_ps(_mm256_castps128_ps256(ven0_lo), ven0_hi, 1); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x48()
115 __m256 vs0 = _mm256_mul_ps(vl0, ven0); in xnn_f32_velu_ukernel__avx_rr2_lut4_p4_perm_x48()
|
| D | velu-wasm-rr2-lut16-p3-x4.c | 60 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4() local
74 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
|
| D | velu-scalar-rr2-lut16-p3-x4.c | 60 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4() local
74 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
|
| D | velu-scalar-rr2-lut16-p3-x5.c | 63 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5() local
80 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
|
| D | velu-wasm-rr2-lut16-p3-x5.c | 63 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5() local
80 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
|
| D | velu-avx512f-rr1-lut16-p3-perm-x32.c | 53 const __m512i ven0 = _mm512_slli_epi32(_mm512_castps_si512(vn0), 19); in xnn_f32_velu_ukernel__avx512f_rr1_lut16_p3_perm_x32() local
58 __m512 vs0 = _mm512_castsi512_ps(_mm512_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx512f_rr1_lut16_p3_perm_x32()
|
| D | velu-avx2-rr1-lut16-p3-gather-x16.c | 58 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 19); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x16() local
63 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x16()
|
| D | velu-avx2-rr1-lut8-p4-perm-x16.c | 53 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 20); in xnn_f32_velu_ukernel__avx2_rr1_lut8_p4_perm_x16() local
60 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut8_p4_perm_x16()
|
| D | velu-avx2-rr1-lut4-p4-perm-x16.c | 54 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 21); in xnn_f32_velu_ukernel__avx2_rr1_lut4_p4_perm_x16() local
61 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut4_p4_perm_x16()
|
| D | velu-wasm-rr2-lut16-p3-x6.c | 66 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x6() local
86 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x6()
|
| D | velu-scalar-rr2-lut16-p3-x6.c | 66 const uint32_t ven0 = fp32_to_bits(vn0) << 19; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6() local
86 float vs0 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx0] + ven0); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
|
| D | velu-avx512f-rr1-lut16-p3-perm-x48.c | 56 const __m512i ven0 = _mm512_slli_epi32(_mm512_castps_si512(vn0), 19); in xnn_f32_velu_ukernel__avx512f_rr1_lut16_p3_perm_x48() local
63 __m512 vs0 = _mm512_castsi512_ps(_mm512_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx512f_rr1_lut16_p3_perm_x48()
|
| D | velu-avx2-rr1-lut16-p3-gather-x24.c | 63 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 19); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x24() local
70 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x24()
|
| D | velu-avx2-rr1-lut8-p4-perm-x24.c | 56 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 20); in xnn_f32_velu_ukernel__avx2_rr1_lut8_p4_perm_x24() local
66 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut8_p4_perm_x24()
|
| D | velu-avx2-rr1-lut4-p4-perm-x24.c | 57 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 21); in xnn_f32_velu_ukernel__avx2_rr1_lut4_p4_perm_x24() local
67 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut4_p4_perm_x24()
|
| D | velu-avx2-rr1-lut16-p3-gather-x32.c | 68 const __m256i ven0 = _mm256_slli_epi32(_mm256_castps_si256(vn0), 19); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x32() local
77 __m256 vs0 = _mm256_castsi256_ps(_mm256_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x32()
|
| D | velu-avx512f-rr1-lut16-p3-perm-x64.c | 59 const __m512i ven0 = _mm512_slli_epi32(_mm512_castps_si512(vn0), 19); in xnn_f32_velu_ukernel__avx512f_rr1_lut16_p3_perm_x64() local
68 __m512 vs0 = _mm512_castsi512_ps(_mm512_add_epi32(vl0, ven0)); in xnn_f32_velu_ukernel__avx512f_rr1_lut16_p3_perm_x64()
|